Merge pull request #3265 from LeSuisse/sign-container-image-cosign

Sign container image using Cosign in keyless mode

.github/ISSUE_TEMPLATE/sprint_issue.md

@@ -22,6 +22,20 @@ Related product discussion:
 
 <!---If necessary, create a list with technical/product steps-->
 
+### Are you modifying a database?
+- [ ] If not, add the `no db change` label to your PR, and you're good to merge.
+- [ ] If yes, add the `db change` label to your PR. You'll receive a message explaining you what to do.
+
+### Reminders when modifying the API
+
+- [ ] Update the openAPI file with utoipa:
+  - [ ] If a new module has been introduced, create a new structure deriving [the OpenAPI proc-macro](https://docs.rs/utoipa/latest/utoipa/derive.OpenApi.html) and nest it in the main [openAPI structure](https://github.com/meilisearch/meilisearch/blob/f2185438eed60fa32d25b15480c5ee064f6fba4a/crates/meilisearch/src/routes/mod.rs#L64-L78).
+  - [ ] If a new route has been introduced, add the [path decorator](https://docs.rs/utoipa/latest/utoipa/attr.path.html) to it and add the route at the top of the file in its openAPI structure.
+  - [ ] If a structure which is deserialized or serialized in the API has been introduced or modified, it must derive the [`schema`](https://docs.rs/utoipa/latest/utoipa/macro.schema.html) or the [`IntoParams`](https://docs.rs/utoipa/latest/utoipa/derive.IntoParams.html) proc-macro.
+        If it's a **new** structure you must also add it to the big list of structures [in the main `OpenApi` structure](https://github.com/meilisearch/meilisearch/blob/f2185438eed60fa32d25b15480c5ee064f6fba4a/crates/meilisearch/src/routes/mod.rs#L88).
+  - [ ] Once everything is done, start Meilisearch with the swagger flag: `cargo run --features swagger`, open `http://localhost:7700/scalar` on your browser, and ensure everything works as expected.
+  - For more info, refer to [this presentation](https://pitch.com/v/generating-the-openapi-file-jrn3nh).
+
 ### Reminders when modifying the Setting API
 
 <!--- Special steps to remind when adding a new index setting -->

.github/workflows/bench-manual.yml

@@ -4,9 +4,9 @@ on:
   workflow_dispatch:
     inputs:
       workload:
-                description: 'The path to the workloads to execute (workloads/...)'
+        description: "The path to the workloads to execute (workloads/...)"
         required: true
-                default: 'workloads/movies.json'
+        default: "workloads/movies.json"
 
 env:
   WORKLOAD_NAME: ${{ github.event.inputs.workload }}
@@ -18,11 +18,10 @@ jobs:
     timeout-minutes: 180 # 3h
     steps:
       - uses: actions/checkout@v3
-            - uses: dtolnay/rust-toolchain@1.81
+      - uses: dtolnay/rust-toolchain@1.85
         with:
           profile: minimal
 
       - name: Run benchmarks - workload ${WORKLOAD_NAME} - branch ${{ github.ref }} - commit ${{ github.sha }}
         run: |
           cargo xtask bench --api-key "${{ secrets.BENCHMARK_API_KEY }}" --dashboard-url "${{ vars.BENCHMARK_DASHBOARD_URL }}" --reason "Manual [Run #${{ github.run_id }}](https://github.com/meilisearch/meilisearch/actions/runs/${{ github.run_id }})" -- ${WORKLOAD_NAME}
-

.github/workflows/bench-pr.yml

@@ -66,7 +66,7 @@ jobs:
           fetch-depth: 0 # fetch full history to be able to get main commit sha
           ref: ${{ steps.comment-branch.outputs.head_ref }}
 
-        - uses: dtolnay/rust-toolchain@1.81
+      - uses: dtolnay/rust-toolchain@1.85
         with:
           profile: minimal
 

.github/workflows/bench-push-indexing.yml

@@ -12,7 +12,7 @@ jobs:
     timeout-minutes: 180 # 3h
     steps:
       - uses: actions/checkout@v3
-          - uses: dtolnay/rust-toolchain@1.81
+      - uses: dtolnay/rust-toolchain@1.85
         with:
           profile: minimal
 
@@ -20,4 +20,3 @@ jobs:
       - name: Run benchmarks - Dataset ${BENCH_NAME} - Branch main - Commit ${{ github.sha }}
         run: |
           cargo xtask bench --api-key "${{ secrets.BENCHMARK_API_KEY }}" --dashboard-url "${{ vars.BENCHMARK_DASHBOARD_URL }}" --reason "Push on `main` [Run #${{ github.run_id }}](https://github.com/meilisearch/meilisearch/actions/runs/${{ github.run_id }})" -- workloads/*.json
-

.github/workflows/benchmarks-manual.yml

@@ -4,9 +4,9 @@ on:
   workflow_dispatch:
     inputs:
       dataset_name:
-        description: 'The name of the dataset used to benchmark (search_songs, search_wiki, search_geo or indexing)'
+        description: "The name of the dataset used to benchmark (search_songs, search_wiki, search_geo or indexing)"
         required: false
-        default: 'search_songs'
+        default: "search_songs"
 
 env:
   BENCH_NAME: ${{ github.event.inputs.dataset_name }}
@@ -18,7 +18,7 @@ jobs:
     timeout-minutes: 4320 # 72h
     steps:
       - uses: actions/checkout@v3
-      - uses: dtolnay/rust-toolchain@1.81
+      - uses: dtolnay/rust-toolchain@1.85
         with:
           profile: minimal
 
@@ -67,7 +67,7 @@ jobs:
           out_dir: critcmp_results
 
       # Helper
-      - name: 'README: compare with another benchmark'
+      - name: "README: compare with another benchmark"
         run: |
           echo "${{ steps.file.outputs.basename }}.json has just been pushed."
           echo 'How to compare this benchmark with another one?'

.github/workflows/benchmarks-pr.yml

@@ -44,7 +44,7 @@ jobs:
             exit 1
           fi
 
-      - uses: dtolnay/rust-toolchain@1.81
+      - uses: dtolnay/rust-toolchain@1.85
         with:
           profile: minimal
 

.github/workflows/benchmarks-push-indexing.yml

@@ -16,7 +16,7 @@ jobs:
     timeout-minutes: 4320 # 72h
     steps:
       - uses: actions/checkout@v3
-      - uses: dtolnay/rust-toolchain@1.81
+      - uses: dtolnay/rust-toolchain@1.85
         with:
           profile: minimal
 
@@ -69,7 +69,7 @@ jobs:
         run: telegraf --config https://eu-central-1-1.aws.cloud2.influxdata.com/api/v2/telegrafs/08b52e34a370b000 --once --debug
 
       # Helper
-      - name: 'README: compare with another benchmark'
+      - name: "README: compare with another benchmark"
         run: |
           echo "${{ steps.file.outputs.basename }}.json has just been pushed."
           echo 'How to compare this benchmark with another one?'

.github/workflows/benchmarks-push-search-geo.yml

@@ -15,7 +15,7 @@ jobs:
     runs-on: benchmarks
     steps:
       - uses: actions/checkout@v3
-      - uses: dtolnay/rust-toolchain@1.81
+      - uses: dtolnay/rust-toolchain@1.85
         with:
           profile: minimal
 
@@ -68,7 +68,7 @@ jobs:
         run: telegraf --config https://eu-central-1-1.aws.cloud2.influxdata.com/api/v2/telegrafs/08b52e34a370b000 --once --debug
 
       # Helper
-      - name: 'README: compare with another benchmark'
+      - name: "README: compare with another benchmark"
         run: |
           echo "${{ steps.file.outputs.basename }}.json has just been pushed."
           echo 'How to compare this benchmark with another one?'

.github/workflows/benchmarks-push-search-songs.yml

@@ -15,7 +15,7 @@ jobs:
     runs-on: benchmarks
     steps:
       - uses: actions/checkout@v3
-      - uses: dtolnay/rust-toolchain@1.81
+      - uses: dtolnay/rust-toolchain@1.85
         with:
           profile: minimal
 
@@ -68,7 +68,7 @@ jobs:
         run: telegraf --config https://eu-central-1-1.aws.cloud2.influxdata.com/api/v2/telegrafs/08b52e34a370b000 --once --debug
 
       # Helper
-      - name: 'README: compare with another benchmark'
+      - name: "README: compare with another benchmark"
         run: |
           echo "${{ steps.file.outputs.basename }}.json has just been pushed."
           echo 'How to compare this benchmark with another one?'

.github/workflows/benchmarks-push-search-wiki.yml

@@ -15,7 +15,7 @@ jobs:
     runs-on: benchmarks
     steps:
       - uses: actions/checkout@v3
-      - uses: dtolnay/rust-toolchain@1.81
+      - uses: dtolnay/rust-toolchain@1.85
         with:
           profile: minimal
 
@@ -68,7 +68,7 @@ jobs:
         run: telegraf --config https://eu-central-1-1.aws.cloud2.influxdata.com/api/v2/telegrafs/08b52e34a370b000 --once --debug
 
       # Helper
-      - name: 'README: compare with another benchmark'
+      - name: "README: compare with another benchmark"
         run: |
           echo "${{ steps.file.outputs.basename }}.json has just been pushed."
           echo 'How to compare this benchmark with another one?'

.github/workflows/check-valid-milestone.yml

@@ -0,0 +1,100 @@
+name: PR Milestone Check
+
+on:
+  pull_request:
+    types: [opened, reopened, edited, synchronize, milestoned, demilestoned]
+    branches:
+      - "main"
+      - "release-v*.*.*"
+
+jobs:
+  check-milestone:
+    name: Check PR Milestone
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v3
+
+      - name: Validate PR milestone
+        uses: actions/github-script@v7
+        with:
+          github-token: ${{ secrets.GITHUB_TOKEN }}
+          script: |
+            // Get PR number directly from the event payload
+            const prNumber = context.payload.pull_request.number;
+
+            // Get PR details
+            const { data: prData } = await github.rest.pulls.get({
+              owner: 'meilisearch',
+              repo: 'meilisearch',
+              pull_number: prNumber
+            });
+
+            // Get base branch name
+            const baseBranch = prData.base.ref;
+            console.log(`Base branch: ${baseBranch}`);
+
+            // Get PR milestone
+            const prMilestone = prData.milestone;
+            if (!prMilestone) {
+              core.setFailed('PR must have a milestone assigned');
+              return;
+            }
+            console.log(`PR milestone: ${prMilestone.title}`);
+
+            // Validate milestone format: vx.y.z
+            const milestoneRegex = /^v\d+\.\d+\.\d+$/;
+            if (!milestoneRegex.test(prMilestone.title)) {
+              core.setFailed(`Milestone "${prMilestone.title}" does not follow the required format vx.y.z`);
+              return;
+            }
+
+            // For main branch PRs, check if the milestone is the highest one
+            if (baseBranch === 'main') {
+              // Get all milestones
+              const { data: milestones } = await github.rest.issues.listMilestones({
+                owner: 'meilisearch',
+                repo: 'meilisearch',
+                state: 'open',
+                sort: 'due_on',
+                direction: 'desc'
+              });
+
+              // Sort milestones by version number (vx.y.z)
+              const sortedMilestones = milestones
+                .filter(m => milestoneRegex.test(m.title))
+                .sort((a, b) => {
+                  const versionA = a.title.substring(1).split('.').map(Number);
+                  const versionB = b.title.substring(1).split('.').map(Number);
+
+                  // Compare major version
+                  if (versionA[0] !== versionB[0]) return versionB[0] - versionA[0];
+                  // Compare minor version
+                  if (versionA[1] !== versionB[1]) return versionB[1] - versionA[1];
+                  // Compare patch version
+                  return versionB[2] - versionA[2];
+                });
+
+              if (sortedMilestones.length === 0) {
+                core.setFailed('No valid milestones found in the repository. Please create at least one milestone with the format vx.y.z');
+                return;
+              }
+
+              const highestMilestone = sortedMilestones[0];
+              console.log(`Highest milestone: ${highestMilestone.title}`);
+
+              if (prMilestone.title !== highestMilestone.title) {
+                core.setFailed(`PRs targeting the main branch must use the highest milestone (${highestMilestone.title}), but this PR uses ${prMilestone.title}`);
+                return;
+              }
+            } else {
+              // For release branches, the milestone should match the branch version
+              const branchVersion = baseBranch.substring(8); // remove 'release-'
+              if (prMilestone.title !== branchVersion) {
+                core.setFailed(`PRs targeting release branch "${baseBranch}" must use the matching milestone "${branchVersion}", but this PR uses "${prMilestone.title}"`);
+                return;
+              }
+            }
+
+            console.log('PR milestone validation passed!');

.github/workflows/db-change-comments.yml

@@ -0,0 +1,57 @@
+name: Comment when db change labels are added
+
+on:
+  pull_request:
+    types: [labeled]
+
+env:
+  MESSAGE: |
+    ### Hello, I'm a bot 🤖 
+
+    You are receiving this message because you declared that this PR make changes to the Meilisearch database.
+    Depending on the nature of the change, additional actions might be required on your part. The following sections detail the additional actions depending on the nature of the change, please copy the relevant section in the description of your PR, and make sure to perform the required actions.
+
+    Thank you for contributing to Meilisearch :heart:
+
+    ## This PR makes forward-compatible changes
+
+    *Forward-compatible changes are changes to the database such that databases created in an older version of Meilisearch are still valid in the new version of Meilisearch. They usually represent additive changes, like adding a new optional attribute or setting.*
+
+    - [ ] Detail the change to the DB format and why they are forward compatible
+    - [ ] Forward-compatibility: A database created before this PR and using the features touched by this PR was able to be opened by a Meilisearch produced by the code of this PR.
+
+
+    ## This PR makes breaking changes
+
+    *Breaking changes are changes to the database such that databases created in an older version of Meilisearch need changes to remain valid in the new version of Meilisearch. This typically happens when the way to store the data changed (change of database, new required key, etc). This can also happen due to breaking changes in the API of an experimental feature. ⚠️ This kind of changes are more difficult to achieve safely, so proceed with caution and test dumpless upgrade right before merging the PR.*
+
+    - [ ] Detail the changes to the DB format,
+        - [ ] which are compatible, and why
+        - [ ] which are not compatible, why, and how they will be fixed up in the upgrade
+     - [ ] /!\ Ensure all the read operations still work!
+        - If the change happened in milli, you may need to check the version of the database before doing any read operation
+        - If the change happened in the index-scheduler, make sure the new code can immediately read the old database
+        - If the change happened in the meilisearch-auth database, reach out to the team; we don't know yet how to handle these changes
+      - [ ] Write the code to go from the old database to the new one
+        - If the change happened in milli, the upgrade function should be written and called [here](https://github.com/meilisearch/meilisearch/blob/3fd86e8d76d7d468b0095d679adb09211ca3b6c0/crates/milli/src/update/upgrade/mod.rs#L24-L47)
+        - If the change happened in the index-scheduler, we've never done it yet, but the right place to do it should be [here](https://github.com/meilisearch/meilisearch/blob/3fd86e8d76d7d468b0095d679adb09211ca3b6c0/crates/index-scheduler/src/scheduler/process_upgrade/mod.rs#L13)
+      - [ ] Write an integration test [here](https://github.com/meilisearch/meilisearch/blob/main/crates/meilisearch/tests/upgrade/mod.rs) ensuring you can read the old database, upgrade to the new database, and read the new database as expected
+    
+
+jobs:
+  add-comment:
+    runs-on: ubuntu-latest
+    if: github.event.label.name == 'db change'
+    steps:
+      - name: Add comment
+        uses: actions/github-script@v7
+        with:
+          github-token: ${{ secrets.GITHUB_TOKEN }}
+          script: |
+            const message = process.env.MESSAGE;
+            github.rest.issues.createComment({
+              issue_number: context.issue.number,
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              body: message
+            })

.github/workflows/db-change-missing.yml

@@ -0,0 +1,28 @@
+name: Check db change labels
+
+on:
+  pull_request:
+    types: [opened, synchronize, reopened, labeled, unlabeled]
+
+jobs:
+  check-labels:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v4
+      - name: Check db change labels
+        id: check_labels
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        run: |
+          URL=/repos/meilisearch/meilisearch/pulls/${{ github.event.pull_request.number }}/labels
+          echo ${{ github.event.pull_request.number }}
+          echo $URL
+          LABELS=$(gh api -H "Accept: application/vnd.github+json" -H "X-GitHub-Api-Version: 2022-11-28" /repos/${{ github.repository }}/issues/${{ github.event.pull_request.number }}/labels -q .[].name)
+          echo "Labels: $LABELS"
+          if [[ ! "$LABELS" =~ "db change" && ! "$LABELS" =~ "no db change" ]]; then
+            echo "::error::Pull request must contain either the 'db change' or 'no db change' label."
+            exit 1
+          else
+            echo "The label is set"
+          fi

.github/workflows/flaky-tests.yml

@@ -9,15 +9,15 @@ jobs:
   flaky:
     runs-on: ubuntu-latest
     container:
-      # Use ubuntu-20.04 to compile with glibc 2.28
-      image: ubuntu:20.04
+      # Use ubuntu-22.04 to compile with glibc 2.35
+      image: ubuntu:22.04
     steps:
       - uses: actions/checkout@v3
       - name: Install needed dependencies
         run: |
           apt-get update && apt-get install -y curl
           apt-get install build-essential -y
-    - uses: dtolnay/rust-toolchain@1.81
+      - uses: dtolnay/rust-toolchain@1.85
       - name: Install cargo-flaky
         run: cargo install cargo-flaky
       - name: Run cargo flaky in the dumps

.github/workflows/fuzzer-indexing.yml

@@ -12,7 +12,7 @@ jobs:
     timeout-minutes: 4320 # 72h
     steps:
       - uses: actions/checkout@v3
-      - uses: dtolnay/rust-toolchain@1.81
+      - uses: dtolnay/rust-toolchain@1.85
         with:
           profile: minimal
 

.github/workflows/milestone-workflow.yml

@@ -5,6 +5,7 @@ name: Milestone's workflow
 # For each Milestone created (not opened!), and if the release is NOT a patch release (only the patch changed)
 # - the roadmap issue is created, see https://github.com/meilisearch/engine-team/blob/main/issue-templates/roadmap-issue.md
 # - the changelog issue is created, see https://github.com/meilisearch/engine-team/blob/main/issue-templates/changelog-issue.md
+# - update the ruleset to add the current release version to the list of allowed versions and be able to use the merge queue.
 
 # For each Milestone closed
 # - the `release_version` label is created
@@ -21,10 +22,9 @@ env:
   GH_TOKEN: ${{ secrets.MEILI_BOT_GH_PAT }}
 
 jobs:
-
-# -----------------
-# MILESTONE CREATED
-# -----------------
+  # -----------------
+  # MILESTONE CREATED
+  # -----------------
 
   get-release-version:
     if: github.event.action == 'created'
@@ -148,9 +148,41 @@ jobs:
             --body-file $ISSUE_TEMPLATE \
             --milestone $MILESTONE_VERSION
 
-# ----------------
-# MILESTONE CLOSED
-# ----------------
+  update-ruleset:
+    runs-on: ubuntu-latest
+    if: github.event.action == 'created'
+    steps:
+      - uses: actions/checkout@v3
+      - name: Install jq
+        run: |
+          sudo apt-get update
+          sudo apt-get install -y jq
+      - name: Update ruleset
+        env:
+          # gh api repos/meilisearch/meilisearch/rulesets --jq '.[] | {name: .name, id: .id}'
+          RULESET_ID: 4253297
+          BRANCH_NAME: ${{ github.event.inputs.branch_name }}
+        run: |
+          echo "RULESET_ID: ${{ env.RULESET_ID }}"
+          echo "BRANCH_NAME: ${{ env.BRANCH_NAME }}"
+
+          # Get current ruleset conditions
+          CONDITIONS=$(gh api repos/meilisearch/meilisearch/rulesets/${{ env.RULESET_ID }} --jq '{ conditions: .conditions }')
+
+          # Update the conditions by appending the milestone version
+          UPDATED_CONDITIONS=$(echo $CONDITIONS | jq '.conditions.ref_name.include += ["refs/heads/release-'${{ env.MILESTONE_VERSION }}'"]')
+
+          # Update the ruleset from stdin (-)
+          echo $UPDATED_CONDITIONS |
+            gh api repos/meilisearch/meilisearch/rulesets/${{ env.RULESET_ID }} \
+              --method PUT \
+              -H "Accept: application/vnd.github+json" \
+              -H "X-GitHub-Api-Version: 2022-11-28" \
+              --input -
+
+  # ----------------
+  # MILESTONE CLOSED
+  # ----------------
 
   create-release-label:
     if: github.event.action == 'closed'

.github/workflows/publish-apt-brew-pkg.yml

@@ -18,21 +18,21 @@ jobs:
     runs-on: ubuntu-latest
     needs: check-version
     container:
-      # Use ubuntu-20.04 to compile with glibc 2.28
-      image: ubuntu:20.04
+      # Use ubuntu-22.04 to compile with glibc 2.35
+      image: ubuntu:22.04
     steps:
       - name: Install needed dependencies
         run: |
           apt-get update && apt-get install -y curl
           apt-get install build-essential -y
-    - uses: dtolnay/rust-toolchain@1.81
+      - uses: dtolnay/rust-toolchain@1.85
       - name: Install cargo-deb
         run: cargo install cargo-deb
       - uses: actions/checkout@v3
       - name: Build deb package
         run: cargo deb -p meilisearch -o target/debian/meilisearch.deb
       - name: Upload debian pkg to release
-      uses: svenstaro/upload-release-action@2.7.0
+        uses: svenstaro/upload-release-action@2.11.1
         with:
           repo_token: ${{ secrets.MEILI_BOT_GH_PAT }}
           file: target/debian/meilisearch.deb

.github/workflows/publish-binaries.yml

@@ -3,7 +3,7 @@ name: Publish binaries to GitHub release
 on:
   workflow_dispatch:
   schedule:
-    - cron: '0 2 * * *' # Every day at 2:00am
+    - cron: "0 2 * * *" # Every day at 2:00am
   release:
     types: [published]
 
@@ -37,21 +37,21 @@ jobs:
     runs-on: ubuntu-latest
     needs: check-version
     container:
-      # Use ubuntu-20.04 to compile with glibc 2.28
-      image: ubuntu:20.04
+      # Use ubuntu-22.04 to compile with glibc 2.35
+      image: ubuntu:22.04
     steps:
       - uses: actions/checkout@v3
       - name: Install needed dependencies
         run: |
           apt-get update && apt-get install -y curl
           apt-get install build-essential -y
-    - uses: dtolnay/rust-toolchain@1.81
+      - uses: dtolnay/rust-toolchain@1.85
       - name: Build
         run: cargo build --release --locked
       # No need to upload binaries for dry run (cron)
       - name: Upload binaries to release
         if: github.event_name == 'release'
-      uses: svenstaro/upload-release-action@2.7.0
+        uses: svenstaro/upload-release-action@2.11.1
         with:
           repo_token: ${{ secrets.MEILI_BOT_GH_PAT }}
           file: target/release/meilisearch
@@ -75,13 +75,13 @@ jobs:
             asset_name: meilisearch-windows-amd64.exe
     steps:
       - uses: actions/checkout@v3
-    - uses: dtolnay/rust-toolchain@1.81
+      - uses: dtolnay/rust-toolchain@1.85
       - name: Build
         run: cargo build --release --locked
       # No need to upload binaries for dry run (cron)
       - name: Upload binaries to release
         if: github.event_name == 'release'
-      uses: svenstaro/upload-release-action@2.7.0
+        uses: svenstaro/upload-release-action@2.11.1
         with:
           repo_token: ${{ secrets.MEILI_BOT_GH_PAT }}
           file: target/release/${{ matrix.artifact_name }}
@@ -101,7 +101,7 @@ jobs:
       - name: Checkout repository
         uses: actions/checkout@v3
       - name: Installing Rust toolchain
-        uses: dtolnay/rust-toolchain@1.81
+        uses: dtolnay/rust-toolchain@1.85
         with:
           profile: minimal
           target: ${{ matrix.target }}
@@ -113,7 +113,7 @@ jobs:
       - name: Upload the binary to release
         # No need to upload binaries for dry run (cron)
         if: github.event_name == 'release'
-        uses: svenstaro/upload-release-action@2.7.0
+        uses: svenstaro/upload-release-action@2.11.1
         with:
           repo_token: ${{ secrets.MEILI_BOT_GH_PAT }}
           file: target/${{ matrix.target }}/release/meilisearch
@@ -127,8 +127,8 @@ jobs:
     env:
       DEBIAN_FRONTEND: noninteractive
     container:
-      # Use ubuntu-20.04 to compile with glibc 2.28
-      image: ubuntu:20.04
+      # Use ubuntu-22.04 to compile with glibc 2.35
+      image: ubuntu:22.04
     strategy:
       matrix:
         include:
@@ -148,7 +148,7 @@ jobs:
           add-apt-repository "deb [arch=$(dpkg --print-architecture)] https://download.docker.com/linux/ubuntu $(lsb_release -cs) stable"
           apt-get update -y && apt-get install -y docker-ce
       - name: Installing Rust toolchain
-        uses: dtolnay/rust-toolchain@1.81
+        uses: dtolnay/rust-toolchain@1.85
         with:
           profile: minimal
           target: ${{ matrix.target }}
@@ -178,7 +178,7 @@ jobs:
       - name: Upload the binary to release
         # No need to upload binaries for dry run (cron)
         if: github.event_name == 'release'
-        uses: svenstaro/upload-release-action@2.7.0
+        uses: svenstaro/upload-release-action@2.11.1
         with:
           repo_token: ${{ secrets.MEILI_BOT_GH_PAT }}
           file: target/${{ matrix.target }}/release/meilisearch

.github/workflows/publish-docker-images.yml

@@ -16,6 +16,8 @@ on:
 jobs:
   docker:
     runs-on: docker
+    permissions:
+      id-token: write # This is needed to use Cosign in keyless mode
     steps:
       - uses: actions/checkout@v3
 
@@ -62,6 +64,9 @@ jobs:
       - name: Set up Docker Buildx
         uses: docker/setup-buildx-action@v3
 
+      - name: Install cosign
+        uses: sigstore/cosign-installer@3454372f43399081ed03b604cb2d021dabca52bb # tag=v3.8.2
+
       - name: Login to Docker Hub
         uses: docker/login-action@v3
         with:
@@ -85,6 +90,7 @@ jobs:
 
       - name: Build and push
         uses: docker/build-push-action@v6
+        id: build-and-push
         with:
           push: true
           platforms: linux/amd64,linux/arm64
@@ -94,6 +100,17 @@ jobs:
             COMMIT_DATE=${{ steps.build-metadata.outputs.date }}
             GIT_TAG=${{ github.ref_name }}
 
+      - name: Sign the images with GitHub OIDC Token
+        env:
+          DIGEST: ${{ steps.build-and-push.outputs.digest }}
+          TAGS: ${{ steps.meta.outputs.tags }}
+        run: |
+          images=""
+          for tag in ${TAGS}; do
+            images+="${tag}@${DIGEST} "
+          done
+          cosign sign --yes ${images}
+
       # /!\ Don't touch this without checking with Cloud team
       - name: Send CI information to Cloud team
         # Do not send if nightly build (i.e. 'schedule' or 'workflow_dispatch' event)
@@ -104,3 +121,22 @@ jobs:
           repository: meilisearch/meilisearch-cloud
           event-type: cloud-docker-build
           client-payload: '{ "meilisearch_version": "${{ github.ref_name }}", "stable": "${{ steps.check-tag-format.outputs.stable }}" }'
+
+      # Send notification to Swarmia to notify of a deployment: https://app.swarmia.com
+      # - name: 'Setup jq'
+      #   uses: dcarbone/install-jq-action
+      # - name: Send deployment to Swarmia
+      #   if: github.event_name == 'push' && success()
+      #   run: |
+      #     JSON_STRING=$( jq --null-input --compact-output \
+      #     --arg version "${{ github.ref_name }}" \
+      #     --arg appName "meilisearch" \
+      #     --arg environment "production" \
+      #     --arg commitSha "${{ github.sha }}" \
+      #     --arg repositoryFullName "${{ github.repository }}" \
+      #     '{"version": $version, "appName": $appName, "environment": $environment, "commitSha": $commitSha, "repositoryFullName": $repositoryFullName}' )
+
+      #     curl -H "Authorization: ${{ secrets.SWARMIA_DEPLOYMENTS_AUTHORIZATION }}" \
+      #       -H "Content-Type: application/json" \
+      #       -d "$JSON_STRING" \
+      #       https://hook.swarmia.com/deployments

.github/workflows/sdks-tests.yml

@@ -22,7 +22,7 @@ jobs:
     outputs:
       docker-image: ${{ steps.define-image.outputs.docker-image }}
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v3
       - name: Define the Docker image we need to use
         id: define-image
         run: |
@@ -46,7 +46,7 @@ jobs:
       MEILISEARCH_VERSION: ${{ needs.define-docker-image.outputs.docker-image }}
 
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v3
         with:
           repository: meilisearch/meilisearch-dotnet
       - name: Setup .NET Core
@@ -75,7 +75,7 @@ jobs:
         ports:
           - '7700:7700'
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v3
         with:
           repository: meilisearch/meilisearch-dart
       - uses: dart-lang/setup-dart@v1
@@ -103,7 +103,7 @@ jobs:
         uses: actions/setup-go@v5
         with:
           go-version: stable
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v3
         with:
           repository: meilisearch/meilisearch-go
       - name: Get dependencies
@@ -129,7 +129,7 @@ jobs:
         ports:
           - '7700:7700'
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v3
         with:
           repository: meilisearch/meilisearch-java
       - name: Set up Java
@@ -156,7 +156,7 @@ jobs:
         ports:
           - '7700:7700'
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v3
         with:
           repository: meilisearch/meilisearch-js
       - name: Setup node
@@ -191,7 +191,7 @@ jobs:
         ports:
           - '7700:7700'
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v3
         with:
           repository: meilisearch/meilisearch-php
       - name: Install PHP
@@ -220,7 +220,7 @@ jobs:
         ports:
           - '7700:7700'
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v3
         with:
           repository: meilisearch/meilisearch-python
       - name: Set up Python
@@ -245,7 +245,7 @@ jobs:
         ports:
           - '7700:7700'
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v3
         with:
           repository: meilisearch/meilisearch-ruby
       - name: Set up Ruby 3
@@ -270,7 +270,7 @@ jobs:
         ports:
           - '7700:7700'
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v3
         with:
           repository: meilisearch/meilisearch-rust
       - name: Build
@@ -291,7 +291,7 @@ jobs:
         ports:
           - '7700:7700'
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v3
         with:
           repository: meilisearch/meilisearch-swift
       - name: Run tests
@@ -314,7 +314,7 @@ jobs:
         ports:
           - '7700:7700'
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v3
         with:
           repository: meilisearch/meilisearch-js-plugins
       - name: Setup node
@@ -344,15 +344,23 @@ jobs:
           MEILI_NO_ANALYTICS: ${{ env.MEILI_NO_ANALYTICS }}
         ports:
           - '7700:7700'
+    env:
+      RAILS_VERSION: '7.0'
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v3
         with:
           repository: meilisearch/meilisearch-rails
-      - name: Set up Ruby 3
+      - name: Install SQLite dependencies
+        run: sudo apt-get update && sudo apt-get install -y libsqlite3-dev
+      - name: Set up Ruby
         uses: ruby/setup-ruby@v1
         with:
           ruby-version: 3
           bundler-cache: true
+      - name: Start MongoDB
+        uses: supercharge/mongodb-github-action@1.12.0
+        with:
+          mongodb-version: 8.0
       - name: Run tests
         run: bundle exec rspec
 
@@ -369,7 +377,7 @@ jobs:
         ports:
           - '7700:7700'
     steps:
-      - uses: actions/checkout@v4
+      - uses: actions/checkout@v3
         with:
           repository: meilisearch/meilisearch-symfony
       - name: Install PHP

.github/workflows/test-suite.yml

@@ -4,13 +4,9 @@ on:
   workflow_dispatch:
   schedule:
     # Everyday at 5:00am
-    - cron: '0 5 * * *'
+    - cron: "0 5 * * *"
   pull_request:
-  push:
-    # trying and staging branches are for Bors config
-    branches:
-      - trying
-      - staging
+  merge_group:
 
 env:
   CARGO_TERM_COLOR: always
@@ -19,11 +15,11 @@ env:
 
 jobs:
   test-linux:
-    name: Tests on ubuntu-20.04
+    name: Tests on ubuntu-22.04
     runs-on: ubuntu-latest
     container:
-      # Use ubuntu-20.04 to compile with glibc 2.28
-      image: ubuntu:20.04
+      # Use ubuntu-22.04 to compile with glibc 2.35
+      image: ubuntu:22.04
     steps:
       - uses: actions/checkout@v3
       - name: Install needed dependencies
@@ -31,9 +27,9 @@ jobs:
           apt-get update && apt-get install -y curl
           apt-get install build-essential -y
       - name: Setup test with Rust stable
-        uses: dtolnay/rust-toolchain@1.81
+        uses: dtolnay/rust-toolchain@1.85
       - name: Cache dependencies
-        uses: Swatinem/rust-cache@v2.7.7
+        uses: Swatinem/rust-cache@v2.8.0
       - name: Run cargo check without any default features
         uses: actions-rs/cargo@v1
         with:
@@ -55,8 +51,8 @@ jobs:
     steps:
       - uses: actions/checkout@v3
       - name: Cache dependencies
-        uses: Swatinem/rust-cache@v2.7.7
-      - uses: dtolnay/rust-toolchain@1.81
+        uses: Swatinem/rust-cache@v2.8.0
+      - uses: dtolnay/rust-toolchain@1.85
       - name: Run cargo check without any default features
         uses: actions-rs/cargo@v1
         with:
@@ -72,8 +68,8 @@ jobs:
     name: Tests almost all features
     runs-on: ubuntu-latest
     container:
-      # Use ubuntu-20.04 to compile with glibc 2.28
-      image: ubuntu:20.04
+      # Use ubuntu-22.04 to compile with glibc 2.35
+      image: ubuntu:22.04
     if: github.event_name == 'schedule' || github.event_name == 'workflow_dispatch'
     steps:
       - uses: actions/checkout@v3
@@ -81,19 +77,51 @@ jobs:
         run: |
           apt-get update
           apt-get install --assume-yes build-essential curl
-      - uses: dtolnay/rust-toolchain@1.81
+      - uses: dtolnay/rust-toolchain@1.85
       - name: Run cargo build with almost all features
         run: |
-          cargo build --workspace --locked --release --features "$(cargo xtask list-features --exclude-feature cuda)"
+          cargo build --workspace --locked --release --features "$(cargo xtask list-features --exclude-feature cuda,test-ollama)"
       - name: Run cargo test with almost all features
         run: |
-          cargo test --workspace --locked --release --features "$(cargo xtask list-features --exclude-feature cuda)"
+          cargo test --workspace --locked --release --features "$(cargo xtask list-features --exclude-feature cuda,test-ollama)"
+
+  ollama-ubuntu:
+    name: Test with Ollama
+    runs-on: ubuntu-latest
+    env:
+      MEILI_TEST_OLLAMA_SERVER: "http://localhost:11434"
+    steps:
+      - uses: actions/checkout@v3
+      - name: Install Ollama
+        run: |
+          curl -fsSL https://ollama.com/install.sh | sudo -E sh
+      - name: Start serving
+        run: |
+          # Run it in the background, there is no way to daemonise at the moment
+          ollama serve &
+
+          # A short pause is required before the HTTP port is opened
+          sleep 5
+
+          # This endpoint blocks until ready
+          time curl -i http://localhost:11434
+
+      - name: Pull nomic-embed-text & all-minilm
+        run: |
+          ollama pull nomic-embed-text
+          ollama pull all-minilm
+
+      - name: Run cargo test
+        uses: actions-rs/cargo@v1
+        with:
+          command: test
+          args: --locked --release --all --features test-ollama ollama
 
   test-disabled-tokenization:
     name: Test disabled tokenization
     runs-on: ubuntu-latest
     container:
-      image: ubuntu:20.04
+      image: ubuntu:22.04
     if: github.event_name == 'schedule' || github.event_name == 'workflow_dispatch'
     steps:
       - uses: actions/checkout@v3
@@ -101,7 +129,7 @@ jobs:
         run: |
           apt-get update
           apt-get install --assume-yes build-essential curl
-      - uses: dtolnay/rust-toolchain@1.81
+      - uses: dtolnay/rust-toolchain@1.85
       - name: Run cargo tree without default features and check lindera is not present
         run: |
           if cargo tree -f '{p} {f}' -e normal --no-default-features | grep -qz lindera; then
@@ -117,17 +145,17 @@ jobs:
     name: Run tests in debug
     runs-on: ubuntu-latest
     container:
-      # Use ubuntu-20.04 to compile with glibc 2.28
-      image: ubuntu:20.04
+      # Use ubuntu-22.04 to compile with glibc 2.35
+      image: ubuntu:22.04
     steps:
       - uses: actions/checkout@v3
       - name: Install needed dependencies
         run: |
           apt-get update && apt-get install -y curl
           apt-get install build-essential -y
-      - uses: dtolnay/rust-toolchain@1.81
+      - uses: dtolnay/rust-toolchain@1.85
       - name: Cache dependencies
-        uses: Swatinem/rust-cache@v2.7.7
+        uses: Swatinem/rust-cache@v2.8.0
       - name: Run tests in debug
         uses: actions-rs/cargo@v1
         with:
@@ -139,12 +167,12 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v3
-      - uses: dtolnay/rust-toolchain@1.81
+      - uses: dtolnay/rust-toolchain@1.85
         with:
           profile: minimal
           components: clippy
       - name: Cache dependencies
-        uses: Swatinem/rust-cache@v2.7.7
+        uses: Swatinem/rust-cache@v2.8.0
       - name: Run cargo clippy
         uses: actions-rs/cargo@v1
         with:
@@ -156,14 +184,14 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v3
-      - uses: dtolnay/rust-toolchain@1.81
+      - uses: dtolnay/rust-toolchain@1.85
         with:
           profile: minimal
           toolchain: nightly-2024-07-09
           override: true
           components: rustfmt
       - name: Cache dependencies
-        uses: Swatinem/rust-cache@v2.7.7
+        uses: Swatinem/rust-cache@v2.8.0
       - name: Run cargo fmt
         # Since we never ran the `build.rs` script in the benchmark directory we are missing one auto-generated import file.
         # Since we want to trigger (and fail) this action as fast as possible, instead of building the benchmark crate

.github/workflows/update-cargo-toml-version.yml

@@ -4,7 +4,7 @@ on:
   workflow_dispatch:
     inputs:
       new_version:
-        description: 'The new version (vX.Y.Z)'
+        description: "The new version (vX.Y.Z)"
         required: true
 
 env:
@@ -18,7 +18,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - uses: actions/checkout@v3
-      - uses: dtolnay/rust-toolchain@1.81
+      - uses: dtolnay/rust-toolchain@1.85
         with:
           profile: minimal
       - name: Install sd

.gitignore

@@ -18,5 +18,8 @@
 ##  ... unreviewed
 *.snap.new
 
+# Database snapshot
+crates/meilisearch/db.snapshot
+
 # Fuzzcheck data for the facet indexing fuzz test
 crates/milli/fuzz/update::facet::incremental::fuzz::fuzz/

CONTRIBUTING.md

@@ -57,9 +57,17 @@ This command will be triggered to each PR as a requirement for merging it.
 You can set the `LINDERA_CACHE` environment variable to speed up your successive builds by up to 2 minutes.
 It'll store some built artifacts in the directory of your choice.
 
-We recommend using the standard `$HOME/.cache/lindera` directory:
+We recommend using the `$HOME/.cache/meili/lindera` directory:
 ```sh
-export LINDERA_CACHE=$HOME/.cache/lindera
+export LINDERA_CACHE=$HOME/.cache/meili/lindera
+```
+
+You can set the `MILLI_BENCH_DATASETS_PATH` environment variable to further speed up your builds.
+It'll store some big files used for the benchmarks in the directory of your choice.
+
+We recommend using the `$HOME/.cache/meili/benches` directory:
+```sh
+export MILLI_BENCH_DATASETS_PATH=$HOME/.cache/meili/benches
 ```
 
 Furthermore, you can improve incremental compilation by setting the `MEILI_NO_VERGEN` environment variable.
@@ -95,6 +103,11 @@ Meilisearch follows the [cargo xtask](https://github.com/matklad/cargo-xtask) wo
 
 Run `cargo xtask --help` from the root of the repository to find out what is available.
 
+#### Update the openAPI file if the API changed
+
+To update the openAPI file in the code, see [sprint_issue.md](https://github.com/meilisearch/meilisearch/blob/main/.github/ISSUE_TEMPLATE/sprint_issue.md#reminders-when-modifying-the-api).
+If you want to update the openAPI file on the [open-api repository](https://github.com/meilisearch/open-api), see [update-openapi-issue.md](https://github.com/meilisearch/engine-team/blob/main/issue-templates/update-openapi-issue.md).
+
 ### Logging
 
 Meilisearch uses [`tracing`](https://lib.rs/crates/tracing) for logging purposes. Tracing logs are structured and can be displayed as JSON to the end user, so prefer passing arguments as fields rather than interpolating them in the message.
@@ -145,7 +158,7 @@ Some notes on GitHub PRs:
 - The PR title should be accurate and descriptive of the changes.
 - [Convert your PR as a draft](https://help.github.com/en/github/collaborating-with-issues-and-pull-requests/changing-the-stage-of-a-pull-request) if your changes are a work in progress: no one will review it until you pass your PR as ready for review.<br>
   The draft PRs are recommended when you want to show that you are working on something and make your work visible.
-- The branch related to the PR must be **up-to-date with `main`** before merging. Fortunately, this project uses [Bors](https://github.com/bors-ng/bors-ng) to automatically enforce this requirement without the PR author having to rebase manually.
+- The branch related to the PR must be **up-to-date with `main`** before merging. Fortunately, this project uses [GitHub Merge Queues](https://github.blog/news-insights/product-news/github-merge-queue-is-generally-available/) to automatically enforce this requirement without the PR author having to rebase manually.
 
 ## Release Process (for internal team only)
 
@@ -153,8 +166,7 @@ Meilisearch tools follow the [Semantic Versioning Convention](https://semver.org
 
 ### Automation to rebase and Merge the PRs
 
-This project integrates a bot that helps us manage pull requests merging.<br>
-_[Read more about this](https://github.com/meilisearch/integration-guides/blob/main/resources/bors.md)._
+This project uses GitHub Merge Queues that helps us manage pull requests merging.
 
 ### How to Publish a new Release
 

Cargo.toml

@@ -22,7 +22,7 @@ members = [
 ]
 
 [workspace.package]
-version = "1.13.0"
+version = "1.16.0"
 authors = [
     "Quentin de Quelen <quentin@dequelen.me>",
     "Clément Renault <clement@meilisearch.com>",
@@ -36,6 +36,12 @@ license = "MIT"
 [profile.release]
 codegen-units = 1
 
+# We now compile heed without the NDEBUG define for better performance.
+# However, we still enable debug assertions for a better detection of
+# disk corruption on the cloud or in OSS.
+[profile.release.package.heed]
+debug-assertions = true
+
 [profile.dev.package.flate2]
 opt-level = 3
 

Dockerfile

@@ -1,5 +1,5 @@
 # Compile
-FROM    rust:1.81.0-alpine3.20 AS compiler
+FROM    rust:1.85-alpine3.20 AS compiler
 
 RUN     apk add -q --no-cache build-base openssl-dev
 

README.md

@@ -20,7 +20,7 @@
 <p align="center">
   <a href="https://deps.rs/repo/github/meilisearch/meilisearch"><img src="https://deps.rs/repo/github/meilisearch/meilisearch/status.svg" alt="Dependency status"></a>
   <a href="https://github.com/meilisearch/meilisearch/blob/main/LICENSE"><img src="https://img.shields.io/badge/license-MIT-informational" alt="License"></a>
-  <a href="https://ms-bors.herokuapp.com/repositories/52"><img src="https://bors.tech/images/badge_small.svg" alt="Bors enabled"></a>
+  <a href="https://github.com/meilisearch/meilisearch/queue"><img alt="Merge Queues enabled" src="https://img.shields.io/badge/Merge_Queues-enabled-%2357cf60?logo=github"></a>
 </p>
 
 <p align="center">⚡ A lightning-fast search engine that fits effortlessly into your apps, websites, and workflow 🔍</p>
@@ -41,7 +41,7 @@
 - [**Movies**](https://where2watch.meilisearch.com/?utm_campaign=oss&utm_source=github&utm_medium=organization) — An application to help you find streaming platforms to watch movies using [hybrid search](https://www.meilisearch.com/solutions/hybrid-search?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=demos).
 - [**Ecommerce**](https://ecommerce.meilisearch.com/?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=demos) — Ecommerce website using disjunctive [facets](https://www.meilisearch.com/docs/learn/fine_tuning_results/faceted_search?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=demos), range and rating filtering, and pagination.
 - [**Songs**](https://music.meilisearch.com/?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=demos) — Search through 47 million of songs.
-- [**SaaS**](https://saas.meilisearch.com/?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=demos) — Search for contacts, deals, and companies in this [multi-tenant](https://www.meilisearch.com/docs/learn/security/multitenancy_tenant_tokens?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=demos) CRM application.
+- [**SaaS**](https://saas.meilisearch.com/?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=demos) — Search for contacts, deals, and companies in this [multi-tenant](https://www.meilisearch.com/docs/learn/security/multitenancy_tenant_tokens?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=demos) CRM application.
 
 See the list of all our example apps in our [demos repository](https://github.com/meilisearch/demos).
 
@@ -99,7 +99,7 @@ If you want to know more about the kind of data we collect and what we use it fo
 
 ## 📫 Get in touch!
 
-Meilisearch is a search engine created by [Meili]([https://www.welcometothejungle.com/en/companies/meilisearch](https://www.meilisearch.com/careers)), a software development company headquartered in France and with team members all over the world. Want to know more about us? [Check out our blog!](https://blog.meilisearch.com/?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=contact)
+Meilisearch is a search engine created by [Meili](https://www.meilisearch.com/careers), a software development company headquartered in France and with team members all over the world. Want to know more about us? [Check out our blog!](https://blog.meilisearch.com/?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=contact)
 
 🗞 [Subscribe to our newsletter](https://meilisearch.us2.list-manage.com/subscribe?u=27870f7b71c908a8b359599fb&id=79582d828e) if you don't want to miss any updates! We promise we won't clutter your mailbox: we only send one edition every two months.
 

assets/grafana-dashboard.json

@@ -1501,6 +1501,300 @@
       "title": "Task queue latency",
       "type": "timeseries"
     },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "${DS_PROMETHEUS}"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "palette-classic"
+          },
+          "custom": {
+            "axisBorderShow": false,
+            "axisCenteredZero": false,
+            "axisColorMode": "text",
+            "axisLabel": "",
+            "axisPlacement": "auto",
+            "barAlignment": 0,
+            "drawStyle": "line",
+            "fillOpacity": 15,
+            "gradientMode": "none",
+            "hideFrom": {
+              "legend": false,
+              "tooltip": false,
+              "viz": false
+            },
+            "insertNulls": false,
+            "lineInterpolation": "linear",
+            "lineWidth": 1,
+            "pointSize": 5,
+            "scaleDistribution": {
+              "type": "linear"
+            },
+            "showPoints": "never",
+            "spanNulls": false,
+            "stacking": {
+              "group": "A",
+              "mode": "none"
+            },
+            "thresholdsStyle": {
+              "mode": "off"
+            }
+          },
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green"
+              },
+              {
+                "color": "red",
+                "value": 80
+              }
+            ]
+          },
+          "unit": "none"
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 11,
+        "w": 12,
+        "x": 12,
+        "y": 51
+      },
+      "id": 29,
+      "interval": "5s",
+      "options": {
+        "legend": {
+          "calcs": [],
+          "displayMode": "list",
+          "placement": "right",
+          "showLegend": true
+        },
+        "tooltip": {
+          "mode": "single",
+          "sort": "none"
+        }
+      },
+      "pluginVersion": "8.1.4",
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "${DS_PROMETHEUS}"
+          },
+          "editorMode": "builder",
+          "exemplar": true,
+          "expr": "meilisearch_task_queue_used_size{instance=\"$instance\", job=\"$job\"}",
+          "interval": "",
+          "legendFormat": "{{value}} ",
+          "range": true,
+          "refId": "A"
+        }
+      ],
+      "title": "Task queue used size in bytes",
+      "type": "timeseries"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "${DS_PROMETHEUS}"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "palette-classic"
+          },
+          "custom": {
+            "axisBorderShow": false,
+            "axisCenteredZero": false,
+            "axisColorMode": "text",
+            "axisLabel": "",
+            "axisPlacement": "auto",
+            "barAlignment": 0,
+            "drawStyle": "line",
+            "fillOpacity": 15,
+            "gradientMode": "none",
+            "hideFrom": {
+              "legend": false,
+              "tooltip": false,
+              "viz": false
+            },
+            "insertNulls": false,
+            "lineInterpolation": "linear",
+            "lineWidth": 1,
+            "pointSize": 5,
+            "scaleDistribution": {
+              "type": "linear"
+            },
+            "showPoints": "never",
+            "spanNulls": false,
+            "stacking": {
+              "group": "A",
+              "mode": "none"
+            },
+            "thresholdsStyle": {
+              "mode": "off"
+            }
+          },
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green"
+              },
+              {
+                "color": "red",
+                "value": 80
+              }
+            ]
+          },
+          "unit": "none"
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 11,
+        "w": 12,
+        "x": 12,
+        "y": 51
+      },
+      "id": 29,
+      "interval": "5s",
+      "options": {
+        "legend": {
+          "calcs": [],
+          "displayMode": "list",
+          "placement": "right",
+          "showLegend": true
+        },
+        "tooltip": {
+          "mode": "single",
+          "sort": "none"
+        }
+      },
+      "pluginVersion": "8.1.4",
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "${DS_PROMETHEUS}"
+          },
+          "editorMode": "builder",
+          "exemplar": true,
+          "expr": "meilisearch_task_queue_size_until_stop_registering{instance=\"$instance\", job=\"$job\"}",
+          "interval": "",
+          "legendFormat": "{{value}} ",
+          "range": true,
+          "refId": "A"
+        }
+      ],
+      "title": "Task queue available size until it stop receiving tasks.",
+      "type": "timeseries"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "${DS_PROMETHEUS}"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "palette-classic"
+          },
+          "custom": {
+            "axisBorderShow": false,
+            "axisCenteredZero": false,
+            "axisColorMode": "text",
+            "axisLabel": "",
+            "axisPlacement": "auto",
+            "barAlignment": 0,
+            "drawStyle": "line",
+            "fillOpacity": 15,
+            "gradientMode": "none",
+            "hideFrom": {
+              "legend": false,
+              "tooltip": false,
+              "viz": false
+            },
+            "insertNulls": false,
+            "lineInterpolation": "linear",
+            "lineWidth": 1,
+            "pointSize": 5,
+            "scaleDistribution": {
+              "type": "linear"
+            },
+            "showPoints": "never",
+            "spanNulls": false,
+            "stacking": {
+              "group": "A",
+              "mode": "none"
+            },
+            "thresholdsStyle": {
+              "mode": "off"
+            }
+          },
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green"
+              },
+              {
+                "color": "red",
+                "value": 80
+              }
+            ]
+          },
+          "unit": "none"
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 11,
+        "w": 12,
+        "x": 12,
+        "y": 51
+      },
+      "id": 29,
+      "interval": "5s",
+      "options": {
+        "legend": {
+          "calcs": [],
+          "displayMode": "list",
+          "placement": "right",
+          "showLegend": true
+        },
+        "tooltip": {
+          "mode": "single",
+          "sort": "none"
+        }
+      },
+      "pluginVersion": "8.1.4",
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "${DS_PROMETHEUS}"
+          },
+          "editorMode": "builder",
+          "exemplar": true,
+          "expr": "meilisearch_task_queue_max_size{instance=\"$instance\", job=\"$job\"}",
+          "interval": "",
+          "legendFormat": "{{value}} ",
+          "range": true,
+          "refId": "A"
+        }
+      ],
+      "title": "Task queue maximum possible size",
+      "type": "stat"
+    },
     {
       "collapsed": true,
       "datasource": {

bors.toml

@@ -1,11 +0,0 @@
-status = [
-    'Tests on ubuntu-20.04',
-    'Tests on macos-13',
-    'Tests on windows-2022',
-    'Run Clippy',
-    'Run Rustfmt',
-    'Run tests in debug',
-]
-pr_status = ['Milestone Check']
-# 3 hours timeout
-timeout-sec = 10800

crates/benchmarks/Cargo.toml

@@ -11,27 +11,27 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-anyhow = "1.0.95"
-bumpalo = "3.16.0"
+anyhow = "1.0.98"
+bumpalo = "3.18.1"
 csv = "1.3.1"
 memmap2 = "0.9.5"
 milli = { path = "../milli" }
-mimalloc = { version = "0.1.43", default-features = false }
-serde_json = { version = "1.0.135", features = ["preserve_order"] }
-tempfile = "3.15.0"
+mimalloc = { version = "0.1.47", default-features = false }
+serde_json = { version = "1.0.140", features = ["preserve_order"] }
+tempfile = "3.20.0"
 
 [dev-dependencies]
-criterion = { version = "0.5.1", features = ["html_reports"] }
+criterion = { version = "0.6.0", features = ["html_reports"] }
 rand = "0.8.5"
 rand_chacha = "0.3.1"
-roaring = "0.10.10"
+roaring = "0.10.12"
 
 [build-dependencies]
-anyhow = "1.0.95"
-bytes = "1.9.0"
-convert_case = "0.6.0"
-flate2 = "1.0.35"
-reqwest = { version = "0.12.12", features = ["blocking", "rustls-tls"], default-features = false }
+anyhow = "1.0.98"
+bytes = "1.10.1"
+convert_case = "0.8.0"
+flate2 = "1.1.2"
+reqwest = { version = "0.12.20", features = ["blocking", "rustls-tls"], default-features = false }
 
 [features]
 default = ["milli/all-tokenizations"]

crates/benchmarks/benches/indexing.rs

@@ -10,9 +10,9 @@ use milli::documents::PrimaryKey;
 use milli::heed::{EnvOpenOptions, RwTxn};
 use milli::progress::Progress;
 use milli::update::new::indexer;
-use milli::update::{IndexDocumentsMethod, IndexerConfig, Settings};
-use milli::vector::EmbeddingConfigs;
-use milli::Index;
+use milli::update::{IndexerConfig, Settings};
+use milli::vector::RuntimeEmbedders;
+use milli::{FilterableAttributesRule, Index};
 use rand::seq::SliceRandom;
 use rand_chacha::rand_core::SeedableRng;
 use roaring::RoaringBitmap;
@@ -35,7 +35,8 @@ fn setup_dir(path: impl AsRef<Path>) {
 fn setup_index() -> Index {
     let path = "benches.mmdb";
     setup_dir(path);
-    let mut options = EnvOpenOptions::new();
+    let options = EnvOpenOptions::new();
+    let mut options = options.read_txn_without_tls();
     options.map_size(100 * 1024 * 1024 * 1024); // 100 GB
     options.max_readers(100);
     Index::new(options, path, true).unwrap()
@@ -57,13 +58,14 @@ fn setup_settings<'t>(
     let searchable_fields = searchable_fields.iter().map(|s| s.to_string()).collect();
     builder.set_searchable_fields(searchable_fields);
 
-    let filterable_fields = filterable_fields.iter().map(|s| s.to_string()).collect();
+    let filterable_fields =
+        filterable_fields.iter().map(|s| FilterableAttributesRule::Field(s.to_string())).collect();
     builder.set_filterable_fields(filterable_fields);
 
     let sortable_fields = sortable_fields.iter().map(|s| s.to_string()).collect();
     builder.set_sortable_fields(sortable_fields);
 
-    builder.execute(|_| (), || false).unwrap();
+    builder.execute(&|| false, &Progress::default(), Default::default()).unwrap();
 }
 
 fn setup_index_with_settings(
@@ -138,10 +140,9 @@ fn indexing_songs_default(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::SMOL_SONGS, "csv");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -165,9 +166,10 @@ fn indexing_songs_default(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -205,10 +207,9 @@ fn reindexing_songs_default(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::SMOL_SONGS, "csv");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -232,9 +233,10 @@ fn reindexing_songs_default(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -250,10 +252,9 @@ fn reindexing_songs_default(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::SMOL_SONGS, "csv");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -277,9 +278,10 @@ fn reindexing_songs_default(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -319,10 +321,9 @@ fn deleting_songs_in_batches_default(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::SMOL_SONGS, "csv");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -346,9 +347,10 @@ fn deleting_songs_in_batches_default(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -396,10 +398,9 @@ fn indexing_songs_in_three_batches_default(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::SMOL_SONGS_1_2, "csv");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -423,9 +424,10 @@ fn indexing_songs_in_three_batches_default(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -441,10 +443,9 @@ fn indexing_songs_in_three_batches_default(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::SMOL_SONGS_3_4, "csv");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -468,9 +469,10 @@ fn indexing_songs_in_three_batches_default(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -482,10 +484,9 @@ fn indexing_songs_in_three_batches_default(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::SMOL_SONGS_4_4, "csv");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -509,9 +510,10 @@ fn indexing_songs_in_three_batches_default(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -549,11 +551,10 @@ fn indexing_songs_without_faceted_numbers(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::SMOL_SONGS, "csv");
 
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -577,9 +578,10 @@ fn indexing_songs_without_faceted_numbers(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -617,10 +619,9 @@ fn indexing_songs_without_faceted_fields(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::SMOL_SONGS, "csv");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -644,9 +645,10 @@ fn indexing_songs_without_faceted_fields(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -684,10 +686,9 @@ fn indexing_wiki(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::SMOL_WIKI_ARTICLES, "csv");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -711,9 +712,10 @@ fn indexing_wiki(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -750,10 +752,9 @@ fn reindexing_wiki(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::SMOL_WIKI_ARTICLES, "csv");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -777,9 +778,10 @@ fn reindexing_wiki(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -795,10 +797,9 @@ fn reindexing_wiki(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::SMOL_WIKI_ARTICLES, "csv");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -822,9 +823,10 @@ fn reindexing_wiki(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -863,10 +865,9 @@ fn deleting_wiki_in_batches_default(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::SMOL_WIKI_ARTICLES, "csv");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -890,9 +891,10 @@ fn deleting_wiki_in_batches_default(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -939,11 +941,10 @@ fn indexing_wiki_in_three_batches(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents =
                     utils::documents_from(datasets_paths::SMOL_WIKI_ARTICLES_1_2, "csv");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -967,9 +968,10 @@ fn indexing_wiki_in_three_batches(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -985,11 +987,10 @@ fn indexing_wiki_in_three_batches(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents =
                     utils::documents_from(datasets_paths::SMOL_WIKI_ARTICLES_3_4, "csv");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -1013,9 +1014,10 @@ fn indexing_wiki_in_three_batches(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -1027,11 +1029,10 @@ fn indexing_wiki_in_three_batches(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents =
                     utils::documents_from(datasets_paths::SMOL_WIKI_ARTICLES_4_4, "csv");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -1055,9 +1056,10 @@ fn indexing_wiki_in_three_batches(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -1095,10 +1097,9 @@ fn indexing_movies_default(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::MOVIES, "json");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -1122,9 +1123,10 @@ fn indexing_movies_default(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -1161,10 +1163,9 @@ fn reindexing_movies_default(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::MOVIES, "json");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -1188,9 +1189,10 @@ fn reindexing_movies_default(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -1206,10 +1208,9 @@ fn reindexing_movies_default(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::MOVIES, "json");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -1233,9 +1234,10 @@ fn reindexing_movies_default(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -1274,10 +1276,9 @@ fn deleting_movies_in_batches_default(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::MOVIES, "json");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -1301,9 +1302,10 @@ fn deleting_movies_in_batches_default(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -1349,9 +1351,10 @@ fn delete_documents_from_ids(index: Index, document_ids_to_delete: Vec<RoaringBi
             new_fields_ids_map,
             Some(primary_key),
             &document_changes,
-            EmbeddingConfigs::default(),
+            RuntimeEmbedders::default(),
             &|| false,
             &Progress::default(),
+            &Default::default(),
         )
         .unwrap();
 
@@ -1387,10 +1390,9 @@ fn indexing_movies_in_three_batches(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::MOVIES_1_2, "json");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -1414,9 +1416,10 @@ fn indexing_movies_in_three_batches(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -1432,10 +1435,9 @@ fn indexing_movies_in_three_batches(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::MOVIES_3_4, "json");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -1459,9 +1461,10 @@ fn indexing_movies_in_three_batches(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -1473,10 +1476,9 @@ fn indexing_movies_in_three_batches(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::MOVIES_4_4, "json");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -1500,9 +1502,10 @@ fn indexing_movies_in_three_batches(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -1563,10 +1566,9 @@ fn indexing_nested_movies_default(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::NESTED_MOVIES, "json");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -1590,9 +1592,10 @@ fn indexing_nested_movies_default(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -1654,10 +1657,9 @@ fn deleting_nested_movies_in_batches_default(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::NESTED_MOVIES, "json");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -1681,9 +1683,10 @@ fn deleting_nested_movies_in_batches_default(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -1737,10 +1740,9 @@ fn indexing_nested_movies_without_faceted_fields(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::NESTED_MOVIES, "json");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -1764,9 +1766,10 @@ fn indexing_nested_movies_without_faceted_fields(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -1804,10 +1807,9 @@ fn indexing_geo(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::SMOL_ALL_COUNTRIES, "jsonl");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -1831,9 +1833,10 @@ fn indexing_geo(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -1870,10 +1873,9 @@ fn reindexing_geo(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::SMOL_ALL_COUNTRIES, "jsonl");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -1897,9 +1899,10 @@ fn reindexing_geo(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -1915,10 +1918,9 @@ fn reindexing_geo(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::SMOL_ALL_COUNTRIES, "jsonl");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -1942,9 +1944,10 @@ fn reindexing_geo(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 
@@ -1983,10 +1986,9 @@ fn deleting_geo_in_batches_default(c: &mut Criterion) {
                 let db_fields_ids_map = index.fields_ids_map(&rtxn).unwrap();
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
-                let mut indexer =
-                    indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
+                let mut indexer = indexer::DocumentOperation::new();
                 let documents = utils::documents_from(datasets_paths::SMOL_ALL_COUNTRIES, "jsonl");
-                indexer.add_documents(&documents).unwrap();
+                indexer.replace_documents(&documents).unwrap();
 
                 let indexer_alloc = Bump::new();
                 let (document_changes, _operation_stats, primary_key) = indexer
@@ -2010,9 +2012,10 @@ fn deleting_geo_in_batches_default(c: &mut Criterion) {
                     new_fields_ids_map,
                     primary_key,
                     &document_changes,
-                    EmbeddingConfigs::default(),
+                    RuntimeEmbedders::default(),
                     &|| false,
                     &Progress::default(),
+                    &Default::default(),
                 )
                 .unwrap();
 

crates/benchmarks/benches/search_geo.rs

@@ -3,6 +3,7 @@ mod utils;
 
 use criterion::{criterion_group, criterion_main};
 use milli::update::Settings;
+use milli::FilterableAttributesRule;
 use utils::Conf;
 
 #[cfg(not(windows))]
@@ -21,8 +22,10 @@ fn base_conf(builder: &mut Settings) {
         ["name", "alternatenames", "elevation"].iter().map(|s| s.to_string()).collect();
     builder.set_searchable_fields(searchable_fields);
 
-    let filterable_fields =
-        ["_geo", "population", "elevation"].iter().map(|s| s.to_string()).collect();
+    let filterable_fields = ["_geo", "population", "elevation"]
+        .iter()
+        .map(|s| FilterableAttributesRule::Field(s.to_string()))
+        .collect();
     builder.set_filterable_fields(filterable_fields);
 
     let sortable_fields =

crates/benchmarks/benches/search_songs.rs

@@ -3,6 +3,7 @@ mod utils;
 
 use criterion::{criterion_group, criterion_main};
 use milli::update::Settings;
+use milli::FilterableAttributesRule;
 use utils::Conf;
 
 #[cfg(not(windows))]
@@ -22,7 +23,7 @@ fn base_conf(builder: &mut Settings) {
 
     let faceted_fields = ["released-timestamp", "duration-float", "genre", "country", "artist"]
         .iter()
-        .map(|s| s.to_string())
+        .map(|s| FilterableAttributesRule::Field(s.to_string()))
         .collect();
     builder.set_filterable_fields(faceted_fields);
 }

crates/benchmarks/benches/utils.rs

@@ -12,8 +12,8 @@ use memmap2::Mmap;
 use milli::heed::EnvOpenOptions;
 use milli::progress::Progress;
 use milli::update::new::indexer;
-use milli::update::{IndexDocumentsMethod, IndexerConfig, Settings};
-use milli::vector::EmbeddingConfigs;
+use milli::update::{IndexerConfig, Settings};
+use milli::vector::RuntimeEmbedders;
 use milli::{Criterion, Filter, Index, Object, TermsMatchingStrategy};
 use serde_json::Value;
 
@@ -65,7 +65,8 @@ pub fn base_setup(conf: &Conf) -> Index {
     }
     create_dir_all(conf.database_name).unwrap();
 
-    let mut options = EnvOpenOptions::new();
+    let options = EnvOpenOptions::new();
+    let mut options = options.read_txn_without_tls();
     options.map_size(100 * 1024 * 1024 * 1024); // 100 GB
     options.max_readers(100);
     let index = Index::new(options, conf.database_name, true).unwrap();
@@ -89,7 +90,7 @@ pub fn base_setup(conf: &Conf) -> Index {
 
     (conf.configure)(&mut builder);
 
-    builder.execute(|_| (), || false).unwrap();
+    builder.execute(&|| false, &Progress::default(), Default::default()).unwrap();
     wtxn.commit().unwrap();
 
     let config = IndexerConfig::default();
@@ -99,8 +100,8 @@ pub fn base_setup(conf: &Conf) -> Index {
     let mut new_fields_ids_map = db_fields_ids_map.clone();
 
     let documents = documents_from(conf.dataset, conf.dataset_format);
-    let mut indexer = indexer::DocumentOperation::new(IndexDocumentsMethod::ReplaceDocuments);
-    indexer.add_documents(&documents).unwrap();
+    let mut indexer = indexer::DocumentOperation::new();
+    indexer.replace_documents(&documents).unwrap();
 
     let indexer_alloc = Bump::new();
     let (document_changes, _operation_stats, primary_key) = indexer
@@ -124,9 +125,10 @@ pub fn base_setup(conf: &Conf) -> Index {
         new_fields_ids_map,
         primary_key,
         &document_changes,
-        EmbeddingConfigs::default(),
+        RuntimeEmbedders::default(),
         &|| false,
         &Progress::default(),
+        &Default::default(),
     )
     .unwrap();
 

crates/benchmarks/build.rs

@@ -67,7 +67,7 @@ fn main() -> anyhow::Result<()> {
         writeln!(
             &mut manifest_paths_file,
             r#"pub const {}: &str = {:?};"#,
-            dataset.to_case(Case::ScreamingSnake),
+            dataset.to_case(Case::UpperSnake),
             out_file.display(),
         )?;
 

crates/build-info/Cargo.toml

@@ -11,8 +11,8 @@ license.workspace = true
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html
 
 [dependencies]
-time = { version = "0.3.37", features = ["parsing"] }
+time = { version = "0.3.41", features = ["parsing"] }
 
 [build-dependencies]
-anyhow = "1.0.95"
-vergen-git2 = "1.0.2"
+anyhow = "1.0.98"
+vergen-git2 = "1.0.7"

crates/dump/Cargo.toml

@@ -11,21 +11,21 @@ readme.workspace = true
 license.workspace = true
 
 [dependencies]
-anyhow = "1.0.95"
-flate2 = "1.0.35"
-http = "1.2.0"
+anyhow = "1.0.98"
+flate2 = "1.1.2"
+http = "1.3.1"
 meilisearch-types = { path = "../meilisearch-types" }
-once_cell = "1.20.2"
+once_cell = "1.21.3"
 regex = "1.11.1"
-roaring = { version = "0.10.10", features = ["serde"] }
-serde = { version = "1.0.217", features = ["derive"] }
-serde_json = { version = "1.0.135", features = ["preserve_order"] }
-tar = "0.4.43"
-tempfile = "3.15.0"
-thiserror = "2.0.9"
-time = { version = "0.3.37", features = ["serde-well-known", "formatting", "parsing", "macros"] }
+roaring = { version = "0.10.12", features = ["serde"] }
+serde = { version = "1.0.219", features = ["derive"] }
+serde_json = { version = "1.0.140", features = ["preserve_order"] }
+tar = "0.4.44"
+tempfile = "3.20.0"
+thiserror = "2.0.12"
+time = { version = "0.3.41", features = ["serde-well-known", "formatting", "parsing", "macros"] }
 tracing = "0.1.41"
-uuid = { version = "1.11.0", features = ["serde", "v4"] }
+uuid = { version = "1.17.0", features = ["serde", "v4"] }
 
 [dev-dependencies]
 big_s = "1.0.2"

crates/dump/src/lib.rs

@@ -1,12 +1,17 @@
 #![allow(clippy::type_complexity)]
 #![allow(clippy::wrong_self_convention)]
 
+use std::collections::BTreeMap;
+
 use meilisearch_types::batches::BatchId;
+use meilisearch_types::byte_unit::Byte;
 use meilisearch_types::error::ResponseError;
 use meilisearch_types::keys::Key;
 use meilisearch_types::milli::update::IndexDocumentsMethod;
 use meilisearch_types::settings::Unchecked;
-use meilisearch_types::tasks::{Details, IndexSwap, KindWithContent, Status, Task, TaskId};
+use meilisearch_types::tasks::{
+    Details, ExportIndexSettings, IndexSwap, KindWithContent, Status, Task, TaskId,
+};
 use meilisearch_types::InstanceUid;
 use roaring::RoaringBitmap;
 use serde::{Deserialize, Serialize};
@@ -141,6 +146,12 @@ pub enum KindDump {
         instance_uid: Option<InstanceUid>,
     },
     SnapshotCreation,
+    Export {
+        url: String,
+        api_key: Option<String>,
+        payload_size: Option<Byte>,
+        indexes: BTreeMap<String, ExportIndexSettings>,
+    },
     UpgradeDatabase {
         from: (u32, u32, u32),
     },
@@ -213,6 +224,15 @@ impl From<KindWithContent> for KindDump {
                 KindDump::DumpCreation { keys, instance_uid }
             }
             KindWithContent::SnapshotCreation => KindDump::SnapshotCreation,
+            KindWithContent::Export { url, api_key, payload_size, indexes } => KindDump::Export {
+                url,
+                api_key,
+                payload_size,
+                indexes: indexes
+                    .into_iter()
+                    .map(|(pattern, settings)| (pattern.to_string(), settings))
+                    .collect(),
+            },
             KindWithContent::UpgradeDatabase { from: version } => {
                 KindDump::UpgradeDatabase { from: version }
             }
@@ -233,11 +253,11 @@ pub(crate) mod test {
     use meilisearch_types::features::{Network, Remote, RuntimeTogglableFeatures};
     use meilisearch_types::index_uid_pattern::IndexUidPattern;
     use meilisearch_types::keys::{Action, Key};
-    use meilisearch_types::milli;
     use meilisearch_types::milli::update::Setting;
+    use meilisearch_types::milli::{self, FilterableAttributesRule};
     use meilisearch_types::settings::{Checked, FacetingSettings, Settings};
     use meilisearch_types::task_view::DetailsView;
-    use meilisearch_types::tasks::{Details, Kind, Status};
+    use meilisearch_types::tasks::{BatchStopReason, Details, Kind, Status};
     use serde_json::{json, Map, Value};
     use time::macros::datetime;
     use uuid::Uuid;
@@ -279,7 +299,10 @@ pub(crate) mod test {
         let settings = Settings {
             displayed_attributes: Setting::Set(vec![S("race"), S("name")]).into(),
             searchable_attributes: Setting::Set(vec![S("name"), S("race")]).into(),
-            filterable_attributes: Setting::Set(btreeset! { S("race"), S("age") }),
+            filterable_attributes: Setting::Set(vec![
+                FilterableAttributesRule::Field(S("race")),
+                FilterableAttributesRule::Field(S("age")),
+            ]),
             sortable_attributes: Setting::Set(btreeset! { S("age") }),
             ranking_rules: Setting::NotSet,
             stop_words: Setting::NotSet,
@@ -302,6 +325,7 @@ pub(crate) mod test {
             localized_attributes: Setting::NotSet,
             facet_search: Setting::NotSet,
             prefix_search: Setting::NotSet,
+            chat: Setting::NotSet,
             _kind: std::marker::PhantomData,
         };
         settings.check()
@@ -321,13 +345,18 @@ pub(crate) mod test {
                 status: maplit::btreemap! { Status::Succeeded => 1 },
                 types: maplit::btreemap! { Kind::DocumentAdditionOrUpdate => 1 },
                 index_uids: maplit::btreemap! { "doggo".to_string() => 1 },
+                progress_trace: Default::default(),
+                write_channel_congestion: None,
+                internal_database_sizes: Default::default(),
             },
+            embedder_stats: Default::default(),
             enqueued_at: Some(BatchEnqueuedAt {
                 earliest: datetime!(2022-11-11 0:00 UTC),
                 oldest: datetime!(2022-11-11 0:00 UTC),
             }),
             started_at: datetime!(2022-11-20 0:00 UTC),
             finished_at: Some(datetime!(2022-11-21 0:00 UTC)),
+            stop_reason: BatchStopReason::Unspecified.to_string(),
         }]
     }
 

crates/dump/src/reader/compat/v5_to_v6.rs

@@ -1,3 +1,4 @@
+use std::num::NonZeroUsize;
 use std::str::FromStr;
 
 use super::v4_to_v5::{CompatIndexV4ToV5, CompatV4ToV5};
@@ -322,7 +323,16 @@ impl<T> From<v5::Settings<T>> for v6::Settings<v6::Unchecked> {
         v6::Settings {
             displayed_attributes: v6::Setting::from(settings.displayed_attributes).into(),
             searchable_attributes: v6::Setting::from(settings.searchable_attributes).into(),
-            filterable_attributes: settings.filterable_attributes.into(),
+            filterable_attributes: match settings.filterable_attributes {
+                v5::settings::Setting::Set(filterable_attributes) => v6::Setting::Set(
+                    filterable_attributes
+                        .into_iter()
+                        .map(v6::FilterableAttributesRule::Field)
+                        .collect(),
+                ),
+                v5::settings::Setting::Reset => v6::Setting::Reset,
+                v5::settings::Setting::NotSet => v6::Setting::NotSet,
+            },
             sortable_attributes: settings.sortable_attributes.into(),
             ranking_rules: {
                 match settings.ranking_rules {
@@ -364,6 +374,7 @@ impl<T> From<v5::Settings<T>> for v6::Settings<v6::Unchecked> {
                     },
                     disable_on_words: typo.disable_on_words.into(),
                     disable_on_attributes: typo.disable_on_attributes.into(),
+                    disable_on_numbers: v6::Setting::NotSet,
                 }),
                 v5::Setting::Reset => v6::Setting::Reset,
                 v5::Setting::NotSet => v6::Setting::NotSet,
@@ -378,7 +389,13 @@ impl<T> From<v5::Settings<T>> for v6::Settings<v6::Unchecked> {
             },
             pagination: match settings.pagination {
                 v5::Setting::Set(pagination) => v6::Setting::Set(v6::PaginationSettings {
-                    max_total_hits: pagination.max_total_hits.into(),
+                    max_total_hits: match pagination.max_total_hits {
+                        v5::Setting::Set(max_total_hits) => v6::Setting::Set(
+                            max_total_hits.try_into().unwrap_or(NonZeroUsize::new(1).unwrap()),
+                        ),
+                        v5::Setting::Reset => v6::Setting::Reset,
+                        v5::Setting::NotSet => v6::Setting::NotSet,
+                    },
                 }),
                 v5::Setting::Reset => v6::Setting::Reset,
                 v5::Setting::NotSet => v6::Setting::NotSet,
@@ -388,6 +405,7 @@ impl<T> From<v5::Settings<T>> for v6::Settings<v6::Unchecked> {
             search_cutoff_ms: v6::Setting::NotSet,
             facet_search: v6::Setting::NotSet,
             prefix_search: v6::Setting::NotSet,
+            chat: v6::Setting::NotSet,
             _kind: std::marker::PhantomData,
         }
     }

crates/dump/src/reader/mod.rs

@@ -116,6 +116,15 @@ impl DumpReader {
         }
     }
 
+    pub fn chat_completions_settings(
+        &mut self,
+    ) -> Result<Box<dyn Iterator<Item = Result<(String, v6::ChatCompletionSettings)>> + '_>> {
+        match self {
+            DumpReader::Current(current) => current.chat_completions_settings(),
+            DumpReader::Compat(_compat) => Ok(Box::new(std::iter::empty())),
+        }
+    }
+
     pub fn features(&self) -> Result<Option<v6::RuntimeTogglableFeatures>> {
         match self {
             DumpReader::Current(current) => Ok(current.features()),

crates/dump/src/reader/snapshots/dump__reader__test__import_dump_v6_with_vectors-5.snap

@@ -1,5 +1,5 @@
 ---
-source: dump/src/reader/mod.rs
+source: crates/dump/src/reader/mod.rs
 expression: vector_index.settings().unwrap()
 ---
 {
@@ -49,6 +49,7 @@ expression: vector_index.settings().unwrap()
       "source": "huggingFace",
       "model": "BAAI/bge-base-en-v1.5",
       "revision": "617ca489d9e86b49b8167676d8220688b99db36e",
+      "pooling": "forceMean",
       "documentTemplate": "{% for field in fields %} {{ field.name }}: {{ field.value }}\n{% endfor %}"
     }
   },

crates/dump/src/reader/v4/meta.rs

@@ -108,7 +108,7 @@ where
         /// not supported on untagged enums.
         struct StarOrVisitor<T>(PhantomData<T>);
 
-        impl<'de, T, FE> Visitor<'de> for StarOrVisitor<T>
+        impl<T, FE> Visitor<'_> for StarOrVisitor<T>
         where
             T: FromStr<Err = FE>,
             FE: Display,

crates/dump/src/reader/v4/tasks.rs

@@ -99,7 +99,7 @@ impl Task {
     /// Return true when a task is finished.
     /// A task is finished when its last state is either `Succeeded` or `Failed`.
     pub fn is_finished(&self) -> bool {
-        self.events.last().map_or(false, |event| {
+        self.events.last().is_some_and(|event| {
             matches!(event, TaskEvent::Succeded { .. } | TaskEvent::Failed { .. })
         })
     }

crates/dump/src/reader/v5/meta.rs

@@ -108,7 +108,7 @@ where
         /// not supported on untagged enums.
         struct StarOrVisitor<T>(PhantomData<T>);
 
-        impl<'de, T, FE> Visitor<'de> for StarOrVisitor<T>
+        impl<T, FE> Visitor<'_> for StarOrVisitor<T>
         where
             T: FromStr<Err = FE>,
             FE: Display,

crates/dump/src/reader/v5/tasks.rs

@@ -114,7 +114,7 @@ impl Task {
     /// Return true when a task is finished.
     /// A task is finished when its last state is either `Succeeded` or `Failed`.
     pub fn is_finished(&self) -> bool {
-        self.events.last().map_or(false, |event| {
+        self.events.last().is_some_and(|event| {
             matches!(event, TaskEvent::Succeeded { .. } | TaskEvent::Failed { .. })
         })
     }

crates/dump/src/reader/v6/mod.rs

@@ -1,8 +1,10 @@
+use std::ffi::OsStr;
 use std::fs::{self, File};
 use std::io::{BufRead, BufReader, ErrorKind};
 use std::path::Path;
 
 pub use meilisearch_types::milli;
+use meilisearch_types::milli::vector::hf::OverridePooling;
 use tempfile::TempDir;
 use time::OffsetDateTime;
 use tracing::debug;
@@ -20,6 +22,7 @@ pub type Unchecked = meilisearch_types::settings::Unchecked;
 pub type Task = crate::TaskDump;
 pub type Batch = meilisearch_types::batches::Batch;
 pub type Key = meilisearch_types::keys::Key;
+pub type ChatCompletionSettings = meilisearch_types::features::ChatCompletionSettings;
 pub type RuntimeTogglableFeatures = meilisearch_types::features::RuntimeTogglableFeatures;
 pub type Network = meilisearch_types::features::Network;
 
@@ -45,6 +48,8 @@ pub type ResponseError = meilisearch_types::error::ResponseError;
 pub type Code = meilisearch_types::error::Code;
 pub type RankingRuleView = meilisearch_types::settings::RankingRuleView;
 
+pub type FilterableAttributesRule = meilisearch_types::milli::FilterableAttributesRule;
+
 pub struct V6Reader {
     dump: TempDir,
     instance_uid: Option<Uuid>,
@@ -189,6 +194,34 @@ impl V6Reader {
         )
     }
 
+    pub fn chat_completions_settings(
+        &mut self,
+    ) -> Result<Box<dyn Iterator<Item = Result<(String, ChatCompletionSettings)>> + '_>> {
+        let entries = match fs::read_dir(self.dump.path().join("chat-completions-settings")) {
+            Ok(entries) => entries,
+            Err(e) if e.kind() == ErrorKind::NotFound => return Ok(Box::new(std::iter::empty())),
+            Err(e) => return Err(e.into()),
+        };
+        Ok(Box::new(
+            entries
+                .map(|entry| -> Result<Option<_>> {
+                    let entry = entry?;
+                    let file_name = entry.file_name();
+                    let path = Path::new(&file_name);
+                    if entry.file_type()?.is_file() && path.extension() == Some(OsStr::new("json"))
+                    {
+                        let name = path.file_stem().unwrap().to_str().unwrap().to_string();
+                        let file = File::open(entry.path())?;
+                        let settings = serde_json::from_reader(file)?;
+                        Ok(Some((name, settings)))
+                    } else {
+                        Ok(None)
+                    }
+                })
+                .filter_map(|entry| entry.transpose()),
+        ))
+    }
+
     pub fn features(&self) -> Option<RuntimeTogglableFeatures> {
         self.features
     }
@@ -252,7 +285,29 @@ impl V6IndexReader {
     }
 
     pub fn settings(&mut self) -> Result<Settings<Checked>> {
-        let settings: Settings<Unchecked> = serde_json::from_reader(&mut self.settings)?;
+        let mut settings: Settings<Unchecked> = serde_json::from_reader(&mut self.settings)?;
+        patch_embedders(&mut settings);
         Ok(settings.check())
     }
 }
+
+fn patch_embedders(settings: &mut Settings<Unchecked>) {
+    if let Setting::Set(embedders) = &mut settings.embedders {
+        for settings in embedders.values_mut() {
+            let Setting::Set(settings) = &mut settings.inner else {
+                continue;
+            };
+            if settings.source != Setting::Set(milli::vector::settings::EmbedderSource::HuggingFace)
+            {
+                continue;
+            }
+            settings.pooling = match settings.pooling {
+                Setting::Set(pooling) => Setting::Set(pooling),
+                // if the pooling for a hugging face embedder is not set, force it to `forceMean`
+                // for backward compatibility with v1.13
+                // dumps created in v1.14 and up will have the setting set for hugging face embedders
+                Setting::Reset | Setting::NotSet => Setting::Set(OverridePooling::ForceMean),
+            };
+        }
+    }
+}

crates/dump/src/writer.rs

@@ -5,7 +5,7 @@ use std::path::PathBuf;
 use flate2::write::GzEncoder;
 use flate2::Compression;
 use meilisearch_types::batches::Batch;
-use meilisearch_types::features::{Network, RuntimeTogglableFeatures};
+use meilisearch_types::features::{ChatCompletionSettings, Network, RuntimeTogglableFeatures};
 use meilisearch_types::keys::Key;
 use meilisearch_types::settings::{Checked, Settings};
 use serde_json::{Map, Value};
@@ -51,6 +51,10 @@ impl DumpWriter {
         KeyWriter::new(self.dir.path().to_path_buf())
     }
 
+    pub fn create_chat_completions_settings(&self) -> Result<ChatCompletionsSettingsWriter> {
+        ChatCompletionsSettingsWriter::new(self.dir.path().join("chat-completions-settings"))
+    }
+
     pub fn create_tasks_queue(&self) -> Result<TaskWriter> {
         TaskWriter::new(self.dir.path().join("tasks"))
     }
@@ -104,6 +108,24 @@ impl KeyWriter {
     }
 }
 
+pub struct ChatCompletionsSettingsWriter {
+    path: PathBuf,
+}
+
+impl ChatCompletionsSettingsWriter {
+    pub(crate) fn new(path: PathBuf) -> Result<Self> {
+        std::fs::create_dir(&path)?;
+        Ok(ChatCompletionsSettingsWriter { path })
+    }
+
+    pub fn push_settings(&mut self, name: &str, settings: &ChatCompletionSettings) -> Result<()> {
+        let mut settings_file = File::create(self.path.join(name).with_extension("json"))?;
+        serde_json::to_writer(&mut settings_file, &settings)?;
+        settings_file.flush()?;
+        Ok(())
+    }
+}
+
 pub struct TaskWriter {
     queue: BufWriter<File>,
     update_files: PathBuf,

crates/file-store/Cargo.toml

@@ -11,7 +11,7 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-tempfile = "3.15.0"
-thiserror = "2.0.9"
+tempfile = "3.20.0"
+thiserror = "2.0.12"
 tracing = "0.1.41"
-uuid = { version = "1.11.0", features = ["serde", "v4"] }
+uuid = { version = "1.17.0", features = ["serde", "v4"] }

crates/filter-parser/Cargo.toml

@@ -14,7 +14,7 @@ license.workspace = true
 [dependencies]
 nom = "7.1.3"
 nom_locate = "4.2.0"
-unescaper = "0.1.5"
+unescaper = "0.1.6"
 
 [dev-dependencies]
 # fixed version due to format breakages in v1.40

crates/filter-parser/src/condition.rs

@@ -30,6 +30,25 @@ pub enum Condition<'a> {
     StartsWith { keyword: Token<'a>, word: Token<'a> },
 }
 
+impl Condition<'_> {
+    pub fn operator(&self) -> &str {
+        match self {
+            Condition::GreaterThan(_) => ">",
+            Condition::GreaterThanOrEqual(_) => ">=",
+            Condition::Equal(_) => "=",
+            Condition::NotEqual(_) => "!=",
+            Condition::Null => "IS NULL",
+            Condition::Empty => "IS EMPTY",
+            Condition::Exists => "EXISTS",
+            Condition::LowerThan(_) => "<",
+            Condition::LowerThanOrEqual(_) => "<=",
+            Condition::Between { .. } => "TO",
+            Condition::Contains { .. } => "CONTAINS",
+            Condition::StartsWith { .. } => "STARTS WITH",
+        }
+    }
+}
+
 /// condition      = value ("==" | ">" ...) value
 pub fn parse_condition(input: Span) -> IResult<FilterCondition> {
     let operator = alt((tag("<="), tag(">="), tag("!="), tag("<"), tag(">"), tag("=")));

crates/filter-parser/src/error.rs

@@ -35,7 +35,7 @@ impl<E> NomErrorExt<E> for nom::Err<E> {
 pub fn cut_with_err<'a, O>(
     mut parser: impl FnMut(Span<'a>) -> IResult<'a, O>,
     mut with: impl FnMut(Error<'a>) -> Error<'a>,
-) -> impl FnMut(Span<'a>) -> IResult<O> {
+) -> impl FnMut(Span<'a>) -> IResult<'a, O> {
     move |input| match parser.parse(input) {
         Err(nom::Err::Error(e)) => Err(nom::Err::Failure(with(e))),
         rest => rest,
@@ -121,7 +121,7 @@ impl<'a> ParseError<Span<'a>> for Error<'a> {
     }
 }
 
-impl<'a> Display for Error<'a> {
+impl Display for Error<'_> {
     fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
         let input = self.context.fragment();
         // When printing our error message we want to escape all `\n` to be sure we keep our format with the

crates/filter-parser/src/lib.rs

@@ -80,7 +80,7 @@ pub struct Token<'a> {
     value: Option<String>,
 }
 
-impl<'a> PartialEq for Token<'a> {
+impl PartialEq for Token<'_> {
     fn eq(&self, other: &Self) -> bool {
         self.span.fragment() == other.span.fragment()
     }
@@ -226,7 +226,7 @@ impl<'a> FilterCondition<'a> {
         }
     }
 
-    pub fn parse(input: &'a str) -> Result<Option<Self>, Error> {
+    pub fn parse(input: &'a str) -> Result<Option<Self>, Error<'a>> {
         if input.trim().is_empty() {
             return Ok(None);
         }
@@ -527,7 +527,7 @@ pub fn parse_filter(input: Span) -> IResult<FilterCondition> {
     terminated(|input| parse_expression(input, 0), eof)(input)
 }
 
-impl<'a> std::fmt::Display for FilterCondition<'a> {
+impl std::fmt::Display for FilterCondition<'_> {
     fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
         match self {
             FilterCondition::Not(filter) => {
@@ -576,7 +576,8 @@ impl<'a> std::fmt::Display for FilterCondition<'a> {
         }
     }
 }
-impl<'a> std::fmt::Display for Condition<'a> {
+
+impl std::fmt::Display for Condition<'_> {
     fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
         match self {
             Condition::GreaterThan(token) => write!(f, "> {token}"),
@@ -594,7 +595,8 @@ impl<'a> std::fmt::Display for Condition<'a> {
         }
     }
 }
-impl<'a> std::fmt::Display for Token<'a> {
+
+impl std::fmt::Display for Token<'_> {
     fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
         write!(f, "{{{}}}", self.value())
     }

crates/filter-parser/src/value.rs

@@ -52,7 +52,7 @@ fn quoted_by(quote: char, input: Span) -> IResult<Token> {
 }
 
 // word           = (alphanumeric | _ | - | .)+    except for reserved keywords
-pub fn word_not_keyword<'a>(input: Span<'a>) -> IResult<Token<'a>> {
+pub fn word_not_keyword<'a>(input: Span<'a>) -> IResult<'a, Token<'a>> {
     let (input, word): (_, Token<'a>) =
         take_while1(is_value_component)(input).map(|(s, t)| (s, t.into()))?;
     if is_keyword(word.value()) {

crates/flatten-serde-json/Cargo.toml

@@ -16,7 +16,7 @@ license.workspace = true
 serde_json = "1.0"
 
 [dev-dependencies]
-criterion = { version = "0.5.1", features = ["html_reports"] }
+criterion = { version = "0.6.0", features = ["html_reports"] }
 
 [[bench]]
 name = "benchmarks"

crates/fuzzers/Cargo.toml

@@ -12,11 +12,11 @@ license.workspace = true
 
 [dependencies]
 arbitrary = { version = "1.4.1", features = ["derive"] }
-bumpalo = "3.16.0"
-clap = { version = "4.5.24", features = ["derive"] }
-either = "1.13.0"
+bumpalo = "3.18.1"
+clap = { version = "4.5.40", features = ["derive"] }
+either = "1.15.0"
 fastrand = "2.3.0"
 milli = { path = "../milli" }
-serde = { version = "1.0.217", features = ["derive"] }
-serde_json = { version = "1.0.135", features = ["preserve_order"] }
-tempfile = "3.15.0"
+serde = { version = "1.0.219", features = ["derive"] }
+serde_json = { version = "1.0.140", features = ["preserve_order"] }
+tempfile = "3.20.0"

crates/fuzzers/src/bin/fuzz-indexing.rs

@@ -12,8 +12,8 @@ use milli::documents::mmap_from_objects;
 use milli::heed::EnvOpenOptions;
 use milli::progress::Progress;
 use milli::update::new::indexer;
-use milli::update::{IndexDocumentsMethod, IndexerConfig};
-use milli::vector::EmbeddingConfigs;
+use milli::update::IndexerConfig;
+use milli::vector::RuntimeEmbedders;
 use milli::Index;
 use serde_json::Value;
 use tempfile::TempDir;
@@ -57,7 +57,8 @@ fn main() {
         let opt = opt.clone();
 
         let handle = std::thread::spawn(move || {
-            let mut options = EnvOpenOptions::new();
+            let options = EnvOpenOptions::new();
+            let mut options = options.read_txn_without_tls();
             options.map_size(1024 * 1024 * 1024 * 1024);
             let tempdir = match opt.path {
                 Some(path) => TempDir::new_in(path).unwrap(),
@@ -88,10 +89,8 @@ fn main() {
                             let mut new_fields_ids_map = db_fields_ids_map.clone();
 
                             let indexer_alloc = Bump::new();
-                            let embedders = EmbeddingConfigs::default();
-                            let mut indexer = indexer::DocumentOperation::new(
-                                IndexDocumentsMethod::ReplaceDocuments,
-                            );
+                            let embedders = RuntimeEmbedders::default();
+                            let mut indexer = indexer::DocumentOperation::new();
 
                             let mut operations = Vec::new();
                             for op in batch.0 {
@@ -115,7 +114,7 @@ fn main() {
                             for op in &operations {
                                 match op {
                                     Either::Left(documents) => {
-                                        indexer.add_documents(documents).unwrap()
+                                        indexer.replace_documents(documents).unwrap()
                                     }
                                     Either::Right(ids) => indexer.delete_documents(ids),
                                 }
@@ -145,6 +144,7 @@ fn main() {
                                 embedders,
                                 &|| false,
                                 &Progress::default(),
+                                &Default::default(),
                             )
                             .unwrap();
 

crates/index-scheduler/Cargo.toml

@@ -11,29 +11,31 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-anyhow = "1.0.95"
+anyhow = "1.0.98"
 bincode = "1.3.3"
-bumpalo = "3.16.0"
+byte-unit = "5.1.6"
+bumpalo = "3.18.1"
 bumparaw-collections = "0.1.4"
-convert_case = "0.6.0"
+convert_case = "0.8.0"
 csv = "1.3.1"
 derive_builder = "0.20.2"
 dump = { path = "../dump" }
 enum-iterator = "2.1.0"
 file-store = { path = "../file-store" }
-flate2 = "1.0.35"
+flate2 = "1.1.2"
+indexmap = "2.9.0"
 meilisearch-auth = { path = "../meilisearch-auth" }
 meilisearch-types = { path = "../meilisearch-types" }
 memmap2 = "0.9.5"
 page_size = "0.6.0"
 rayon = "1.10.0"
-roaring = { version = "0.10.10", features = ["serde"] }
-serde = { version = "1.0.217", features = ["derive"] }
-serde_json = { version = "1.0.135", features = ["preserve_order"] }
+roaring = { version = "0.10.12", features = ["serde"] }
+serde = { version = "1.0.219", features = ["derive"] }
+serde_json = { version = "1.0.140", features = ["preserve_order"] }
 synchronoise = "1.0.1"
-tempfile = "3.15.0"
-thiserror = "2.0.9"
-time = { version = "0.3.37", features = [
+tempfile = "3.20.0"
+thiserror = "2.0.12"
+time = { version = "0.3.41", features = [
     "serde-well-known",
     "formatting",
     "parsing",
@@ -41,12 +43,12 @@ time = { version = "0.3.37", features = [
 ] }
 tracing = "0.1.41"
 ureq = "2.12.1"
-uuid = { version = "1.11.0", features = ["serde", "v4"] }
+uuid = { version = "1.17.0", features = ["serde", "v4"] }
+backoff = "0.4.0"
 
 [dev-dependencies]
-arroy = "0.5.0"
 big_s = "1.0.2"
-crossbeam-channel = "0.5.14"
+crossbeam-channel = "0.5.15"
 # fixed version due to format breakages in v1.40
 insta = { version = "=1.39.0", features = ["json", "redactions"] }
 maplit = "1.0.2"

crates/index-scheduler/src/dump.rs

@@ -4,6 +4,7 @@ use std::io;
 use dump::{KindDump, TaskDump, UpdateFile};
 use meilisearch_types::batches::{Batch, BatchId};
 use meilisearch_types::heed::RwTxn;
+use meilisearch_types::index_uid_pattern::IndexUidPattern;
 use meilisearch_types::milli;
 use meilisearch_types::tasks::{Kind, KindWithContent, Status, Task};
 use roaring::RoaringBitmap;
@@ -211,6 +212,23 @@ impl<'a> Dump<'a> {
                     KindWithContent::DumpCreation { keys, instance_uid }
                 }
                 KindDump::SnapshotCreation => KindWithContent::SnapshotCreation,
+                KindDump::Export { url, api_key, payload_size, indexes } => {
+                    KindWithContent::Export {
+                        url,
+                        api_key,
+                        payload_size,
+                        indexes: indexes
+                            .into_iter()
+                            .map(|(pattern, settings)| {
+                                Ok((
+                                    IndexUidPattern::try_from(pattern)
+                                        .map_err(|_| Error::CorruptedDump)?,
+                                    settings,
+                                ))
+                            })
+                            .collect::<Result<_, Error>>()?,
+                    }
+                }
                 KindDump::UpgradeDatabase { from } => KindWithContent::UpgradeDatabase { from },
             },
         };

crates/index-scheduler/src/error.rs

@@ -2,6 +2,7 @@ use std::fmt::Display;
 
 use meilisearch_types::batches::BatchId;
 use meilisearch_types::error::{Code, ErrorCode};
+use meilisearch_types::milli::index::RollbackOutcome;
 use meilisearch_types::tasks::{Kind, Status};
 use meilisearch_types::{heed, milli};
 use thiserror::Error;
@@ -151,7 +152,27 @@ pub enum Error {
     #[error(transparent)]
     DatabaseUpgrade(Box<Self>),
     #[error(transparent)]
+    Export(Box<Self>),
+    #[error("Failed to export documents to remote server {code} ({type}): {message} <{link}>")]
+    FromRemoteWhenExporting { message: String, code: String, r#type: String, link: String },
+    #[error("Failed to rollback for index `{index}`: {rollback_outcome} ")]
+    RollbackFailed { index: String, rollback_outcome: RollbackOutcome },
+    #[error(transparent)]
     UnrecoverableError(Box<Self>),
+    #[error("The index scheduler is in version v{}.{}.{}, but Meilisearch is in version v{}.{}.{}.\n  - hint: start the correct version of Meilisearch, or consider updating your database. See also <https://www.meilisearch.com/docs/learn/update_and_migration/updating>",
+    index_scheduler_version.0, index_scheduler_version.1, index_scheduler_version.2,
+    package_version.0, package_version.1, package_version.2)]
+    IndexSchedulerVersionMismatch {
+        index_scheduler_version: (u32, u32, u32),
+        package_version: (u32, u32, u32),
+    },
+    #[error("Index `{index}` is in version v{}.{}.{}, but Meilisearch is in version v{}.{}.{}.\n  - note: this is an internal error, please consider filing a bug report: <https://github.com/meilisearch/meilisearch/issues/new?template=bug_report.md>",
+    index_version.0, index_version.1, index_version.2, package_version.0, package_version.1, package_version.2)]
+    IndexVersionMismatch {
+        index: String,
+        index_version: (u32, u32, u32),
+        package_version: (u32, u32, u32),
+    },
     #[error(transparent)]
     HeedTransaction(heed::Error),
 
@@ -195,6 +216,7 @@ impl Error {
             | Error::BatchNotFound(_)
             | Error::TaskDeletionWithEmptyQuery
             | Error::TaskCancelationWithEmptyQuery
+            | Error::FromRemoteWhenExporting { .. }
             | Error::AbortedTask
             | Error::Dump(_)
             | Error::Heed(_)
@@ -204,11 +226,15 @@ impl Error {
             | Error::IoError(_)
             | Error::Persist(_)
             | Error::FeatureNotEnabled(_)
+            | Error::Export(_)
             | Error::Anyhow(_) => true,
             Error::CreateBatch(_)
             | Error::CorruptedTaskQueue
             | Error::DatabaseUpgrade(_)
             | Error::UnrecoverableError(_)
+            | Error::IndexSchedulerVersionMismatch { .. }
+            | Error::IndexVersionMismatch { .. }
+            | Error::RollbackFailed { .. }
             | Error::HeedTransaction(_) => false,
             #[cfg(test)]
             Error::PlannedFailure => false,
@@ -262,6 +288,7 @@ impl ErrorCode for Error {
             Error::Dump(e) => e.error_code(),
             Error::Milli { error, .. } => error.error_code(),
             Error::ProcessBatchPanicked(_) => Code::Internal,
+            Error::FromRemoteWhenExporting { .. } => Code::Internal,
             Error::Heed(e) => e.error_code(),
             Error::HeedTransaction(e) => e.error_code(),
             Error::FileStore(e) => e.error_code(),
@@ -274,7 +301,11 @@ impl ErrorCode for Error {
             Error::CorruptedTaskQueue => Code::Internal,
             Error::CorruptedDump => Code::Internal,
             Error::DatabaseUpgrade(_) => Code::Internal,
+            Error::Export(_) => Code::Internal,
+            Error::RollbackFailed { .. } => Code::Internal,
             Error::UnrecoverableError(_) => Code::Internal,
+            Error::IndexSchedulerVersionMismatch { .. } => Code::Internal,
+            Error::IndexVersionMismatch { .. } => Code::Internal,
             Error::CreateBatch(_) => Code::Internal,
 
             // This one should never be seen by the end user

crates/index-scheduler/src/features.rs

@@ -2,7 +2,7 @@ use std::sync::{Arc, RwLock};
 
 use meilisearch_types::features::{InstanceTogglableFeatures, Network, RuntimeTogglableFeatures};
 use meilisearch_types::heed::types::{SerdeJson, Str};
-use meilisearch_types::heed::{Database, Env, RwTxn};
+use meilisearch_types::heed::{Database, Env, RwTxn, WithoutTls};
 
 use crate::error::FeatureNotEnabledError;
 use crate::Result;
@@ -118,6 +118,45 @@ impl RoFeatures {
             .into())
         }
     }
+
+    pub fn check_composite_embedders(&self, disabled_action: &'static str) -> Result<()> {
+        if self.runtime.composite_embedders {
+            Ok(())
+        } else {
+            Err(FeatureNotEnabledError {
+                disabled_action,
+                feature: "composite embedders",
+                issue_link: "https://github.com/orgs/meilisearch/discussions/816",
+            }
+            .into())
+        }
+    }
+
+    pub fn check_chat_completions(&self, disabled_action: &'static str) -> Result<()> {
+        if self.runtime.chat_completions {
+            Ok(())
+        } else {
+            Err(FeatureNotEnabledError {
+                disabled_action,
+                feature: "chat completions",
+                issue_link: "https://github.com/orgs/meilisearch/discussions/835",
+            }
+            .into())
+        }
+    }
+
+    pub fn check_multimodal(&self, disabled_action: &'static str) -> Result<()> {
+        if self.runtime.multimodal {
+            Ok(())
+        } else {
+            Err(FeatureNotEnabledError {
+                disabled_action,
+                feature: "multimodal",
+                issue_link: "https://github.com/orgs/meilisearch/discussions/846",
+            }
+            .into())
+        }
+    }
 }
 
 impl FeatureData {
@@ -126,7 +165,7 @@ impl FeatureData {
     }
 
     pub fn new(
-        env: &Env,
+        env: &Env<WithoutTls>,
         wtxn: &mut RwTxn,
         instance_features: InstanceTogglableFeatures,
     ) -> Result<Self> {

crates/index-scheduler/src/index_mapper/index_map.rs

@@ -304,7 +304,8 @@ fn create_or_open_index(
     map_size: usize,
     creation: bool,
 ) -> Result<Index> {
-    let mut options = EnvOpenOptions::new();
+    let options = EnvOpenOptions::new();
+    let mut options = options.read_txn_without_tls();
     options.map_size(clamp_to_page_size(map_size));
 
     // You can find more details about this experimental
@@ -333,7 +334,7 @@ fn create_or_open_index(
 #[cfg(test)]
 mod tests {
 
-    use meilisearch_types::heed::Env;
+    use meilisearch_types::heed::{Env, WithoutTls};
     use meilisearch_types::Index;
     use uuid::Uuid;
 
@@ -343,7 +344,7 @@ mod tests {
     use crate::IndexScheduler;
 
     impl IndexMapper {
-        fn test() -> (Self, Env, IndexSchedulerHandle) {
+        fn test() -> (Self, Env<WithoutTls>, IndexSchedulerHandle) {
             let (index_scheduler, handle) = IndexScheduler::test(true, vec![]);
             (index_scheduler.index_mapper, index_scheduler.env, handle)
         }

crates/index-scheduler/src/index_mapper/mod.rs

@@ -4,8 +4,10 @@ use std::time::Duration;
 use std::{fs, thread};
 
 use meilisearch_types::heed::types::{SerdeJson, Str};
-use meilisearch_types::heed::{Database, Env, RoTxn, RwTxn};
+use meilisearch_types::heed::{Database, Env, RoTxn, RwTxn, WithoutTls};
 use meilisearch_types::milli;
+use meilisearch_types::milli::database_stats::DatabaseStats;
+use meilisearch_types::milli::index::RollbackOutcome;
 use meilisearch_types::milli::update::IndexerConfig;
 use meilisearch_types::milli::{FieldDistribution, Index};
 use serde::{Deserialize, Serialize};
@@ -98,8 +100,13 @@ pub enum IndexStatus {
 /// The statistics that can be computed from an `Index` object.
 #[derive(Serialize, Deserialize, Debug)]
 pub struct IndexStats {
-    /// Number of documents in the index.
-    pub number_of_documents: u64,
+    /// Stats of the documents database.
+    #[serde(default)]
+    pub documents_database_stats: DatabaseStats,
+
+    #[serde(default, skip_serializing)]
+    pub number_of_documents: Option<u64>,
+
     /// Size taken up by the index' DB, in bytes.
     ///
     /// This includes the size taken by both the used and free pages of the DB, and as the free pages
@@ -138,9 +145,10 @@ impl IndexStats {
     pub fn new(index: &Index, rtxn: &RoTxn) -> milli::Result<Self> {
         let arroy_stats = index.arroy_stats(rtxn)?;
         Ok(IndexStats {
-            number_of_documents: index.number_of_documents(rtxn)?,
             number_of_embeddings: Some(arroy_stats.number_of_embeddings),
             number_of_embedded_documents: Some(arroy_stats.documents.len()),
+            documents_database_stats: index.documents_stats(rtxn)?.unwrap_or_default(),
+            number_of_documents: None,
             database_size: index.on_disk_size()?,
             used_database_size: index.used_size()?,
             primary_key: index.primary_key(rtxn)?.map(|s| s.to_string()),
@@ -157,7 +165,7 @@ impl IndexMapper {
     }
 
     pub fn new(
-        env: &Env,
+        env: &Env<WithoutTls>,
         wtxn: &mut RwTxn,
         options: &IndexSchedulerOptions,
         budget: IndexBudget,
@@ -424,6 +432,51 @@ impl IndexMapper {
         Ok(index)
     }
 
+    pub fn rollback_index(
+        &self,
+        rtxn: &RoTxn,
+        name: &str,
+        to: (u32, u32, u32),
+    ) -> Result<RollbackOutcome> {
+        // remove any currently updating index to make sure that we aren't keeping a reference to the index somewhere
+        drop(self.currently_updating_index.write().unwrap().take());
+
+        let uuid = self
+            .index_mapping
+            .get(rtxn, name)?
+            .ok_or_else(|| Error::IndexNotFound(name.to_string()))?;
+
+        // take the lock to make sure noone is messing with the indexes while we rollback
+        // this will block any search or other operation, but we are rollbacking so this is probably acceptable.
+        let mut index_map = self.index_map.write().unwrap();
+
+        'close_index: loop {
+            match index_map.get(&uuid) {
+                Available(_) => {
+                    index_map.close_for_resize(&uuid, self.enable_mdb_writemap, 0);
+                    // index should now be `Closing`; try again
+                    continue;
+                }
+                // index already closed
+                Missing => break 'close_index,
+                // closing requested by this thread or another one; wait for closing to complete, then exit
+                Closing(closing_index) => {
+                    if closing_index.wait_timeout(Duration::from_secs(100)).is_none() {
+                        // release the lock so it doesn't get poisoned
+                        drop(index_map);
+                        panic!("cannot close index")
+                    }
+                    break;
+                }
+                BeingDeleted => return Err(Error::IndexNotFound(name.to_string())),
+            };
+        }
+
+        let index_path = self.base_path.join(uuid.to_string());
+        Index::rollback(milli::heed::EnvOpenOptions::new().read_txn_without_tls(), index_path, to)
+            .map_err(|err| crate::Error::from_milli(err, Some(name.to_string())))
+    }
+
     /// Attempts `f` for each index that exists in the index mapper.
     ///
     /// It is preferable to use this function rather than a loop that opens all indexes, as a way to avoid having all indexes opened,

crates/index-scheduler/src/insta_snapshot.rs

@@ -1,7 +1,7 @@
 use std::collections::BTreeSet;
 use std::fmt::Write;
 
-use meilisearch_types::batches::{Batch, BatchEnqueuedAt};
+use meilisearch_types::batches::{Batch, BatchEnqueuedAt, BatchStats};
 use meilisearch_types::heed::types::{SerdeBincode, SerdeJson, Str};
 use meilisearch_types::heed::{Database, RoTxn};
 use meilisearch_types::milli::{CboRoaringBitmapCodec, RoaringBitmapCodec, BEU32};
@@ -34,6 +34,7 @@ pub fn snapshot_index_scheduler(scheduler: &IndexScheduler) -> String {
         planned_failures: _,
         run_loop_iteration: _,
         embedders: _,
+        chat_settings: _,
     } = scheduler;
 
     let rtxn = env.read_txn().unwrap();
@@ -41,11 +42,8 @@ pub fn snapshot_index_scheduler(scheduler: &IndexScheduler) -> String {
     let mut snap = String::new();
 
     let indx_sched_version = version.get_version(&rtxn).unwrap();
-    let latest_version = (
-        versioning::VERSION_MAJOR.parse().unwrap(),
-        versioning::VERSION_MINOR.parse().unwrap(),
-        versioning::VERSION_PATCH.parse().unwrap(),
-    );
+    let latest_version =
+        (versioning::VERSION_MAJOR, versioning::VERSION_MINOR, versioning::VERSION_PATCH);
     if indx_sched_version != Some(latest_version) {
         snap.push_str(&format!("index scheduler running on version {indx_sched_version:?}\n"));
     }
@@ -291,6 +289,9 @@ fn snapshot_details(d: &Details) -> String {
         Details::IndexSwap { swaps } => {
             format!("{{ swaps: {swaps:?} }}")
         }
+        Details::Export { url, api_key, payload_size, indexes } => {
+            format!("{{ url: {url:?}, api_key: {api_key:?}, payload_size: {payload_size:?}, indexes: {indexes:?} }}")
+        }
         Details::UpgradeDatabase { from, to } => {
             format!("{{ from: {from:?}, to: {to:?} }}")
         }
@@ -341,7 +342,23 @@ pub fn snapshot_canceled_by(rtxn: &RoTxn, db: Database<BEU32, RoaringBitmapCodec
 
 pub fn snapshot_batch(batch: &Batch) -> String {
     let mut snap = String::new();
-    let Batch { uid, details, stats, started_at, finished_at, progress: _, enqueued_at } = batch;
+    let Batch {
+        uid,
+        details,
+        stats,
+        embedder_stats,
+        started_at,
+        finished_at,
+        progress: _,
+        enqueued_at,
+        stop_reason,
+    } = batch;
+    let stats = BatchStats {
+        progress_trace: Default::default(),
+        internal_database_sizes: Default::default(),
+        write_channel_congestion: None,
+        ..stats.clone()
+    };
     if let Some(finished_at) = finished_at {
         assert!(finished_at > started_at);
     }
@@ -352,7 +369,14 @@ pub fn snapshot_batch(batch: &Batch) -> String {
     snap.push('{');
     snap.push_str(&format!("uid: {uid}, "));
     snap.push_str(&format!("details: {}, ", serde_json::to_string(details).unwrap()));
-    snap.push_str(&format!("stats: {}, ", serde_json::to_string(stats).unwrap()));
+    snap.push_str(&format!("stats: {}, ", serde_json::to_string(&stats).unwrap()));
+    if !embedder_stats.skip_serializing() {
+        snap.push_str(&format!(
+            "embedder stats: {}, ",
+            serde_json::to_string(&embedder_stats).unwrap()
+        ));
+    }
+    snap.push_str(&format!("stop reason: {}, ", serde_json::to_string(&stop_reason).unwrap()));
     snap.push('}');
     snap
 }
@@ -365,7 +389,8 @@ pub fn snapshot_index_mapper(rtxn: &RoTxn, mapper: &IndexMapper) -> String {
         let stats = mapper.stats_of(rtxn, &name).unwrap();
         s.push_str(&format!(
             "{name}: {{ number_of_documents: {}, field_distribution: {:?} }}\n",
-            stats.number_of_documents, stats.field_distribution
+            stats.documents_database_stats.number_of_entries(),
+            stats.field_distribution
         ));
     }
 

crates/index-scheduler/src/lib.rs

@@ -51,16 +51,21 @@ pub use features::RoFeatures;
 use flate2::bufread::GzEncoder;
 use flate2::Compression;
 use meilisearch_types::batches::Batch;
-use meilisearch_types::features::{InstanceTogglableFeatures, Network, RuntimeTogglableFeatures};
+use meilisearch_types::features::{
+    ChatCompletionSettings, InstanceTogglableFeatures, Network, RuntimeTogglableFeatures,
+};
 use meilisearch_types::heed::byteorder::BE;
-use meilisearch_types::heed::types::I128;
-use meilisearch_types::heed::{self, Env, RoTxn};
-use meilisearch_types::milli::index::IndexEmbeddingConfig;
+use meilisearch_types::heed::types::{DecodeIgnore, SerdeJson, Str, I128};
+use meilisearch_types::heed::{self, Database, Env, RoTxn, WithoutTls};
 use meilisearch_types::milli::update::IndexerConfig;
-use meilisearch_types::milli::vector::{Embedder, EmbedderOptions, EmbeddingConfigs};
+use meilisearch_types::milli::vector::json_template::JsonTemplate;
+use meilisearch_types::milli::vector::{
+    Embedder, EmbedderOptions, RuntimeEmbedder, RuntimeEmbedders, RuntimeFragment,
+};
 use meilisearch_types::milli::{self, Index};
 use meilisearch_types::task_view::TaskView;
 use meilisearch_types::tasks::{KindWithContent, Task};
+use milli::vector::db::IndexEmbeddingConfig;
 use processing::ProcessingTasks;
 pub use queue::Query;
 use queue::Queue;
@@ -74,6 +79,9 @@ use crate::utils::clamp_to_page_size;
 
 pub(crate) type BEI128 = I128<BE>;
 
+const TASK_SCHEDULER_SIZE_THRESHOLD_PERCENT_INT: u64 = 40;
+const CHAT_SETTINGS_DB_NAME: &str = "chat-settings";
+
 #[derive(Debug)]
 pub struct IndexSchedulerOptions {
     /// The path to the version file of Meilisearch.
@@ -125,13 +133,19 @@ pub struct IndexSchedulerOptions {
     pub instance_features: InstanceTogglableFeatures,
     /// The experimental features enabled for this instance.
     pub auto_upgrade: bool,
+    /// The maximal number of entries in the search query cache of an embedder.
+    ///
+    /// 0 disables the cache.
+    pub embedding_cache_cap: usize,
+    /// Snapshot compaction status.
+    pub experimental_no_snapshot_compaction: bool,
 }
 
 /// Structure which holds meilisearch's indexes and schedules the tasks
 /// to be performed on them.
 pub struct IndexScheduler {
     /// The LMDB environment which the DBs are associated with.
-    pub(crate) env: Env,
+    pub(crate) env: Env<WithoutTls>,
 
     /// The list of tasks currently processing
     pub(crate) processing_tasks: Arc<RwLock<ProcessingTasks>>,
@@ -145,6 +159,9 @@ pub struct IndexScheduler {
     /// In charge of fetching and setting the status of experimental features.
     features: features::FeatureData,
 
+    /// Stores the custom chat prompts and other settings of the indexes.
+    pub(crate) chat_settings: Database<Str, SerdeJson<ChatCompletionSettings>>,
+
     /// Everything related to the processing of the tasks
     pub scheduler: scheduler::Scheduler,
 
@@ -156,6 +173,11 @@ pub struct IndexScheduler {
     /// The Authorization header to send to the webhook URL.
     pub(crate) webhook_authorization_header: Option<String>,
 
+    /// A map to retrieve the runtime representation of an embedder depending on its configuration.
+    ///
+    /// This map may return the same embedder object for two different indexes or embedder settings,
+    /// but it will only do this if the embedder configuration options are the same, leading
+    /// to the same embeddings for the same input text.
     embedders: Arc<RwLock<HashMap<EmbedderOptions, Arc<Embedder>>>>,
 
     // ================= test
@@ -198,17 +220,23 @@ impl IndexScheduler {
             #[cfg(test)]
             run_loop_iteration: self.run_loop_iteration.clone(),
             features: self.features.clone(),
+            chat_settings: self.chat_settings,
         }
     }
 
     pub(crate) const fn nb_db() -> u32 {
-        Versioning::nb_db() + Queue::nb_db() + IndexMapper::nb_db() + features::FeatureData::nb_db()
+        Versioning::nb_db()
+            + Queue::nb_db()
+            + IndexMapper::nb_db()
+            + features::FeatureData::nb_db()
+            + 1 // chat-prompts
     }
 
     /// Create an index scheduler and start its run loop.
     #[allow(private_interfaces)] // because test_utils is private
     pub fn new(
         options: IndexSchedulerOptions,
+        auth_env: Env<WithoutTls>,
         from_db_version: (u32, u32, u32),
         #[cfg(test)] test_breakpoint_sdr: crossbeam_channel::Sender<(test_utils::Breakpoint, bool)>,
         #[cfg(test)] planned_failures: Vec<(usize, test_utils::FailureLocation)>,
@@ -240,7 +268,9 @@ impl IndexScheduler {
         };
 
         let env = unsafe {
-            heed::EnvOpenOptions::new()
+            let env_options = heed::EnvOpenOptions::new();
+            let mut env_options = env_options.read_txn_without_tls();
+            env_options
                 .max_dbs(Self::nb_db())
                 .map_size(budget.task_db_size)
                 .open(&options.tasks_path)
@@ -253,6 +283,7 @@ impl IndexScheduler {
         let features = features::FeatureData::new(&env, &mut wtxn, options.instance_features)?;
         let queue = Queue::new(&env, &mut wtxn, &options)?;
         let index_mapper = IndexMapper::new(&env, &mut wtxn, &options, budget)?;
+        let chat_settings = env.create_database(&mut wtxn, Some(CHAT_SETTINGS_DB_NAME))?;
         wtxn.commit()?;
 
         // allow unreachable_code to get rids of the warning in the case of a test build.
@@ -260,7 +291,7 @@ impl IndexScheduler {
             processing_tasks: Arc::new(RwLock::new(ProcessingTasks::new())),
             version,
             queue,
-            scheduler: Scheduler::new(&options),
+            scheduler: Scheduler::new(&options, auth_env),
 
             index_mapper,
             env,
@@ -276,12 +307,17 @@ impl IndexScheduler {
             #[cfg(test)]
             run_loop_iteration: Arc::new(RwLock::new(0)),
             features,
+            chat_settings,
         };
 
         this.run();
         Ok(this)
     }
 
+    fn read_txn(&self) -> Result<RoTxn<WithoutTls>> {
+        self.env.read_txn().map_err(|e| e.into())
+    }
+
     /// Return `Ok(())` if the index scheduler is able to access one of its database.
     pub fn health(&self) -> Result<()> {
         let rtxn = self.env.read_txn()?;
@@ -358,15 +394,16 @@ impl IndexScheduler {
         }
     }
 
-    pub fn read_txn(&self) -> Result<RoTxn> {
-        self.env.read_txn().map_err(|e| e.into())
-    }
-
     /// Start the run loop for the given index scheduler.
     ///
     /// This function will execute in a different thread and must be called
     /// only once per index scheduler.
     fn run(&self) {
+        // If the number of batched tasks is 0, we don't need to run the scheduler at all.
+        // It will never be able to process any tasks.
+        if self.scheduler.max_number_of_batched_tasks == 0 {
+            return;
+        }
         let run = self.private_clone();
         std::thread::Builder::new()
             .name(String::from("scheduler"))
@@ -384,9 +421,9 @@ impl IndexScheduler {
                         Ok(Ok(TickOutcome::StopProcessingForever)) => break,
                         Ok(Err(e)) => {
                             tracing::error!("{e}");
-                            // Wait one second when an irrecoverable error occurs.
+                            // Wait when an irrecoverable error occurs.
                             if !e.is_recoverable() {
-                                std::thread::sleep(Duration::from_secs(1));
+                                std::thread::sleep(Duration::from_secs(10));
                             }
                         }
                         Err(_panic) => {
@@ -413,6 +450,17 @@ impl IndexScheduler {
         Ok(self.env.non_free_pages_size()?)
     }
 
+    /// Return the maximum possible database size
+    pub fn max_size(&self) -> Result<u64> {
+        Ok(self.env.info().map_size as u64)
+    }
+
+    /// Return the max size of task allowed until the task queue stop receiving.
+    pub fn remaining_size_until_task_queue_stop(&self) -> Result<u64> {
+        Ok((self.env.info().map_size as u64 * TASK_SCHEDULER_SIZE_THRESHOLD_PERCENT_INT / 100)
+            .saturating_sub(self.used_size()?))
+    }
+
     /// Return the index corresponding to the name.
     ///
     /// * If the index wasn't opened before, the index will be opened.
@@ -427,12 +475,14 @@ impl IndexScheduler {
     /// If you need to fetch information from or perform an action on all indexes,
     /// see the `try_for_each_index` function.
     pub fn index(&self, name: &str) -> Result<Index> {
-        self.index_mapper.index(&self.env.read_txn()?, name)
+        let rtxn = self.env.read_txn()?;
+        self.index_mapper.index(&rtxn, name)
     }
 
     /// Return the boolean referring if index exists.
     pub fn index_exists(&self, name: &str) -> Result<bool> {
-        self.index_mapper.index_exists(&self.env.read_txn()?, name)
+        let rtxn = self.env.read_txn()?;
+        self.index_mapper.index_exists(&rtxn, name)
     }
 
     /// Return the name of all indexes without opening them.
@@ -461,7 +511,7 @@ impl IndexScheduler {
 
     /// Returns the total number of indexes available for the specified filter.
     /// And a `Vec` of the index_uid + its stats
-    pub fn get_paginated_indexes_stats(
+    pub fn paginated_indexes_stats(
         &self,
         filters: &meilisearch_auth::AuthFilter,
         from: usize,
@@ -502,12 +552,31 @@ impl IndexScheduler {
         ret.map(|ret| (total, ret))
     }
 
+    /// Returns the total number of chat workspaces available ~~for the specified filter~~.
+    /// And a `Vec` of the workspace_uids
+    pub fn paginated_chat_workspace_uids(
+        &self,
+        from: usize,
+        limit: usize,
+    ) -> Result<(usize, Vec<String>)> {
+        let rtxn = self.read_txn()?;
+        let total = self.chat_settings.len(&rtxn)?;
+        let mut iter = self.chat_settings.iter(&rtxn)?.skip(from);
+        iter.by_ref()
+            .take(limit)
+            .map(|ret| ret.map_err(Error::from))
+            .map(|ret| ret.map(|(uid, _)| uid.to_string()))
+            .collect::<Result<Vec<_>, Error>>()
+            .map(|ret| (total as usize, ret))
+    }
+
     /// The returned structure contains:
     /// 1. The name of the property being observed can be `statuses`, `types`, or `indexes`.
     /// 2. The name of the specific data related to the property can be `enqueued` for the `statuses`, `settingsUpdate` for the `types`, or the name of the index for the `indexes`, for example.
     /// 3. The number of times the properties appeared.
     pub fn get_stats(&self) -> Result<BTreeMap<String, BTreeMap<String, u64>>> {
-        self.queue.get_stats(&self.read_txn()?, &self.processing_tasks.read().unwrap())
+        let rtxn = self.read_txn()?;
+        self.queue.get_stats(&rtxn, &self.processing_tasks.read().unwrap())
     }
 
     // Return true if there is at least one task that is processing.
@@ -610,9 +679,10 @@ impl IndexScheduler {
         task_id: Option<TaskId>,
         dry_run: bool,
     ) -> Result<Task> {
-        // if the task doesn't delete anything and 50% of the task queue is full, we must refuse to enqueue the incomming task
-        if !matches!(&kind, KindWithContent::TaskDeletion { tasks, .. } if !tasks.is_empty())
-            && (self.env.non_free_pages_size()? * 100) / self.env.info().map_size as u64 > 40
+        // if the task doesn't delete or cancel anything and 40% of the task queue is full, we must refuse to enqueue the incoming task
+        if !matches!(&kind, KindWithContent::TaskDeletion { tasks, .. } | KindWithContent::TaskCancelation { tasks, .. } if !tasks.is_empty())
+            && (self.env.non_free_pages_size()? * 100) / self.env.info().map_size as u64
+                > TASK_SCHEDULER_SIZE_THRESHOLD_PERCENT_INT
         {
             return Err(Error::NoSpaceLeftInTaskQueue);
         }
@@ -681,7 +751,7 @@ impl IndexScheduler {
                 written: usize,
             }
 
-            impl<'a, 'b> Read for TaskReader<'a, 'b> {
+            impl Read for TaskReader<'_, '_> {
                 fn read(&mut self, mut buf: &mut [u8]) -> std::io::Result<usize> {
                     if self.buffer.is_empty() {
                         match self.tasks.next() {
@@ -784,35 +854,48 @@ impl IndexScheduler {
         &self,
         index_uid: String,
         embedding_configs: Vec<IndexEmbeddingConfig>,
-    ) -> Result<EmbeddingConfigs> {
+    ) -> Result<RuntimeEmbedders> {
         let res: Result<_> = embedding_configs
             .into_iter()
             .map(
                 |IndexEmbeddingConfig {
                      name,
                      config: milli::vector::EmbeddingConfig { embedder_options, prompt, quantized },
-                     ..
-                 }| {
-                    let prompt = Arc::new(
-                        prompt
+                     fragments,
+                 }|
+                 -> Result<(String, Arc<RuntimeEmbedder>)> {
+                    let document_template = prompt
                         .try_into()
                         .map_err(meilisearch_types::milli::Error::from)
-                            .map_err(|err| Error::from_milli(err, Some(index_uid.clone())))?,
-                    );
+                        .map_err(|err| Error::from_milli(err, Some(index_uid.clone())))?;
+
+                    let fragments = fragments
+                        .into_inner()
+                        .into_iter()
+                        .map(|fragment| {
+                            let value = embedder_options.fragment(&fragment.name).unwrap();
+                            let template = JsonTemplate::new(value.clone()).unwrap();
+                            RuntimeFragment { name: fragment.name, id: fragment.id, template }
+                        })
+                        .collect();
                     // optimistically return existing embedder
                     {
                         let embedders = self.embedders.read().unwrap();
                         if let Some(embedder) = embedders.get(&embedder_options) {
-                            return Ok((
-                                name,
-                                (embedder.clone(), prompt, quantized.unwrap_or_default()),
+                            let runtime = Arc::new(RuntimeEmbedder::new(
+                                embedder.clone(),
+                                document_template,
+                                fragments,
+                                quantized.unwrap_or_default(),
                             ));
+
+                            return Ok((name, runtime));
                         }
                     }
 
                     // add missing embedder
                     let embedder = Arc::new(
-                        Embedder::new(embedder_options.clone())
+                        Embedder::new(embedder_options.clone(), self.scheduler.embedding_cache_cap)
                             .map_err(meilisearch_types::milli::vector::Error::from)
                             .map_err(|err| {
                                 Error::from_milli(err.into(), Some(index_uid.clone()))
@@ -822,11 +905,44 @@ impl IndexScheduler {
                         let mut embedders = self.embedders.write().unwrap();
                         embedders.insert(embedder_options, embedder.clone());
                     }
-                    Ok((name, (embedder, prompt, quantized.unwrap_or_default())))
+
+                    let runtime = Arc::new(RuntimeEmbedder::new(
+                        embedder.clone(),
+                        document_template,
+                        fragments,
+                        quantized.unwrap_or_default(),
+                    ));
+
+                    Ok((name, runtime))
                 },
             )
             .collect();
-        res.map(EmbeddingConfigs::new)
+        res.map(RuntimeEmbedders::new)
+    }
+
+    pub fn chat_settings(&self, uid: &str) -> Result<Option<ChatCompletionSettings>> {
+        let rtxn = self.env.read_txn()?;
+        self.chat_settings.get(&rtxn, uid).map_err(Into::into)
+    }
+
+    /// Return true if chat workspace exists.
+    pub fn chat_workspace_exists(&self, name: &str) -> Result<bool> {
+        let rtxn = self.env.read_txn()?;
+        Ok(self.chat_settings.remap_data_type::<DecodeIgnore>().get(&rtxn, name)?.is_some())
+    }
+
+    pub fn put_chat_settings(&self, uid: &str, settings: &ChatCompletionSettings) -> Result<()> {
+        let mut wtxn = self.env.write_txn()?;
+        self.chat_settings.put(&mut wtxn, uid, settings)?;
+        wtxn.commit()?;
+        Ok(())
+    }
+
+    pub fn delete_chat_settings(&self, uid: &str) -> Result<bool> {
+        let mut wtxn = self.env.write_txn()?;
+        let deleted = self.chat_settings.delete(&mut wtxn, uid)?;
+        wtxn.commit()?;
+        Ok(deleted)
     }
 }
 

crates/index-scheduler/src/processing.rs

@@ -64,9 +64,17 @@ make_enum_progress! {
     }
 }
 
+make_enum_progress! {
+    pub enum FinalizingIndexStep {
+        Committing,
+        ComputingStats,
+    }
+}
+
 make_enum_progress! {
     pub enum TaskCancelationProgress {
         RetrievingTasks,
+        CancelingUpgrade,
         UpdatingTasks,
     }
 }
@@ -95,6 +103,7 @@ make_enum_progress! {
     pub enum DumpCreationProgress {
         StartTheDumpCreation,
         DumpTheApiKeys,
+        DumpTheChatCompletionSettings,
         DumpTheTasks,
         DumpTheBatches,
         DumpTheIndexes,
@@ -167,8 +176,17 @@ make_enum_progress! {
     }
 }
 
+make_enum_progress! {
+    pub enum Export {
+        EnsuringCorrectnessOfTheTarget,
+        ExportingTheSettings,
+        ExportingTheDocuments,
+    }
+}
+
 make_atomic_progress!(Task alias AtomicTaskStep => "task" );
 make_atomic_progress!(Document alias AtomicDocumentStep => "document" );
+make_atomic_progress!(Index alias AtomicIndexStep => "index" );
 make_atomic_progress!(Batch alias AtomicBatchStep => "batch" );
 make_atomic_progress!(UpdateFile alias AtomicUpdateFileStep => "update file" );
 

crates/index-scheduler/src/queue/batches.rs

@@ -3,7 +3,7 @@ use std::ops::{Bound, RangeBounds};
 
 use meilisearch_types::batches::{Batch, BatchId};
 use meilisearch_types::heed::types::{DecodeIgnore, SerdeBincode, SerdeJson, Str};
-use meilisearch_types::heed::{Database, Env, RoTxn, RwTxn};
+use meilisearch_types::heed::{Database, Env, RoTxn, RwTxn, WithoutTls};
 use meilisearch_types::milli::{CboRoaringBitmapCodec, RoaringBitmapCodec, BEU32};
 use meilisearch_types::tasks::{Kind, Status};
 use roaring::{MultiOps, RoaringBitmap};
@@ -66,7 +66,7 @@ impl BatchQueue {
         NUMBER_OF_DATABASES
     }
 
-    pub(super) fn new(env: &Env, wtxn: &mut RwTxn) -> Result<Self> {
+    pub(super) fn new(env: &Env<WithoutTls>, wtxn: &mut RwTxn) -> Result<Self> {
         Ok(Self {
             all_batches: env.create_database(wtxn, Some(db_name::ALL_BATCHES))?,
             status: env.create_database(wtxn, Some(db_name::BATCH_STATUS))?,
@@ -179,9 +179,11 @@ impl BatchQueue {
                 progress: None,
                 details: batch.details,
                 stats: batch.stats,
+                embedder_stats: batch.embedder_stats.as_ref().into(),
                 started_at: batch.started_at,
                 finished_at: batch.finished_at,
                 enqueued_at: batch.enqueued_at,
+                stop_reason: batch.reason.to_string(),
             },
         )?;
 

crates/index-scheduler/src/queue/batches_test.rs

@@ -106,7 +106,7 @@ fn query_batches_simple() {
     batches[0].enqueued_at = None;
     // Insta cannot snapshot our batches because the batch stats contains an enum as key: https://github.com/mitsuhiko/insta/issues/689
     let batch = serde_json::to_string_pretty(&batches[0]).unwrap();
-    snapshot!(batch, @r#"
+    snapshot!(batch, @r###"
     {
       "uid": 0,
       "details": {
@@ -126,9 +126,10 @@ fn query_batches_simple() {
       },
       "startedAt": "1970-01-01T00:00:00Z",
       "finishedAt": null,
-      "enqueuedAt": null
+      "enqueuedAt": null,
+      "stopReason": "created batch containing only task with id 0 of type `indexCreation` that cannot be batched with any other task."
     }
-    "#);
+    "###);
 
     let query = Query { statuses: Some(vec![Status::Enqueued]), ..Default::default() };
     let (batches, _) = index_scheduler

crates/index-scheduler/src/queue/mod.rs

@@ -13,7 +13,7 @@ use std::time::Duration;
 
 use file_store::FileStore;
 use meilisearch_types::batches::BatchId;
-use meilisearch_types::heed::{Database, Env, RoTxn, RwTxn};
+use meilisearch_types::heed::{Database, Env, RoTxn, RwTxn, WithoutTls};
 use meilisearch_types::milli::{CboRoaringBitmapCodec, BEU32};
 use meilisearch_types::tasks::{Kind, KindWithContent, Status, Task};
 use roaring::RoaringBitmap;
@@ -157,7 +157,7 @@ impl Queue {
 
     /// Create an index scheduler and start its run loop.
     pub(crate) fn new(
-        env: &Env,
+        env: &Env<WithoutTls>,
         wtxn: &mut RwTxn,
         options: &IndexSchedulerOptions,
     ) -> Result<Self> {
@@ -292,8 +292,6 @@ impl Queue {
             return Ok(task);
         }
 
-        // Get rid of the mutability.
-        let task = task;
         self.tasks.register(wtxn, &task)?;
 
         Ok(task)

crates/index-scheduler/src/queue/snapshots/batches_test.rs/query_batches_canceled_by/start.snap

@@ -1,6 +1,5 @@
 ---
 source: crates/index-scheduler/src/queue/batches_test.rs
-snapshot_kind: text
 ---
 ### Autobatching Enabled = true
 ### Processing batch None:
@@ -49,8 +48,8 @@ catto: { number_of_documents: 0, field_distribution: {} }
 [timestamp] [1,2,3,]
 ----------------------------------------------------------------------
 ### All Batches:
-0 {uid: 0, details: {"primaryKey":"mouse"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"catto":1}}, }
-1 {uid: 1, details: {"primaryKey":"sheep","matchedTasks":3,"canceledTasks":2,"originalFilter":"test_query","swaps":[{"indexes":["catto","doggo"]}]}, stats: {"totalNbTasks":3,"status":{"succeeded":1,"canceled":2},"types":{"indexCreation":1,"indexSwap":1,"taskCancelation":1},"indexUids":{"doggo":1}}, }
+0 {uid: 0, details: {"primaryKey":"mouse"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"catto":1}}, stop reason: "created batch containing only task with id 0 of type `indexCreation` that cannot be batched with any other task.", }
+1 {uid: 1, details: {"primaryKey":"sheep","matchedTasks":3,"canceledTasks":2,"originalFilter":"test_query","swaps":[{"indexes":["catto","doggo"]}]}, stats: {"totalNbTasks":3,"status":{"succeeded":1,"canceled":2},"types":{"indexCreation":1,"indexSwap":1,"taskCancelation":1},"indexUids":{"doggo":1}}, stop reason: "created batch containing only task with id 3 of type `taskCancelation` that cannot be batched with any other task.", }
 ----------------------------------------------------------------------
 ### Batch to tasks mapping:
 0 [0,]

crates/index-scheduler/src/queue/snapshots/batches_test.rs/query_batches_from_and_limit/processed_all_tasks.snap

@@ -1,6 +1,5 @@
 ---
 source: crates/index-scheduler/src/queue/batches_test.rs
-snapshot_kind: text
 ---
 ### Autobatching Enabled = true
 ### Processing batch None:
@@ -48,9 +47,9 @@ whalo: { number_of_documents: 0, field_distribution: {} }
 [timestamp] [2,]
 ----------------------------------------------------------------------
 ### All Batches:
-0 {uid: 0, details: {"primaryKey":"bone"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"doggo":1}}, }
-1 {uid: 1, details: {"primaryKey":"plankton"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"whalo":1}}, }
-2 {uid: 2, details: {"primaryKey":"his_own_vomit"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"catto":1}}, }
+0 {uid: 0, details: {"primaryKey":"bone"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"doggo":1}}, stop reason: "created batch containing only task with id 0 of type `indexCreation` that cannot be batched with any other task.", }
+1 {uid: 1, details: {"primaryKey":"plankton"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"whalo":1}}, stop reason: "created batch containing only task with id 1 of type `indexCreation` that cannot be batched with any other task.", }
+2 {uid: 2, details: {"primaryKey":"his_own_vomit"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"catto":1}}, stop reason: "created batch containing only task with id 2 of type `indexCreation` that cannot be batched with any other task.", }
 ----------------------------------------------------------------------
 ### Batch to tasks mapping:
 0 [0,]

crates/index-scheduler/src/queue/snapshots/batches_test.rs/query_batches_simple/after-advancing-a-bit.snap

@@ -1,11 +1,10 @@
 ---
 source: crates/index-scheduler/src/queue/batches_test.rs
-snapshot_kind: text
 ---
 ### Autobatching Enabled = true
 ### Processing batch Some(1):
 [1,]
-{uid: 1, details: {"primaryKey":"sheep"}, stats: {"totalNbTasks":1,"status":{"processing":1},"types":{"indexCreation":1},"indexUids":{"doggo":1}}, }
+{uid: 1, details: {"primaryKey":"sheep"}, stats: {"totalNbTasks":1,"status":{"processing":1},"types":{"indexCreation":1},"indexUids":{"doggo":1}}, stop reason: "created batch containing only task with id 1 of type `indexCreation` that cannot be batched with any other task.", }
 ----------------------------------------------------------------------
 ### All Tasks:
 0 {uid: 0, batch_uid: 0, status: succeeded, details: { primary_key: Some("mouse") }, kind: IndexCreation { index_uid: "catto", primary_key: Some("mouse") }}
@@ -43,7 +42,7 @@ catto: { number_of_documents: 0, field_distribution: {} }
 [timestamp] [0,]
 ----------------------------------------------------------------------
 ### All Batches:
-0 {uid: 0, details: {"primaryKey":"mouse"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"catto":1}}, }
+0 {uid: 0, details: {"primaryKey":"mouse"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"catto":1}}, stop reason: "created batch containing only task with id 0 of type `indexCreation` that cannot be batched with any other task.", }
 ----------------------------------------------------------------------
 ### Batch to tasks mapping:
 0 [0,]

crates/index-scheduler/src/queue/snapshots/batches_test.rs/query_batches_simple/end.snap

@@ -1,6 +1,5 @@
 ---
 source: crates/index-scheduler/src/queue/batches_test.rs
-snapshot_kind: text
 ---
 ### Autobatching Enabled = true
 ### Processing batch None:
@@ -48,9 +47,9 @@ doggo: { number_of_documents: 0, field_distribution: {} }
 [timestamp] [2,]
 ----------------------------------------------------------------------
 ### All Batches:
-0 {uid: 0, details: {"primaryKey":"mouse"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"catto":1}}, }
-1 {uid: 1, details: {"primaryKey":"sheep"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"doggo":1}}, }
-2 {uid: 2, details: {"primaryKey":"fish"}, stats: {"totalNbTasks":1,"status":{"failed":1},"types":{"indexCreation":1},"indexUids":{"whalo":1}}, }
+0 {uid: 0, details: {"primaryKey":"mouse"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"catto":1}}, stop reason: "created batch containing only task with id 0 of type `indexCreation` that cannot be batched with any other task.", }
+1 {uid: 1, details: {"primaryKey":"sheep"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"doggo":1}}, stop reason: "created batch containing only task with id 1 of type `indexCreation` that cannot be batched with any other task.", }
+2 {uid: 2, details: {"primaryKey":"fish"}, stats: {"totalNbTasks":1,"status":{"failed":1},"types":{"indexCreation":1},"indexUids":{"whalo":1}}, stop reason: "created batch containing only task with id 2 of type `indexCreation` that cannot be batched with any other task.", }
 ----------------------------------------------------------------------
 ### Batch to tasks mapping:
 0 [0,]

crates/index-scheduler/src/queue/snapshots/batches_test.rs/query_batches_special_rules/after-processing-everything.snap

@@ -1,6 +1,5 @@
 ---
 source: crates/index-scheduler/src/queue/batches_test.rs
-snapshot_kind: text
 ---
 ### Autobatching Enabled = true
 ### Processing batch None:
@@ -53,10 +52,10 @@ doggo: { number_of_documents: 0, field_distribution: {} }
 [timestamp] [3,]
 ----------------------------------------------------------------------
 ### All Batches:
-0 {uid: 0, details: {"primaryKey":"mouse"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"catto":1}}, }
-1 {uid: 1, details: {"primaryKey":"sheep"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"doggo":1}}, }
-2 {uid: 2, details: {"swaps":[{"indexes":["catto","doggo"]}]}, stats: {"totalNbTasks":1,"status":{"failed":1},"types":{"indexSwap":1},"indexUids":{}}, }
-3 {uid: 3, details: {"swaps":[{"indexes":["catto","whalo"]}]}, stats: {"totalNbTasks":1,"status":{"failed":1},"types":{"indexSwap":1},"indexUids":{}}, }
+0 {uid: 0, details: {"primaryKey":"mouse"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"catto":1}}, stop reason: "created batch containing only task with id 0 of type `indexCreation` that cannot be batched with any other task.", }
+1 {uid: 1, details: {"primaryKey":"sheep"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"doggo":1}}, stop reason: "created batch containing only task with id 1 of type `indexCreation` that cannot be batched with any other task.", }
+2 {uid: 2, details: {"swaps":[{"indexes":["catto","doggo"]}]}, stats: {"totalNbTasks":1,"status":{"failed":1},"types":{"indexSwap":1},"indexUids":{}}, stop reason: "created batch containing only task with id 2 of type `indexSwap` that cannot be batched with any other task.", }
+3 {uid: 3, details: {"swaps":[{"indexes":["catto","whalo"]}]}, stats: {"totalNbTasks":1,"status":{"failed":1},"types":{"indexSwap":1},"indexUids":{}}, stop reason: "created batch containing only task with id 3 of type `indexSwap` that cannot be batched with any other task.", }
 ----------------------------------------------------------------------
 ### Batch to tasks mapping:
 0 [0,]

crates/index-scheduler/src/queue/snapshots/tasks_test.rs/query_tasks_canceled_by/start.snap

@@ -1,6 +1,5 @@
 ---
 source: crates/index-scheduler/src/queue/tasks_test.rs
-snapshot_kind: text
 ---
 ### Autobatching Enabled = true
 ### Processing batch None:
@@ -49,8 +48,8 @@ catto: { number_of_documents: 0, field_distribution: {} }
 [timestamp] [1,2,3,]
 ----------------------------------------------------------------------
 ### All Batches:
-0 {uid: 0, details: {"primaryKey":"mouse"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"catto":1}}, }
-1 {uid: 1, details: {"primaryKey":"sheep","matchedTasks":3,"canceledTasks":2,"originalFilter":"test_query","swaps":[{"indexes":["catto","doggo"]}]}, stats: {"totalNbTasks":3,"status":{"succeeded":1,"canceled":2},"types":{"indexCreation":1,"indexSwap":1,"taskCancelation":1},"indexUids":{"doggo":1}}, }
+0 {uid: 0, details: {"primaryKey":"mouse"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"catto":1}}, stop reason: "created batch containing only task with id 0 of type `indexCreation` that cannot be batched with any other task.", }
+1 {uid: 1, details: {"primaryKey":"sheep","matchedTasks":3,"canceledTasks":2,"originalFilter":"test_query","swaps":[{"indexes":["catto","doggo"]}]}, stats: {"totalNbTasks":3,"status":{"succeeded":1,"canceled":2},"types":{"indexCreation":1,"indexSwap":1,"taskCancelation":1},"indexUids":{"doggo":1}}, stop reason: "created batch containing only task with id 3 of type `taskCancelation` that cannot be batched with any other task.", }
 ----------------------------------------------------------------------
 ### Batch to tasks mapping:
 0 [0,]

crates/index-scheduler/src/queue/snapshots/tasks_test.rs/query_tasks_from_and_limit/processed_all_tasks.snap

@@ -1,6 +1,5 @@
 ---
 source: crates/index-scheduler/src/queue/tasks_test.rs
-snapshot_kind: text
 ---
 ### Autobatching Enabled = true
 ### Processing batch None:
@@ -48,9 +47,9 @@ whalo: { number_of_documents: 0, field_distribution: {} }
 [timestamp] [2,]
 ----------------------------------------------------------------------
 ### All Batches:
-0 {uid: 0, details: {"primaryKey":"bone"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"doggo":1}}, }
-1 {uid: 1, details: {"primaryKey":"plankton"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"whalo":1}}, }
-2 {uid: 2, details: {"primaryKey":"his_own_vomit"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"catto":1}}, }
+0 {uid: 0, details: {"primaryKey":"bone"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"doggo":1}}, stop reason: "created batch containing only task with id 0 of type `indexCreation` that cannot be batched with any other task.", }
+1 {uid: 1, details: {"primaryKey":"plankton"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"whalo":1}}, stop reason: "created batch containing only task with id 1 of type `indexCreation` that cannot be batched with any other task.", }
+2 {uid: 2, details: {"primaryKey":"his_own_vomit"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"catto":1}}, stop reason: "created batch containing only task with id 2 of type `indexCreation` that cannot be batched with any other task.", }
 ----------------------------------------------------------------------
 ### Batch to tasks mapping:
 0 [0,]

crates/index-scheduler/src/queue/snapshots/tasks_test.rs/query_tasks_simple/end.snap

@@ -1,6 +1,5 @@
 ---
 source: crates/index-scheduler/src/queue/tasks_test.rs
-snapshot_kind: text
 ---
 ### Autobatching Enabled = true
 ### Processing batch None:
@@ -48,9 +47,9 @@ doggo: { number_of_documents: 0, field_distribution: {} }
 [timestamp] [2,]
 ----------------------------------------------------------------------
 ### All Batches:
-0 {uid: 0, details: {"primaryKey":"mouse"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"catto":1}}, }
-1 {uid: 1, details: {"primaryKey":"sheep"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"doggo":1}}, }
-2 {uid: 2, details: {"primaryKey":"fish"}, stats: {"totalNbTasks":1,"status":{"failed":1},"types":{"indexCreation":1},"indexUids":{"whalo":1}}, }
+0 {uid: 0, details: {"primaryKey":"mouse"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"catto":1}}, stop reason: "created batch containing only task with id 0 of type `indexCreation` that cannot be batched with any other task.", }
+1 {uid: 1, details: {"primaryKey":"sheep"}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"indexCreation":1},"indexUids":{"doggo":1}}, stop reason: "created batch containing only task with id 1 of type `indexCreation` that cannot be batched with any other task.", }
+2 {uid: 2, details: {"primaryKey":"fish"}, stats: {"totalNbTasks":1,"status":{"failed":1},"types":{"indexCreation":1},"indexUids":{"whalo":1}}, stop reason: "created batch containing only task with id 2 of type `indexCreation` that cannot be batched with any other task.", }
 ----------------------------------------------------------------------
 ### Batch to tasks mapping:
 0 [0,]

crates/index-scheduler/src/queue/tasks.rs

@@ -1,7 +1,7 @@
 use std::ops::{Bound, RangeBounds};
 
 use meilisearch_types::heed::types::{DecodeIgnore, SerdeBincode, SerdeJson, Str};
-use meilisearch_types::heed::{Database, Env, RoTxn, RwTxn};
+use meilisearch_types::heed::{Database, Env, RoTxn, RwTxn, WithoutTls};
 use meilisearch_types::milli::{CboRoaringBitmapCodec, RoaringBitmapCodec, BEU32};
 use meilisearch_types::tasks::{Kind, Status, Task};
 use roaring::{MultiOps, RoaringBitmap};
@@ -68,7 +68,7 @@ impl TaskQueue {
         NUMBER_OF_DATABASES
     }
 
-    pub(crate) fn new(env: &Env, wtxn: &mut RwTxn) -> Result<Self> {
+    pub(crate) fn new(env: &Env<WithoutTls>, wtxn: &mut RwTxn) -> Result<Self> {
         Ok(Self {
             all_tasks: env.create_database(wtxn, Some(db_name::ALL_TASKS))?,
             status: env.create_database(wtxn, Some(db_name::STATUS))?,
@@ -315,7 +315,7 @@ impl Queue {
         if let Some(batch_uids) = batch_uids {
             let mut batch_tasks = RoaringBitmap::new();
             for batch_uid in batch_uids {
-                if processing_batch.as_ref().map_or(false, |batch| batch.uid == *batch_uid) {
+                if processing_batch.as_ref().is_some_and(|batch| batch.uid == *batch_uid) {
                     batch_tasks |= &**processing_tasks;
                 } else {
                     batch_tasks |= self.tasks_in_batch(rtxn, *batch_uid)?;

crates/index-scheduler/src/queue/test.rs

@@ -364,7 +364,7 @@ fn test_task_queue_is_full() {
     // we won't be able to test this error in an integration test thus as a best effort test I still ensure the error return the expected error code
     snapshot!(format!("{:?}", result.error_code()), @"NoSpaceLeftOnDevice");
 
-    // Even the task deletion that doesn't delete anything shouldn't be accepted
+    // Even the task deletion and cancelation that don't delete anything should be refused
     let result = index_scheduler
         .register(
             KindWithContent::TaskDeletion { query: S("test"), tasks: RoaringBitmap::new() },
@@ -373,10 +373,39 @@ fn test_task_queue_is_full() {
         )
         .unwrap_err();
     snapshot!(result, @"Meilisearch cannot receive write operations because the limit of the task database has been reached. Please delete tasks to continue performing write operations.");
+    let result = index_scheduler
+        .register(
+            KindWithContent::TaskCancelation { query: S("test"), tasks: RoaringBitmap::new() },
+            None,
+            false,
+        )
+        .unwrap_err();
+    snapshot!(result, @"Meilisearch cannot receive write operations because the limit of the task database has been reached. Please delete tasks to continue performing write operations.");
+
     // we won't be able to test this error in an integration test thus as a best effort test I still ensure the error return the expected error code
     snapshot!(format!("{:?}", result.error_code()), @"NoSpaceLeftOnDevice");
 
-    // But a task deletion that delete something should works
+    // But a task cancelation that cancel something should work
+    index_scheduler
+        .register(
+            KindWithContent::TaskCancelation { query: S("test"), tasks: (0..100).collect() },
+            None,
+            false,
+        )
+        .unwrap();
+    handle.advance_one_successful_batch();
+
+    // But we should still be forbidden from enqueuing new tasks
+    let result = index_scheduler
+        .register(
+            KindWithContent::IndexCreation { index_uid: S("doggo"), primary_key: None },
+            None,
+            false,
+        )
+        .unwrap_err();
+    snapshot!(result, @"Meilisearch cannot receive write operations because the limit of the task database has been reached. Please delete tasks to continue performing write operations.");
+
+    // And a task deletion that delete something should works
     index_scheduler
         .register(
             KindWithContent::TaskDeletion { query: S("test"), tasks: (0..100).collect() },

crates/index-scheduler/src/scheduler/autobatcher.rs

@@ -7,10 +7,7 @@ The main function of the autobatcher is [`next_autobatch`].
 
 use std::ops::ControlFlow::{self, Break, Continue};
 
-use meilisearch_types::milli::update::IndexDocumentsMethod::{
-    self, ReplaceDocuments, UpdateDocuments,
-};
-use meilisearch_types::tasks::TaskId;
+use meilisearch_types::tasks::{BatchStopReason, PrimaryKeyMismatchReason, TaskId};
 
 use crate::KindWithContent;
 
@@ -19,19 +16,11 @@ use crate::KindWithContent;
 ///
 /// Only the non-prioritised tasks that can be grouped in a batch have a corresponding [`AutobatchKind`]
 enum AutobatchKind {
-    DocumentImport {
-        method: IndexDocumentsMethod,
-        allow_index_creation: bool,
-        primary_key: Option<String>,
-    },
+    DocumentImport { allow_index_creation: bool, primary_key: Option<String> },
     DocumentEdition,
-    DocumentDeletion {
-        by_filter: bool,
-    },
+    DocumentDeletion { by_filter: bool },
     DocumentClear,
-    Settings {
-        allow_index_creation: bool,
-    },
+    Settings { allow_index_creation: bool },
     IndexCreation,
     IndexDeletion,
     IndexUpdate,
@@ -60,11 +49,8 @@ impl From<KindWithContent> for AutobatchKind {
     fn from(kind: KindWithContent) -> Self {
         match kind {
             KindWithContent::DocumentAdditionOrUpdate {
-                method,
-                allow_index_creation,
-                primary_key,
-                ..
-            } => AutobatchKind::DocumentImport { method, allow_index_creation, primary_key },
+                allow_index_creation, primary_key, ..
+            } => AutobatchKind::DocumentImport { allow_index_creation, primary_key },
             KindWithContent::DocumentEdition { .. } => AutobatchKind::DocumentEdition,
             KindWithContent::DocumentDeletion { .. } => {
                 AutobatchKind::DocumentDeletion { by_filter: false }
@@ -85,6 +71,7 @@ impl From<KindWithContent> for AutobatchKind {
             KindWithContent::TaskCancelation { .. }
             | KindWithContent::TaskDeletion { .. }
             | KindWithContent::DumpCreation { .. }
+            | KindWithContent::Export { .. }
             | KindWithContent::UpgradeDatabase { .. }
             | KindWithContent::SnapshotCreation => {
                 panic!("The autobatcher should never be called with tasks that don't apply to an index.")
@@ -99,7 +86,6 @@ pub enum BatchKind {
         ids: Vec<TaskId>,
     },
     DocumentOperation {
-        method: IndexDocumentsMethod,
         allow_index_creation: bool,
         primary_key: Option<String>,
         operation_ids: Vec<TaskId>,
@@ -161,23 +147,48 @@ impl BatchKind {
     // TODO use an AutoBatchKind as input
     pub fn new(
         task_id: TaskId,
-        kind: KindWithContent,
+        kind_with_content: KindWithContent,
         primary_key: Option<&str>,
-    ) -> (ControlFlow<BatchKind, BatchKind>, bool) {
+    ) -> (ControlFlow<(BatchKind, BatchStopReason), BatchKind>, bool) {
         use AutobatchKind as K;
 
-        match AutobatchKind::from(kind) {
-            K::IndexCreation => (Break(BatchKind::IndexCreation { id: task_id }), true),
-            K::IndexDeletion => (Break(BatchKind::IndexDeletion { ids: vec![task_id] }), false),
-            K::IndexUpdate => (Break(BatchKind::IndexUpdate { id: task_id }), false),
-            K::IndexSwap => (Break(BatchKind::IndexSwap { id: task_id }), false),
+        let kind = kind_with_content.as_kind();
+
+        match AutobatchKind::from(kind_with_content) {
+            K::IndexCreation => (
+                Break((
+                    BatchKind::IndexCreation { id: task_id },
+                    BatchStopReason::TaskCannotBeBatched { kind, id: task_id },
+                )),
+                true,
+            ),
+            K::IndexDeletion => (
+                Break((
+                    BatchKind::IndexDeletion { ids: vec![task_id] },
+                    BatchStopReason::IndexDeletion { id: task_id },
+                )),
+                false,
+            ),
+            K::IndexUpdate => (
+                Break((
+                    BatchKind::IndexUpdate { id: task_id },
+                    BatchStopReason::TaskCannotBeBatched { kind, id: task_id },
+                )),
+                false,
+            ),
+            K::IndexSwap => (
+                Break((
+                    BatchKind::IndexSwap { id: task_id },
+                    BatchStopReason::TaskCannotBeBatched { kind, id: task_id },
+                )),
+                false,
+            ),
             K::DocumentClear => (Continue(BatchKind::DocumentClear { ids: vec![task_id] }), false),
-            K::DocumentImport { method, allow_index_creation, primary_key: pk }
+            K::DocumentImport { allow_index_creation, primary_key: pk }
                 if primary_key.is_none() || pk.is_none() || primary_key == pk.as_deref() =>
             {
                 (
                     Continue(BatchKind::DocumentOperation {
-                        method,
                         allow_index_creation,
                         primary_key: pk,
                         operation_ids: vec![task_id],
@@ -186,16 +197,28 @@ impl BatchKind {
                 )
             }
             // if the primary key set in the task was different than ours we should stop and make this batch fail asap.
-            K::DocumentImport { method, allow_index_creation, primary_key } => (
-                Break(BatchKind::DocumentOperation {
-                    method,
+            K::DocumentImport { allow_index_creation, primary_key: pk } => (
+                Break((
+                    BatchKind::DocumentOperation {
                         allow_index_creation,
-                    primary_key,
+                        primary_key: pk.clone(),
                         operation_ids: vec![task_id],
-                }),
+                    },
+                    BatchStopReason::PrimaryKeyIndexMismatch {
+                        id: task_id,
+                        in_index: primary_key.unwrap().to_owned(),
+                        in_task: pk.unwrap(),
+                    },
+                )),
                 allow_index_creation,
             ),
-            K::DocumentEdition => (Break(BatchKind::DocumentEdition { id: task_id }), false),
+            K::DocumentEdition => (
+                Break((
+                    BatchKind::DocumentEdition { id: task_id },
+                    BatchStopReason::TaskCannotBeBatched { kind, id: task_id },
+                )),
+                false,
+            ),
             K::DocumentDeletion { by_filter: includes_by_filter } => (
                 Continue(BatchKind::DocumentDeletion {
                     deletion_ids: vec![task_id],
@@ -215,54 +238,71 @@ impl BatchKind {
     /// To ease the writing of the code. `true` can be returned when you don't need to create an index
     /// but false can't be returned if you needs to create an index.
     #[rustfmt::skip]
-    fn accumulate(self, id: TaskId, kind: AutobatchKind, index_already_exists: bool, primary_key: Option<&str>) -> ControlFlow<BatchKind, BatchKind> {
+    fn accumulate(self, id: TaskId, kind_with_content: KindWithContent, index_already_exists: bool, primary_key: Option<&str>) -> ControlFlow<(BatchKind, BatchStopReason), BatchKind> {
         use AutobatchKind as K;
 
-        match (self, kind) {
+        let kind = kind_with_content.as_kind();
+        let autobatch_kind = AutobatchKind::from(kind_with_content);
+
+        let pk: Option<String> = match (self.primary_key(), autobatch_kind.primary_key(), primary_key) {
+            // 1. If incoming task don't interact with primary key -> we can continue
+            (batch_pk, None | Some(None), _) => {
+                batch_pk.flatten().map(ToOwned::to_owned)
+            },
+            // 2.1 If we already have a primary-key ->
+            // 2.1.1 If the task we're trying to accumulate have a pk it must be equal to our primary key
+            (_batch_pk, Some(Some(task_pk)), Some(index_pk)) => if task_pk == index_pk {
+                Some(task_pk.to_owned())
+            } else {
+                return Break((self, BatchStopReason::PrimaryKeyMismatch {
+                    id,
+                    reason: PrimaryKeyMismatchReason::TaskPrimaryKeyDifferFromIndexPrimaryKey {
+                        task_pk: task_pk.to_owned(),
+                        index_pk: index_pk.to_owned(),
+                    },
+                }))
+            },
+            // 2.2 If we don't have a primary-key ->
+            // 2.2.2 If the batch is set to Some(None), the task should be too
+            (Some(None), Some(Some(task_pk)), None) => return Break((self, BatchStopReason::PrimaryKeyMismatch {
+                id,
+                reason: PrimaryKeyMismatchReason::CannotInterfereWithPrimaryKeyGuessing {
+                    task_pk: task_pk.to_owned(),
+                },
+            })),
+            (Some(Some(batch_pk)), Some(Some(task_pk)), None) => if task_pk == batch_pk {
+                Some(task_pk.to_owned())
+            } else {
+                let batch_pk = batch_pk.to_owned();
+                let task_pk = task_pk.to_owned();
+                return Break((self, BatchStopReason::PrimaryKeyMismatch {
+                    id,
+                    reason: PrimaryKeyMismatchReason::TaskPrimaryKeyDifferFromCurrentBatchPrimaryKey {
+                        batch_pk,
+                        task_pk
+                    },
+                }))
+            },
+            (None, Some(Some(task_pk)), None) => Some(task_pk.to_owned())
+        };
+
+        match (self, autobatch_kind) {
             // We don't batch any of these operations
-            (this, K::IndexCreation | K::IndexUpdate | K::IndexSwap | K::DocumentEdition) => Break(this),
+            (this, K::IndexCreation | K::IndexUpdate | K::IndexSwap | K::DocumentEdition) => Break((this, BatchStopReason::TaskCannotBeBatched { kind, id })),
             // We must not batch tasks that don't have the same index creation rights if the index doesn't already exists.
             (this, kind) if !index_already_exists && this.allow_index_creation() == Some(false) && kind.allow_index_creation() == Some(true) => {
-                Break(this)
-            },
-            // NOTE: We need to negate the whole condition since we're checking if we need to break instead of continue.
-            //       I wrote it this way because it's easier to understand than the other way around.
-            (this, kind) if !(
-                // 1. If both task don't interact with primary key -> we can continue
-                (this.primary_key().is_none() && kind.primary_key().is_none()) ||
-                // 2. Else ->
-                (
-                    // 2.1 If we already have a primary-key ->
-                    (
-                        primary_key.is_some() &&
-                        // 2.1.1 If the task we're trying to accumulate have a pk it must be equal to our primary key
-                        // 2.1.2 If the task don't have a primary-key -> we can continue
-                        kind.primary_key().map_or(true, |pk| pk == primary_key)
-                    ) ||
-                    // 2.2 If we don't have a primary-key ->
-                    (
-                        // 2.2.1 If both the batch and the task have a primary key they should be equal
-                        // 2.2.2 If the batch is set to Some(None), the task should be too
-                        // 2.2.3 If the batch is set to None -> we can continue
-                        this.primary_key().zip(kind.primary_key()).map_or(true, |(this, kind)| this == kind)
-                    )
-                )
-
-                ) // closing the negation
-
-            => {
-                Break(this)
+                Break((this, BatchStopReason::IndexCreationMismatch { id }))
             },
             // The index deletion can batch with everything but must stop after
             (
                 BatchKind::DocumentClear { mut ids }
                 | BatchKind::DocumentDeletion { deletion_ids: mut ids, includes_by_filter: _ }
-                | BatchKind::DocumentOperation { method: _, allow_index_creation: _, primary_key: _, operation_ids: mut ids }
+                | BatchKind::DocumentOperation { allow_index_creation: _, primary_key: _, operation_ids: mut ids }
                 | BatchKind::Settings { allow_index_creation: _, settings_ids: mut ids },
                 K::IndexDeletion,
             ) => {
                 ids.push(id);
-                Break(BatchKind::IndexDeletion { ids })
+                Break((BatchKind::IndexDeletion { ids }, BatchStopReason::IndexDeletion { id }))
             }
             (
                 BatchKind::ClearAndSettings { settings_ids: mut ids, allow_index_creation: _, mut other },
@@ -270,7 +310,7 @@ impl BatchKind {
             ) => {
                 ids.push(id);
                 ids.append(&mut other);
-                Break(BatchKind::IndexDeletion { ids })
+                Break((BatchKind::IndexDeletion { ids }, BatchStopReason::IndexDeletion { id }))
             }
 
             (
@@ -283,51 +323,37 @@ impl BatchKind {
             (
                 this @ BatchKind::DocumentClear { .. },
                 K::DocumentImport { .. } | K::Settings { .. },
-            ) => Break(this),
+            ) => Break((this, BatchStopReason::DocumentOperationWithSettings { id })),
             (
-                BatchKind::DocumentOperation { method: _, allow_index_creation: _, primary_key: _, mut operation_ids },
+                BatchKind::DocumentOperation { allow_index_creation: _, primary_key: _, mut operation_ids },
                 K::DocumentClear,
             ) => {
                 operation_ids.push(id);
                 Continue(BatchKind::DocumentClear { ids: operation_ids })
             }
 
-            // we can autobatch the same kind of document additions / updates
+            // we can autobatch different kind of document operations and mix replacements with updates
             (
-                BatchKind::DocumentOperation { method: ReplaceDocuments, allow_index_creation, primary_key: _, mut operation_ids },
-                K::DocumentImport { method: ReplaceDocuments, primary_key: pk, .. },
+                BatchKind::DocumentOperation { allow_index_creation, primary_key: _, mut operation_ids },
+                K::DocumentImport { primary_key: _, .. },
             ) => {
                 operation_ids.push(id);
                 Continue(BatchKind::DocumentOperation {
-                    method: ReplaceDocuments,
                     allow_index_creation,
                     operation_ids,
                     primary_key: pk,
                 })
             }
             (
-                BatchKind::DocumentOperation { method: UpdateDocuments, allow_index_creation, primary_key: _, mut operation_ids },
-                K::DocumentImport { method: UpdateDocuments, primary_key: pk, .. },
-            ) => {
-                operation_ids.push(id);
-                Continue(BatchKind::DocumentOperation {
-                    method: UpdateDocuments,
-                    allow_index_creation,
-                    primary_key: pk,
-                    operation_ids,
-                })
-            }
-            (
-                BatchKind::DocumentOperation { method, allow_index_creation, primary_key, mut operation_ids },
+                BatchKind::DocumentOperation { allow_index_creation, primary_key: _, mut operation_ids },
                 K::DocumentDeletion { by_filter: false },
             ) => {
                 operation_ids.push(id);
 
                 Continue(BatchKind::DocumentOperation {
-                    method,
                     allow_index_creation,
-                    primary_key,
                     operation_ids,
+                    primary_key: pk,
                 })
             }
             // We can't batch a document operation with a delete by filter
@@ -335,19 +361,12 @@ impl BatchKind {
                 this @ BatchKind::DocumentOperation { .. },
                 K::DocumentDeletion { by_filter: true },
             ) => {
-                Break(this)
+                Break((this, BatchStopReason::DocumentOperationWithDeletionByFilter { id }))
             }
-            // but we can't autobatch documents if it's not the same kind
-            // this match branch MUST be AFTER the previous one
-            (
-                this @ BatchKind::DocumentOperation { .. },
-                K::DocumentImport { .. },
-            ) => Break(this),
-
             (
                 this @ BatchKind::DocumentOperation { .. },
                 K::Settings { .. },
-            ) => Break(this),
+            ) => Break((this, BatchStopReason::DocumentOperationWithSettings { id })),
 
             (BatchKind::DocumentDeletion { mut deletion_ids, includes_by_filter: _ }, K::DocumentClear) => {
                 deletion_ids.push(id);
@@ -357,16 +376,15 @@ impl BatchKind {
             (
                 this @ BatchKind::DocumentDeletion { deletion_ids: _, includes_by_filter: true },
                 K::DocumentImport { .. }
-            ) => Break(this),
+            ) => Break((this, BatchStopReason::DeletionByFilterWithDocumentOperation { id })),
             // we can autobatch the deletion and import if the index already exists
             (
                 BatchKind::DocumentDeletion { mut deletion_ids, includes_by_filter: false },
-                K::DocumentImport { method, allow_index_creation, primary_key }
+                K::DocumentImport { allow_index_creation, primary_key }
             ) if index_already_exists => {
                 deletion_ids.push(id);
 
                 Continue(BatchKind::DocumentOperation {
-                    method,
                     allow_index_creation,
                     primary_key,
                     operation_ids: deletion_ids,
@@ -375,29 +393,28 @@ impl BatchKind {
             // we can autobatch the deletion and import if both can't create an index
             (
                 BatchKind::DocumentDeletion { mut deletion_ids, includes_by_filter: false },
-                K::DocumentImport { method, allow_index_creation, primary_key }
+                K::DocumentImport { allow_index_creation, primary_key }
             ) if !allow_index_creation => {
                 deletion_ids.push(id);
 
                 Continue(BatchKind::DocumentOperation {
-                    method,
                     allow_index_creation,
                     primary_key,
                     operation_ids: deletion_ids,
                 })
             }
-            // we can't autobatch a deletion and an import if the index does not exists but would be created by an addition
+            // we can't autobatch a deletion and an import if the index does not exist but would be created by an addition
             (
                 this @ BatchKind::DocumentDeletion { .. },
                 K::DocumentImport { .. }
             ) => {
-                Break(this)
+                Break((this, BatchStopReason::IndexCreationMismatch { id }))
             }
             (BatchKind::DocumentDeletion { mut deletion_ids, includes_by_filter }, K::DocumentDeletion { by_filter }) => {
                 deletion_ids.push(id);
                 Continue(BatchKind::DocumentDeletion { deletion_ids, includes_by_filter: includes_by_filter | by_filter })
             }
-            (this @ BatchKind::DocumentDeletion { .. }, K::Settings { .. }) => Break(this),
+            (this @ BatchKind::DocumentDeletion { .. }, K::Settings { .. }) => Break((this, BatchStopReason::DocumentOperationWithSettings { id })),
 
             (
                 BatchKind::Settings { settings_ids, allow_index_creation },
@@ -410,7 +427,7 @@ impl BatchKind {
             (
                 this @ BatchKind::Settings { .. },
                 K::DocumentImport { .. } | K::DocumentDeletion { .. },
-            ) => Break(this),
+            ) => Break((this, BatchStopReason::SettingsWithDocumentOperation { id })),
             (
                 BatchKind::Settings { mut settings_ids, allow_index_creation },
                 K::Settings { .. },
@@ -433,7 +450,7 @@ impl BatchKind {
                     allow_index_creation,
                 })
             }
-            (this @ BatchKind::ClearAndSettings { .. }, K::DocumentImport { .. }) => Break(this),
+            (this @ BatchKind::ClearAndSettings { .. }, K::DocumentImport { .. }) => Break((this, BatchStopReason::SettingsWithDocumentOperation { id })),
             (
                 BatchKind::ClearAndSettings {
                     mut other,
@@ -489,7 +506,7 @@ pub fn autobatch(
     enqueued: Vec<(TaskId, KindWithContent)>,
     index_already_exists: bool,
     primary_key: Option<&str>,
-) -> Option<(BatchKind, bool)> {
+) -> Option<(BatchKind, bool, Option<BatchStopReason>)> {
     let mut enqueued = enqueued.into_iter();
     let (id, kind) = enqueued.next()?;
 
@@ -498,18 +515,22 @@ pub fn autobatch(
 
     let (mut acc, must_create_index) = match BatchKind::new(id, kind, primary_key) {
         (Continue(acc), create) => (acc, create),
-        (Break(acc), create) => return Some((acc, create)),
+        (Break((acc, batch_stop_reason)), create) => {
+            return Some((acc, create, Some(batch_stop_reason)))
+        }
     };
 
     // if an index has been created in the previous step we can consider it as existing.
     index_exist |= must_create_index;
 
-    for (id, kind) in enqueued {
-        acc = match acc.accumulate(id, kind.into(), index_exist, primary_key) {
+    for (id, kind_with_content) in enqueued {
+        acc = match acc.accumulate(id, kind_with_content, index_exist, primary_key) {
             Continue(acc) => acc,
-            Break(acc) => return Some((acc, must_create_index)),
+            Break((acc, batch_stop_reason)) => {
+                return Some((acc, must_create_index, Some(batch_stop_reason)))
+            }
         };
     }
 
-    Some((acc, must_create_index))
+    Some((acc, must_create_index, None))
 }

crates/index-scheduler/src/scheduler/autobatcher_test.rs

@@ -1,7 +1,7 @@
 use meilisearch_types::milli::update::IndexDocumentsMethod::{
     self, ReplaceDocuments, UpdateDocuments,
 };
-use meilisearch_types::tasks::{IndexSwap, KindWithContent};
+use meilisearch_types::tasks::{BatchStopReason, IndexSwap, KindWithContent};
 use uuid::Uuid;
 
 use self::autobatcher::{autobatch, BatchKind};
@@ -20,7 +20,7 @@ fn autobatch_from(
     index_already_exists: bool,
     primary_key: Option<&str>,
     input: impl IntoIterator<Item = KindWithContent>,
-) -> Option<(BatchKind, bool)> {
+) -> Option<(BatchKind, bool, Option<BatchStopReason>)> {
     autobatch(
         input.into_iter().enumerate().map(|(id, kind)| (id as TaskId, kind)).collect(),
         index_already_exists,
@@ -92,304 +92,304 @@ fn idx_swap() -> KindWithContent {
 fn autobatch_simple_operation_together() {
     // we can autobatch one or multiple `ReplaceDocuments` together.
     // if the index exists.
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, false, None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_imp( ReplaceDocuments, true , None), doc_imp(ReplaceDocuments, true , None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0, 1, 2] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, false, None), doc_imp( ReplaceDocuments, false , None), doc_imp(ReplaceDocuments, false , None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0, 1, 2] }, false))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, false, None)]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0] }, false, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_imp( ReplaceDocuments, true , None), doc_imp(ReplaceDocuments, true , None)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0, 1, 2] }, true, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, false, None), doc_imp( ReplaceDocuments, false , None), doc_imp(ReplaceDocuments, false , None)]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0, 1, 2] }, false, None))");
 
     // if it doesn't exists.
-    debug_snapshot!(autobatch_from(false,None,  [doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(false,None,  [doc_imp(ReplaceDocuments, false, None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(false,None,  [doc_imp(ReplaceDocuments, true, None), doc_imp( ReplaceDocuments, true , None), doc_imp(ReplaceDocuments, true , None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0, 1, 2] }, true))");
-    debug_snapshot!(autobatch_from(false,None,  [doc_imp(ReplaceDocuments, false, None), doc_imp( ReplaceDocuments, true , None), doc_imp(ReplaceDocuments, true , None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0] }, false))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, None))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_imp(ReplaceDocuments, false, None)]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0] }, false, None))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_imp(ReplaceDocuments, true, None), doc_imp( ReplaceDocuments, true , None), doc_imp(ReplaceDocuments, true , None)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0, 1, 2] }, true, None))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_imp(ReplaceDocuments, false, None), doc_imp( ReplaceDocuments, true , None), doc_imp(ReplaceDocuments, true , None)]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0] }, false, Some(IndexCreationMismatch { id: 1 })))");
 
     // we can autobatch one or multiple `UpdateDocuments` together.
     // if the index exists.
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), doc_imp(UpdateDocuments, true, None), doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0, 1, 2] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, false, None)]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, false, None), doc_imp(UpdateDocuments, false, None), doc_imp(UpdateDocuments, false, None)]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0, 1, 2] }, false))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), doc_imp(UpdateDocuments, true, None), doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0, 1, 2] }, true, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, false, None)]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0] }, false, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, false, None), doc_imp(UpdateDocuments, false, None), doc_imp(UpdateDocuments, false, None)]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0, 1, 2] }, false, None))");
 
     // if it doesn't exists.
-    debug_snapshot!(autobatch_from(false,None,  [doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(false,None,  [doc_imp(UpdateDocuments, true, None), doc_imp(UpdateDocuments, true, None), doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0, 1, 2] }, true))");
-    debug_snapshot!(autobatch_from(false,None,  [doc_imp(UpdateDocuments, false, None)]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(false,None,  [doc_imp(UpdateDocuments, false, None), doc_imp(UpdateDocuments, false, None), doc_imp(UpdateDocuments, false, None)]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0, 1, 2] }, false))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, None))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_imp(UpdateDocuments, true, None), doc_imp(UpdateDocuments, true, None), doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0, 1, 2] }, true, None))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_imp(UpdateDocuments, false, None)]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0] }, false, None))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_imp(UpdateDocuments, false, None), doc_imp(UpdateDocuments, false, None), doc_imp(UpdateDocuments, false, None)]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0, 1, 2] }, false, None))");
 
     // we can autobatch one or multiple DocumentDeletion together
-    debug_snapshot!(autobatch_from(true, None, [doc_del()]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: false }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del(), doc_del(), doc_del()]), @"Some((DocumentDeletion { deletion_ids: [0, 1, 2], includes_by_filter: false }, false))");
-    debug_snapshot!(autobatch_from(false,None,  [doc_del()]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: false }, false))");
-    debug_snapshot!(autobatch_from(false,None,  [doc_del(), doc_del(), doc_del()]), @"Some((DocumentDeletion { deletion_ids: [0, 1, 2], includes_by_filter: false }, false))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del()]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: false }, false, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del(), doc_del(), doc_del()]), @"Some((DocumentDeletion { deletion_ids: [0, 1, 2], includes_by_filter: false }, false, None))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_del()]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: false }, false, None))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_del(), doc_del(), doc_del()]), @"Some((DocumentDeletion { deletion_ids: [0, 1, 2], includes_by_filter: false }, false, None))");
 
     // we can autobatch one or multiple DocumentDeletionByFilter together
-    debug_snapshot!(autobatch_from(true, None, [doc_del_fil()]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), doc_del_fil(), doc_del_fil()]), @"Some((DocumentDeletion { deletion_ids: [0, 1, 2], includes_by_filter: true }, false))");
-    debug_snapshot!(autobatch_from(false,None,  [doc_del_fil()]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false))");
-    debug_snapshot!(autobatch_from(false,None,  [doc_del_fil(), doc_del_fil(), doc_del_fil()]), @"Some((DocumentDeletion { deletion_ids: [0, 1, 2], includes_by_filter: true }, false))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del_fil()]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), doc_del_fil(), doc_del_fil()]), @"Some((DocumentDeletion { deletion_ids: [0, 1, 2], includes_by_filter: true }, false, None))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_del_fil()]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false, None))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_del_fil(), doc_del_fil(), doc_del_fil()]), @"Some((DocumentDeletion { deletion_ids: [0, 1, 2], includes_by_filter: true }, false, None))");
 
     // we can autobatch one or multiple Settings together
-    debug_snapshot!(autobatch_from(true, None, [settings(true)]), @"Some((Settings { allow_index_creation: true, settings_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [settings(true), settings(true), settings(true)]), @"Some((Settings { allow_index_creation: true, settings_ids: [0, 1, 2] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [settings(false)]), @"Some((Settings { allow_index_creation: false, settings_ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [settings(false), settings(false), settings(false)]), @"Some((Settings { allow_index_creation: false, settings_ids: [0, 1, 2] }, false))");
+    debug_snapshot!(autobatch_from(true, None, [settings(true)]), @"Some((Settings { allow_index_creation: true, settings_ids: [0] }, true, None))");
+    debug_snapshot!(autobatch_from(true, None, [settings(true), settings(true), settings(true)]), @"Some((Settings { allow_index_creation: true, settings_ids: [0, 1, 2] }, true, None))");
+    debug_snapshot!(autobatch_from(true, None, [settings(false)]), @"Some((Settings { allow_index_creation: false, settings_ids: [0] }, false, None))");
+    debug_snapshot!(autobatch_from(true, None, [settings(false), settings(false), settings(false)]), @"Some((Settings { allow_index_creation: false, settings_ids: [0, 1, 2] }, false, None))");
 
-    debug_snapshot!(autobatch_from(false,None,  [settings(true)]), @"Some((Settings { allow_index_creation: true, settings_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(false,None,  [settings(true), settings(true), settings(true)]), @"Some((Settings { allow_index_creation: true, settings_ids: [0, 1, 2] }, true))");
-    debug_snapshot!(autobatch_from(false,None,  [settings(false)]), @"Some((Settings { allow_index_creation: false, settings_ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(false,None,  [settings(false), settings(false), settings(false)]), @"Some((Settings { allow_index_creation: false, settings_ids: [0, 1, 2] }, false))");
+    debug_snapshot!(autobatch_from(false,None,  [settings(true)]), @"Some((Settings { allow_index_creation: true, settings_ids: [0] }, true, None))");
+    debug_snapshot!(autobatch_from(false,None,  [settings(true), settings(true), settings(true)]), @"Some((Settings { allow_index_creation: true, settings_ids: [0, 1, 2] }, true, None))");
+    debug_snapshot!(autobatch_from(false,None,  [settings(false)]), @"Some((Settings { allow_index_creation: false, settings_ids: [0] }, false, None))");
+    debug_snapshot!(autobatch_from(false,None,  [settings(false), settings(false), settings(false)]), @"Some((Settings { allow_index_creation: false, settings_ids: [0, 1, 2] }, false, None))");
 
     // We can autobatch document addition with document deletion
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_del()]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), doc_del()]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, false, None), doc_del()]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, false, None), doc_del()]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, Some("catto")), doc_del()]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("catto"), operation_ids: [0, 1] }, true))"###);
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, Some("catto")), doc_del()]), @r###"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: Some("catto"), operation_ids: [0, 1] }, true))"###);
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, false, Some("catto")), doc_del()]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0, 1] }, false))"###);
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, false, Some("catto")), doc_del()]), @r###"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0, 1] }, false))"###);
-    debug_snapshot!(autobatch_from(false, None, [doc_imp(ReplaceDocuments, true, None), doc_del()]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, true))");
-    debug_snapshot!(autobatch_from(false, None, [doc_imp(UpdateDocuments, true, None), doc_del()]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, true))");
-    debug_snapshot!(autobatch_from(false, None, [doc_imp(ReplaceDocuments, false, None), doc_del()]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(false, None, [doc_imp(UpdateDocuments, false, None), doc_del()]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(false, None, [doc_imp(ReplaceDocuments, true, Some("catto")), doc_del()]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("catto"), operation_ids: [0, 1] }, true))"###);
-    debug_snapshot!(autobatch_from(false, None, [doc_imp(UpdateDocuments, true, Some("catto")), doc_del()]), @r###"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: Some("catto"), operation_ids: [0, 1] }, true))"###);
-    debug_snapshot!(autobatch_from(false, None, [doc_imp(ReplaceDocuments, false, Some("catto")), doc_del()]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0, 1] }, false))"###);
-    debug_snapshot!(autobatch_from(false, None, [doc_imp(UpdateDocuments, false, Some("catto")), doc_del()]), @r###"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0, 1] }, false))"###);
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_del()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, true, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), doc_del()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, true, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, false, None), doc_del()]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0, 1] }, false, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, false, None), doc_del()]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0, 1] }, false, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, Some("catto")), doc_del()]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("catto"), operation_ids: [0, 1] }, true, None))"###);
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, Some("catto")), doc_del()]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("catto"), operation_ids: [0, 1] }, true, None))"###);
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, false, Some("catto")), doc_del()]), @r###"Some((DocumentOperation { allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0, 1] }, false, None))"###);
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, false, Some("catto")), doc_del()]), @r###"Some((DocumentOperation { allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0, 1] }, false, None))"###);
+    debug_snapshot!(autobatch_from(false, None, [doc_imp(ReplaceDocuments, true, None), doc_del()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, true, None))");
+    debug_snapshot!(autobatch_from(false, None, [doc_imp(UpdateDocuments, true, None), doc_del()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, true, None))");
+    debug_snapshot!(autobatch_from(false, None, [doc_imp(ReplaceDocuments, false, None), doc_del()]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0, 1] }, false, None))");
+    debug_snapshot!(autobatch_from(false, None, [doc_imp(UpdateDocuments, false, None), doc_del()]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0, 1] }, false, None))");
+    debug_snapshot!(autobatch_from(false, None, [doc_imp(ReplaceDocuments, true, Some("catto")), doc_del()]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("catto"), operation_ids: [0, 1] }, true, None))"###);
+    debug_snapshot!(autobatch_from(false, None, [doc_imp(UpdateDocuments, true, Some("catto")), doc_del()]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("catto"), operation_ids: [0, 1] }, true, None))"###);
+    debug_snapshot!(autobatch_from(false, None, [doc_imp(ReplaceDocuments, false, Some("catto")), doc_del()]), @r###"Some((DocumentOperation { allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0, 1] }, false, None))"###);
+    debug_snapshot!(autobatch_from(false, None, [doc_imp(UpdateDocuments, false, Some("catto")), doc_del()]), @r###"Some((DocumentOperation { allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0, 1] }, false, None))"###);
     // And the other way around
-    debug_snapshot!(autobatch_from(true, None, [doc_del(), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del(), doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del(), doc_imp(ReplaceDocuments, false, None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del(), doc_imp(UpdateDocuments, false, None)]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del(), doc_imp(ReplaceDocuments, true, Some("catto"))]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("catto"), operation_ids: [0, 1] }, false))"###);
-    debug_snapshot!(autobatch_from(true, None, [doc_del(), doc_imp(UpdateDocuments, true, Some("catto"))]), @r###"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: Some("catto"), operation_ids: [0, 1] }, false))"###);
-    debug_snapshot!(autobatch_from(true, None, [doc_del(), doc_imp(ReplaceDocuments, false, Some("catto"))]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0, 1] }, false))"###);
-    debug_snapshot!(autobatch_from(true, None, [doc_del(), doc_imp(UpdateDocuments, false, Some("catto"))]), @r###"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0, 1] }, false))"###);
-    debug_snapshot!(autobatch_from(false, None, [doc_del(), doc_imp(ReplaceDocuments, false, None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(false, None, [doc_del(), doc_imp(UpdateDocuments, false, None)]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(false, None, [doc_del(), doc_imp(ReplaceDocuments, false, Some("catto"))]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0, 1] }, false))"###);
-    debug_snapshot!(autobatch_from(false, None, [doc_del(), doc_imp(UpdateDocuments, false, Some("catto"))]), @r###"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0, 1] }, false))"###);
+    debug_snapshot!(autobatch_from(true, None, [doc_del(), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, false, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del(), doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, false, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del(), doc_imp(ReplaceDocuments, false, None)]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0, 1] }, false, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del(), doc_imp(UpdateDocuments, false, None)]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0, 1] }, false, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del(), doc_imp(ReplaceDocuments, true, Some("catto"))]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("catto"), operation_ids: [0, 1] }, false, None))"###);
+    debug_snapshot!(autobatch_from(true, None, [doc_del(), doc_imp(UpdateDocuments, true, Some("catto"))]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("catto"), operation_ids: [0, 1] }, false, None))"###);
+    debug_snapshot!(autobatch_from(true, None, [doc_del(), doc_imp(ReplaceDocuments, false, Some("catto"))]), @r###"Some((DocumentOperation { allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0, 1] }, false, None))"###);
+    debug_snapshot!(autobatch_from(true, None, [doc_del(), doc_imp(UpdateDocuments, false, Some("catto"))]), @r###"Some((DocumentOperation { allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0, 1] }, false, None))"###);
+    debug_snapshot!(autobatch_from(false, None, [doc_del(), doc_imp(ReplaceDocuments, false, None)]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0, 1] }, false, None))");
+    debug_snapshot!(autobatch_from(false, None, [doc_del(), doc_imp(UpdateDocuments, false, None)]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0, 1] }, false, None))");
+    debug_snapshot!(autobatch_from(false, None, [doc_del(), doc_imp(ReplaceDocuments, false, Some("catto"))]), @r###"Some((DocumentOperation { allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0, 1] }, false, None))"###);
+    debug_snapshot!(autobatch_from(false, None, [doc_del(), doc_imp(UpdateDocuments, false, Some("catto"))]), @r###"Some((DocumentOperation { allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0, 1] }, false, None))"###);
 
     // But we can't autobatch document addition with document deletion by filter
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_del_fil()]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), doc_del_fil()]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, false, None), doc_del_fil()]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, false, None), doc_del_fil()]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, Some("catto")), doc_del_fil()]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("catto"), operation_ids: [0] }, true))"###);
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, Some("catto")), doc_del_fil()]), @r###"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: Some("catto"), operation_ids: [0] }, true))"###);
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, false, Some("catto")), doc_del_fil()]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0] }, false))"###);
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, false, Some("catto")), doc_del_fil()]), @r###"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0] }, false))"###);
-    debug_snapshot!(autobatch_from(false, None, [doc_imp(ReplaceDocuments, true, None), doc_del_fil()]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(false, None, [doc_imp(UpdateDocuments, true, None), doc_del_fil()]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(false, None, [doc_imp(ReplaceDocuments, false, None), doc_del_fil()]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(false, None, [doc_imp(UpdateDocuments, false, None), doc_del_fil()]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(false, None, [doc_imp(ReplaceDocuments, true, Some("catto")), doc_del_fil()]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("catto"), operation_ids: [0] }, true))"###);
-    debug_snapshot!(autobatch_from(false, None, [doc_imp(UpdateDocuments, true, Some("catto")), doc_del_fil()]), @r###"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: Some("catto"), operation_ids: [0] }, true))"###);
-    debug_snapshot!(autobatch_from(false, None, [doc_imp(ReplaceDocuments, false, Some("catto")), doc_del_fil()]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0] }, false))"###);
-    debug_snapshot!(autobatch_from(false, None, [doc_imp(UpdateDocuments, false, Some("catto")), doc_del_fil()]), @r###"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0] }, false))"###);
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_del_fil()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithDeletionByFilter { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), doc_del_fil()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithDeletionByFilter { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, false, None), doc_del_fil()]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0] }, false, Some(DocumentOperationWithDeletionByFilter { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, false, None), doc_del_fil()]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0] }, false, Some(DocumentOperationWithDeletionByFilter { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, Some("catto")), doc_del_fil()]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("catto"), operation_ids: [0] }, true, Some(DocumentOperationWithDeletionByFilter { id: 1 })))"###);
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, Some("catto")), doc_del_fil()]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("catto"), operation_ids: [0] }, true, Some(DocumentOperationWithDeletionByFilter { id: 1 })))"###);
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, false, Some("catto")), doc_del_fil()]), @r###"Some((DocumentOperation { allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0] }, false, Some(DocumentOperationWithDeletionByFilter { id: 1 })))"###);
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, false, Some("catto")), doc_del_fil()]), @r###"Some((DocumentOperation { allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0] }, false, Some(DocumentOperationWithDeletionByFilter { id: 1 })))"###);
+    debug_snapshot!(autobatch_from(false, None, [doc_imp(ReplaceDocuments, true, None), doc_del_fil()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithDeletionByFilter { id: 1 })))");
+    debug_snapshot!(autobatch_from(false, None, [doc_imp(UpdateDocuments, true, None), doc_del_fil()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithDeletionByFilter { id: 1 })))");
+    debug_snapshot!(autobatch_from(false, None, [doc_imp(ReplaceDocuments, false, None), doc_del_fil()]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0] }, false, Some(DocumentOperationWithDeletionByFilter { id: 1 })))");
+    debug_snapshot!(autobatch_from(false, None, [doc_imp(UpdateDocuments, false, None), doc_del_fil()]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0] }, false, Some(DocumentOperationWithDeletionByFilter { id: 1 })))");
+    debug_snapshot!(autobatch_from(false, None, [doc_imp(ReplaceDocuments, true, Some("catto")), doc_del_fil()]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("catto"), operation_ids: [0] }, true, Some(DocumentOperationWithDeletionByFilter { id: 1 })))"###);
+    debug_snapshot!(autobatch_from(false, None, [doc_imp(UpdateDocuments, true, Some("catto")), doc_del_fil()]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("catto"), operation_ids: [0] }, true, Some(DocumentOperationWithDeletionByFilter { id: 1 })))"###);
+    debug_snapshot!(autobatch_from(false, None, [doc_imp(ReplaceDocuments, false, Some("catto")), doc_del_fil()]), @r###"Some((DocumentOperation { allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0] }, false, Some(DocumentOperationWithDeletionByFilter { id: 1 })))"###);
+    debug_snapshot!(autobatch_from(false, None, [doc_imp(UpdateDocuments, false, Some("catto")), doc_del_fil()]), @r###"Some((DocumentOperation { allow_index_creation: false, primary_key: Some("catto"), operation_ids: [0] }, false, Some(DocumentOperationWithDeletionByFilter { id: 1 })))"###);
     // And the other way around
-    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), doc_imp(ReplaceDocuments, false, None)]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), doc_imp(UpdateDocuments, false, None)]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), doc_imp(ReplaceDocuments, true, Some("catto"))]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), doc_imp(UpdateDocuments, true, Some("catto"))]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), doc_imp(ReplaceDocuments, false, Some("catto"))]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), doc_imp(UpdateDocuments, false, Some("catto"))]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false))");
-    debug_snapshot!(autobatch_from(false, None, [doc_del_fil(), doc_imp(ReplaceDocuments, false, None)]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false))");
-    debug_snapshot!(autobatch_from(false, None, [doc_del_fil(), doc_imp(UpdateDocuments, false, None)]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false))");
-    debug_snapshot!(autobatch_from(false, None, [doc_del_fil(), doc_imp(ReplaceDocuments, false, Some("catto"))]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false))");
-    debug_snapshot!(autobatch_from(false, None, [doc_del_fil(), doc_imp(UpdateDocuments, false, Some("catto"))]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false, Some(DeletionByFilterWithDocumentOperation { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false, Some(DeletionByFilterWithDocumentOperation { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), doc_imp(ReplaceDocuments, false, None)]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false, Some(DeletionByFilterWithDocumentOperation { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), doc_imp(UpdateDocuments, false, None)]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false, Some(DeletionByFilterWithDocumentOperation { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), doc_imp(ReplaceDocuments, true, Some("catto"))]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false, Some(DeletionByFilterWithDocumentOperation { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), doc_imp(UpdateDocuments, true, Some("catto"))]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false, Some(DeletionByFilterWithDocumentOperation { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), doc_imp(ReplaceDocuments, false, Some("catto"))]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false, Some(DeletionByFilterWithDocumentOperation { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), doc_imp(UpdateDocuments, false, Some("catto"))]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false, Some(DeletionByFilterWithDocumentOperation { id: 1 })))");
+    debug_snapshot!(autobatch_from(false, None, [doc_del_fil(), doc_imp(ReplaceDocuments, false, None)]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false, Some(DeletionByFilterWithDocumentOperation { id: 1 })))");
+    debug_snapshot!(autobatch_from(false, None, [doc_del_fil(), doc_imp(UpdateDocuments, false, None)]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false, Some(DeletionByFilterWithDocumentOperation { id: 1 })))");
+    debug_snapshot!(autobatch_from(false, None, [doc_del_fil(), doc_imp(ReplaceDocuments, false, Some("catto"))]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false, Some(DeletionByFilterWithDocumentOperation { id: 1 })))");
+    debug_snapshot!(autobatch_from(false, None, [doc_del_fil(), doc_imp(UpdateDocuments, false, Some("catto"))]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false, Some(DeletionByFilterWithDocumentOperation { id: 1 })))");
 }
 
 #[test]
-fn simple_document_operation_dont_autobatch_with_other() {
-    // addition, updates and deletion by filter can't batch together
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), doc_del_fil()]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_del_fil()]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false))");
+fn simple_different_document_operations_autobatch_together() {
+    // addition and updates with deletion by filter can't batch together
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, true, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, true, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), doc_del_fil()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithDeletionByFilter { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_del_fil()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithDeletionByFilter { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false, Some(DeletionByFilterWithDocumentOperation { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false, Some(DeletionByFilterWithDocumentOperation { id: 1 })))");
 
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), idx_create()]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), idx_create()]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del(), idx_create()]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: false }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), idx_create()]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), idx_create()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(TaskCannotBeBatched { kind: IndexCreation, id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), idx_create()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(TaskCannotBeBatched { kind: IndexCreation, id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del(), idx_create()]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: false }, false, Some(TaskCannotBeBatched { kind: IndexCreation, id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), idx_create()]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false, Some(TaskCannotBeBatched { kind: IndexCreation, id: 1 })))");
 
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), idx_update()]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), idx_update()]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del(), idx_update()]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: false }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), idx_update()]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), idx_update()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(TaskCannotBeBatched { kind: IndexUpdate, id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), idx_update()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(TaskCannotBeBatched { kind: IndexUpdate, id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del(), idx_update()]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: false }, false, Some(TaskCannotBeBatched { kind: IndexUpdate, id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), idx_update()]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false, Some(TaskCannotBeBatched { kind: IndexUpdate, id: 1 })))");
 
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), idx_swap()]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), idx_swap()]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del(), idx_swap()]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: false }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), idx_swap()]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), idx_swap()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(TaskCannotBeBatched { kind: IndexSwap, id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), idx_swap()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(TaskCannotBeBatched { kind: IndexSwap, id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del(), idx_swap()]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: false }, false, Some(TaskCannotBeBatched { kind: IndexSwap, id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), idx_swap()]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: true }, false, Some(TaskCannotBeBatched { kind: IndexSwap, id: 1 })))");
 }
 
 #[test]
 fn document_addition_doesnt_batch_with_settings() {
     // simple case
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), settings(true)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), settings(true)]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), settings(true)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithSettings { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), settings(true)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithSettings { id: 1 })))");
 
     // multiple settings and doc addition
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, None), settings(true), settings(true)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, None), settings(true), settings(true)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, true))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, None), settings(true), settings(true)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, true, Some(DocumentOperationWithSettings { id: 2 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, None), settings(true), settings(true)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, true, Some(DocumentOperationWithSettings { id: 2 })))");
 
     // addition and setting unordered
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), settings(true), doc_imp(ReplaceDocuments, true, None), settings(true)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), settings(true), doc_imp(UpdateDocuments, true, None), settings(true)]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), settings(true), doc_imp(ReplaceDocuments, true, None), settings(true)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithSettings { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), settings(true), doc_imp(UpdateDocuments, true, None), settings(true)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithSettings { id: 1 })))");
 
     // Doesn't batch with other forbidden operations
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), settings(true), doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), settings(true), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), settings(true), doc_del()]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), settings(true), doc_del()]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), settings(true), idx_create()]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), settings(true), idx_create()]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), settings(true), idx_update()]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), settings(true), idx_update()]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), settings(true), idx_swap()]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), settings(true), idx_swap()]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), settings(true), doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithSettings { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), settings(true), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithSettings { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), settings(true), doc_del()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithSettings { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), settings(true), doc_del()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithSettings { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), settings(true), idx_create()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithSettings { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), settings(true), idx_create()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithSettings { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), settings(true), idx_update()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithSettings { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), settings(true), idx_update()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithSettings { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), settings(true), idx_swap()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithSettings { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), settings(true), idx_swap()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithSettings { id: 1 })))");
 }
 
 #[test]
 fn clear_and_additions() {
     // these two doesn't need to batch
-    debug_snapshot!(autobatch_from(true, None, [doc_clr(), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentClear { ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_clr(), doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentClear { ids: [0] }, false))");
+    debug_snapshot!(autobatch_from(true, None, [doc_clr(), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentClear { ids: [0] }, false, Some(DocumentOperationWithSettings { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_clr(), doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentClear { ids: [0] }, false, Some(DocumentOperationWithSettings { id: 1 })))");
 
     // Basic use case
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, None), doc_clr()]), @"Some((DocumentClear { ids: [0, 1, 2] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), doc_imp(UpdateDocuments, true, None), doc_clr()]), @"Some((DocumentClear { ids: [0, 1, 2] }, true))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, None), doc_clr()]), @"Some((DocumentClear { ids: [0, 1, 2] }, true, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), doc_imp(UpdateDocuments, true, None), doc_clr()]), @"Some((DocumentClear { ids: [0, 1, 2] }, true, None))");
 
     // This batch kind doesn't mix with other document addition
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, None), doc_clr(), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentClear { ids: [0, 1, 2] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), doc_imp(UpdateDocuments, true, None), doc_clr(), doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentClear { ids: [0, 1, 2] }, true))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, None), doc_clr(), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentClear { ids: [0, 1, 2] }, true, Some(DocumentOperationWithSettings { id: 3 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), doc_imp(UpdateDocuments, true, None), doc_clr(), doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentClear { ids: [0, 1, 2] }, true, Some(DocumentOperationWithSettings { id: 3 })))");
 
     // But you can batch multiple clear together
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, None), doc_clr(), doc_clr(), doc_clr()]), @"Some((DocumentClear { ids: [0, 1, 2, 3, 4] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), doc_imp(UpdateDocuments, true, None), doc_clr(), doc_clr(), doc_clr()]), @"Some((DocumentClear { ids: [0, 1, 2, 3, 4] }, true))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, None), doc_clr(), doc_clr(), doc_clr()]), @"Some((DocumentClear { ids: [0, 1, 2, 3, 4] }, true, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), doc_imp(UpdateDocuments, true, None), doc_clr(), doc_clr(), doc_clr()]), @"Some((DocumentClear { ids: [0, 1, 2, 3, 4] }, true, None))");
 }
 
 #[test]
 fn clear_and_additions_and_settings() {
     // A clear don't need to autobatch the settings that happens AFTER there is no documents
-    debug_snapshot!(autobatch_from(true, None, [doc_clr(), settings(true)]), @"Some((DocumentClear { ids: [0] }, false))");
+    debug_snapshot!(autobatch_from(true, None, [doc_clr(), settings(true)]), @"Some((DocumentClear { ids: [0] }, false, Some(DocumentOperationWithSettings { id: 1 })))");
 
-    debug_snapshot!(autobatch_from(true, None, [settings(true), doc_clr(), settings(true)]), @"Some((ClearAndSettings { other: [1], allow_index_creation: true, settings_ids: [0, 2] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), settings(true), doc_clr()]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), settings(true), doc_clr()]), @"Some((DocumentOperation { method: UpdateDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
+    debug_snapshot!(autobatch_from(true, None, [settings(true), doc_clr(), settings(true)]), @"Some((ClearAndSettings { other: [1], allow_index_creation: true, settings_ids: [0, 2] }, true, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), settings(true), doc_clr()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithSettings { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), settings(true), doc_clr()]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithSettings { id: 1 })))");
 }
 
 #[test]
 fn anything_and_index_deletion() {
     // The `IndexDeletion` doesn't batch with anything that happens AFTER.
-    debug_snapshot!(autobatch_from(true, None, [idx_del(), doc_imp(ReplaceDocuments, true, None)]), @"Some((IndexDeletion { ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [idx_del(), doc_imp(UpdateDocuments, true, None)]), @"Some((IndexDeletion { ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [idx_del(), doc_imp(ReplaceDocuments, false, None)]), @"Some((IndexDeletion { ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [idx_del(), doc_imp(UpdateDocuments, false, None)]), @"Some((IndexDeletion { ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [idx_del(), doc_del()]), @"Some((IndexDeletion { ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [idx_del(), doc_del_fil()]), @"Some((IndexDeletion { ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [idx_del(), doc_clr()]), @"Some((IndexDeletion { ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [idx_del(), settings(true)]), @"Some((IndexDeletion { ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [idx_del(), settings(false)]), @"Some((IndexDeletion { ids: [0] }, false))");
+    debug_snapshot!(autobatch_from(true, None, [idx_del(), doc_imp(ReplaceDocuments, true, None)]), @"Some((IndexDeletion { ids: [0] }, false, Some(IndexDeletion { id: 0 })))");
+    debug_snapshot!(autobatch_from(true, None, [idx_del(), doc_imp(UpdateDocuments, true, None)]), @"Some((IndexDeletion { ids: [0] }, false, Some(IndexDeletion { id: 0 })))");
+    debug_snapshot!(autobatch_from(true, None, [idx_del(), doc_imp(ReplaceDocuments, false, None)]), @"Some((IndexDeletion { ids: [0] }, false, Some(IndexDeletion { id: 0 })))");
+    debug_snapshot!(autobatch_from(true, None, [idx_del(), doc_imp(UpdateDocuments, false, None)]), @"Some((IndexDeletion { ids: [0] }, false, Some(IndexDeletion { id: 0 })))");
+    debug_snapshot!(autobatch_from(true, None, [idx_del(), doc_del()]), @"Some((IndexDeletion { ids: [0] }, false, Some(IndexDeletion { id: 0 })))");
+    debug_snapshot!(autobatch_from(true, None, [idx_del(), doc_del_fil()]), @"Some((IndexDeletion { ids: [0] }, false, Some(IndexDeletion { id: 0 })))");
+    debug_snapshot!(autobatch_from(true, None, [idx_del(), doc_clr()]), @"Some((IndexDeletion { ids: [0] }, false, Some(IndexDeletion { id: 0 })))");
+    debug_snapshot!(autobatch_from(true, None, [idx_del(), settings(true)]), @"Some((IndexDeletion { ids: [0] }, false, Some(IndexDeletion { id: 0 })))");
+    debug_snapshot!(autobatch_from(true, None, [idx_del(), settings(false)]), @"Some((IndexDeletion { ids: [0] }, false, Some(IndexDeletion { id: 0 })))");
 
-    debug_snapshot!(autobatch_from(false,None,  [idx_del(), doc_imp(ReplaceDocuments, true, None)]), @"Some((IndexDeletion { ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(false,None,  [idx_del(), doc_imp(UpdateDocuments, true, None)]), @"Some((IndexDeletion { ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(false,None,  [idx_del(), doc_imp(ReplaceDocuments, false, None)]), @"Some((IndexDeletion { ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(false,None,  [idx_del(), doc_imp(UpdateDocuments, false, None)]), @"Some((IndexDeletion { ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(false,None,  [idx_del(), doc_del()]), @"Some((IndexDeletion { ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(false,None,  [idx_del(), doc_del_fil()]), @"Some((IndexDeletion { ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(false,None,  [idx_del(), doc_clr()]), @"Some((IndexDeletion { ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(false,None,  [idx_del(), settings(true)]), @"Some((IndexDeletion { ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(false,None,  [idx_del(), settings(false)]), @"Some((IndexDeletion { ids: [0] }, false))");
+    debug_snapshot!(autobatch_from(false,None,  [idx_del(), doc_imp(ReplaceDocuments, true, None)]), @"Some((IndexDeletion { ids: [0] }, false, Some(IndexDeletion { id: 0 })))");
+    debug_snapshot!(autobatch_from(false,None,  [idx_del(), doc_imp(UpdateDocuments, true, None)]), @"Some((IndexDeletion { ids: [0] }, false, Some(IndexDeletion { id: 0 })))");
+    debug_snapshot!(autobatch_from(false,None,  [idx_del(), doc_imp(ReplaceDocuments, false, None)]), @"Some((IndexDeletion { ids: [0] }, false, Some(IndexDeletion { id: 0 })))");
+    debug_snapshot!(autobatch_from(false,None,  [idx_del(), doc_imp(UpdateDocuments, false, None)]), @"Some((IndexDeletion { ids: [0] }, false, Some(IndexDeletion { id: 0 })))");
+    debug_snapshot!(autobatch_from(false,None,  [idx_del(), doc_del()]), @"Some((IndexDeletion { ids: [0] }, false, Some(IndexDeletion { id: 0 })))");
+    debug_snapshot!(autobatch_from(false,None,  [idx_del(), doc_del_fil()]), @"Some((IndexDeletion { ids: [0] }, false, Some(IndexDeletion { id: 0 })))");
+    debug_snapshot!(autobatch_from(false,None,  [idx_del(), doc_clr()]), @"Some((IndexDeletion { ids: [0] }, false, Some(IndexDeletion { id: 0 })))");
+    debug_snapshot!(autobatch_from(false,None,  [idx_del(), settings(true)]), @"Some((IndexDeletion { ids: [0] }, false, Some(IndexDeletion { id: 0 })))");
+    debug_snapshot!(autobatch_from(false,None,  [idx_del(), settings(false)]), @"Some((IndexDeletion { ids: [0] }, false, Some(IndexDeletion { id: 0 })))");
 
     // The index deletion can accept almost any type of `BatchKind` and transform it to an `IndexDeletion`.
     // First, the basic cases
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, false, None), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, false, None), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del(), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_clr(), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [settings(true), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [settings(false), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, true, Some(IndexDeletion { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, true, None), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, true, Some(IndexDeletion { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, false, None), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false, Some(IndexDeletion { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(UpdateDocuments, false, None), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false, Some(IndexDeletion { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del(), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false, Some(IndexDeletion { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_del_fil(), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false, Some(IndexDeletion { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_clr(), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false, Some(IndexDeletion { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [settings(true), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, true, Some(IndexDeletion { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [settings(false), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false, Some(IndexDeletion { id: 1 })))");
 
-    debug_snapshot!(autobatch_from(false,None,  [doc_imp(ReplaceDocuments, true, None), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, true))");
-    debug_snapshot!(autobatch_from(false,None,  [doc_imp(UpdateDocuments, true, None), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, true))");
-    debug_snapshot!(autobatch_from(false,None,  [doc_imp(ReplaceDocuments, false, None), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(false,None,  [doc_imp(UpdateDocuments, false, None), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(false,None,  [doc_del(), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(false,None,  [doc_del_fil(), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(false,None,  [doc_clr(), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(false,None,  [settings(true), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, true))");
-    debug_snapshot!(autobatch_from(false,None,  [settings(false), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_imp(ReplaceDocuments, true, None), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, true, Some(IndexDeletion { id: 1 })))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_imp(UpdateDocuments, true, None), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, true, Some(IndexDeletion { id: 1 })))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_imp(ReplaceDocuments, false, None), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false, Some(IndexDeletion { id: 1 })))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_imp(UpdateDocuments, false, None), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false, Some(IndexDeletion { id: 1 })))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_del(), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false, Some(IndexDeletion { id: 1 })))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_del_fil(), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false, Some(IndexDeletion { id: 1 })))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_clr(), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false, Some(IndexDeletion { id: 1 })))");
+    debug_snapshot!(autobatch_from(false,None,  [settings(true), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, true, Some(IndexDeletion { id: 1 })))");
+    debug_snapshot!(autobatch_from(false,None,  [settings(false), idx_del()]), @"Some((IndexDeletion { ids: [0, 1] }, false, Some(IndexDeletion { id: 1 })))");
 }
 
 #[test]
 fn allowed_and_disallowed_index_creation() {
     // `DocumentImport` can't be mixed with those disallowed to do so except if the index already exists.
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, false, None), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, false, None), doc_imp(ReplaceDocuments, false, None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), settings(true)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, false, None), settings(true)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0] }, false))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, false, None), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0, 1] }, false, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, true, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, false, None), doc_imp(ReplaceDocuments, false, None)]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0, 1] }, false, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), settings(true)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithSettings { id: 1 })))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, false, None), settings(true)]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0] }, false, Some(DocumentOperationWithSettings { id: 1 })))");
 
-    debug_snapshot!(autobatch_from(false,None,  [doc_imp(ReplaceDocuments, false, None), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0] }, false))");
-    debug_snapshot!(autobatch_from(false,None,  [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, true))");
-    debug_snapshot!(autobatch_from(false,None,  [doc_imp(ReplaceDocuments, false, None), doc_imp(ReplaceDocuments, false, None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0, 1] }, false))");
-    debug_snapshot!(autobatch_from(false,None,  [doc_imp(ReplaceDocuments, true, None), settings(true)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(false,None,  [doc_imp(ReplaceDocuments, false, None), settings(true)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: false, primary_key: None, operation_ids: [0] }, false))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_imp(ReplaceDocuments, false, None), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0] }, false, Some(IndexCreationMismatch { id: 1 })))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, true, None))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_imp(ReplaceDocuments, false, None), doc_imp(ReplaceDocuments, false, None)]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0, 1] }, false, None))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_imp(ReplaceDocuments, true, None), settings(true)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(DocumentOperationWithSettings { id: 1 })))");
+    debug_snapshot!(autobatch_from(false,None,  [doc_imp(ReplaceDocuments, false, None), settings(true)]), @"Some((DocumentOperation { allow_index_creation: false, primary_key: None, operation_ids: [0] }, false, Some(IndexCreationMismatch { id: 1 })))");
 
     // batch deletion and addition
-    debug_snapshot!(autobatch_from(false, None, [doc_del(), doc_imp(ReplaceDocuments, true, Some("catto"))]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: false }, false))");
-    debug_snapshot!(autobatch_from(false, None, [doc_del(), doc_imp(UpdateDocuments, true, Some("catto"))]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: false }, false))");
-    debug_snapshot!(autobatch_from(false, None, [doc_del(), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: false }, false))");
-    debug_snapshot!(autobatch_from(false, None, [doc_del(), doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: false }, false))");
+    debug_snapshot!(autobatch_from(false, None, [doc_del(), doc_imp(ReplaceDocuments, true, Some("catto"))]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: false }, false, Some(IndexCreationMismatch { id: 1 })))");
+    debug_snapshot!(autobatch_from(false, None, [doc_del(), doc_imp(UpdateDocuments, true, Some("catto"))]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: false }, false, Some(IndexCreationMismatch { id: 1 })))");
+    debug_snapshot!(autobatch_from(false, None, [doc_del(), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: false }, false, Some(IndexCreationMismatch { id: 1 })))");
+    debug_snapshot!(autobatch_from(false, None, [doc_del(), doc_imp(UpdateDocuments, true, None)]), @"Some((DocumentDeletion { deletion_ids: [0], includes_by_filter: false }, false, Some(IndexCreationMismatch { id: 1 })))");
 }
 
 #[test]
 fn autobatch_primary_key() {
     // ==> If I have a pk
     // With a single update
-    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("id"))]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("id"), operation_ids: [0] }, true))"###);
-    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("other"))]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("other"), operation_ids: [0] }, true))"###);
+    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, None))");
+    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("id"))]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("id"), operation_ids: [0] }, true, None))"###);
+    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("other"))]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("other"), operation_ids: [0] }, true, Some(PrimaryKeyIndexMismatch { id: 0, in_index: "id", in_task: "other" })))"###);
 
     // With a multiple updates
-    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, true))");
-    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, Some("id"))]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("id"), operation_ids: [0, 1] }, true))"###);
-    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, Some("id")), doc_imp(ReplaceDocuments, true, None)]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("id"), operation_ids: [0, 1] }, true))"###);
-    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, Some("other"))]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, Some("id"))]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
+    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, true, None))");
+    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, Some("id"))]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("id"), operation_ids: [0, 1] }, true, None))"###);
+    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, Some("id")), doc_imp(ReplaceDocuments, true, None)]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("id"), operation_ids: [0, 1, 2] }, true, None))"###);
+    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, Some("other"))]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(PrimaryKeyMismatch { id: 1, reason: TaskPrimaryKeyDifferFromIndexPrimaryKey { task_pk: "other", index_pk: "id" } })))"###);
+    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, None)]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(PrimaryKeyMismatch { id: 1, reason: TaskPrimaryKeyDifferFromIndexPrimaryKey { task_pk: "other", index_pk: "id" } })))"###);
+    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, Some("id"))]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(PrimaryKeyMismatch { id: 1, reason: TaskPrimaryKeyDifferFromIndexPrimaryKey { task_pk: "other", index_pk: "id" } })))"###);
 
-    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("id")), doc_imp(ReplaceDocuments, true, None)]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("id"), operation_ids: [0] }, true))"###);
-    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("id")), doc_imp(ReplaceDocuments, true, Some("id"))]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("id"), operation_ids: [0, 1] }, true))"###);
-    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("id")), doc_imp(ReplaceDocuments, true, Some("id")), doc_imp(ReplaceDocuments, true, None)]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("id"), operation_ids: [0, 1] }, true))"###);
-    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("id")), doc_imp(ReplaceDocuments, true, Some("other"))]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("id"), operation_ids: [0] }, true))"###);
-    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("id")), doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, None)]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("id"), operation_ids: [0] }, true))"###);
-    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("id")), doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, Some("id"))]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("id"), operation_ids: [0] }, true))"###);
+    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("id")), doc_imp(ReplaceDocuments, true, None)]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("id"), operation_ids: [0, 1] }, true, None))"###);
+    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("id")), doc_imp(ReplaceDocuments, true, Some("id"))]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("id"), operation_ids: [0, 1] }, true, None))"###);
+    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("id")), doc_imp(ReplaceDocuments, true, Some("id")), doc_imp(ReplaceDocuments, true, None)]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("id"), operation_ids: [0, 1, 2] }, true, None))"###);
+    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("id")), doc_imp(ReplaceDocuments, true, Some("other"))]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("id"), operation_ids: [0] }, true, Some(PrimaryKeyMismatch { id: 1, reason: TaskPrimaryKeyDifferFromIndexPrimaryKey { task_pk: "other", index_pk: "id" } })))"###);
+    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("id")), doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, None)]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("id"), operation_ids: [0] }, true, Some(PrimaryKeyMismatch { id: 1, reason: TaskPrimaryKeyDifferFromIndexPrimaryKey { task_pk: "other", index_pk: "id" } })))"###);
+    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("id")), doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, Some("id"))]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("id"), operation_ids: [0] }, true, Some(PrimaryKeyMismatch { id: 1, reason: TaskPrimaryKeyDifferFromIndexPrimaryKey { task_pk: "other", index_pk: "id" } })))"###);
 
-    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, None)]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("other"), operation_ids: [0] }, true))"###);
-    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, Some("id"))]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("other"), operation_ids: [0] }, true))"###);
-    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, Some("id")), doc_imp(ReplaceDocuments, true, None)]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("other"), operation_ids: [0] }, true))"###);
-    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, Some("other"))]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("other"), operation_ids: [0] }, true))"###);
-    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, None)]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("other"), operation_ids: [0] }, true))"###);
-    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, Some("id"))]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("other"), operation_ids: [0] }, true))"###);
+    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, None)]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("other"), operation_ids: [0] }, true, Some(PrimaryKeyIndexMismatch { id: 0, in_index: "id", in_task: "other" })))"###);
+    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, Some("id"))]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("other"), operation_ids: [0] }, true, Some(PrimaryKeyIndexMismatch { id: 0, in_index: "id", in_task: "other" })))"###);
+    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, Some("id")), doc_imp(ReplaceDocuments, true, None)]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("other"), operation_ids: [0] }, true, Some(PrimaryKeyIndexMismatch { id: 0, in_index: "id", in_task: "other" })))"###);
+    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, Some("other"))]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("other"), operation_ids: [0] }, true, Some(PrimaryKeyIndexMismatch { id: 0, in_index: "id", in_task: "other" })))"###);
+    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, None)]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("other"), operation_ids: [0] }, true, Some(PrimaryKeyIndexMismatch { id: 0, in_index: "id", in_task: "other" })))"###);
+    debug_snapshot!(autobatch_from(true, Some("id"), [doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, Some("other")), doc_imp(ReplaceDocuments, true, Some("id"))]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("other"), operation_ids: [0] }, true, Some(PrimaryKeyIndexMismatch { id: 0, in_index: "id", in_task: "other" })))"###);
 
     // ==> If I don't have a pk
     // With a single update
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, Some("id"))]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("id"), operation_ids: [0] }, true))"###);
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, Some("other"))]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("other"), operation_ids: [0] }, true))"###);
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, Some("id"))]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("id"), operation_ids: [0] }, true, None))"###);
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, Some("other"))]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("other"), operation_ids: [0] }, true, None))"###);
 
     // With a multiple updates
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, Some("id"))]), @"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: None, operation_ids: [0] }, true))");
-    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, Some("id")), doc_imp(ReplaceDocuments, true, None)]), @r###"Some((DocumentOperation { method: ReplaceDocuments, allow_index_creation: true, primary_key: Some("id"), operation_ids: [0] }, true))"###);
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, None)]), @"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0, 1] }, true, None))");
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, None), doc_imp(ReplaceDocuments, true, Some("id"))]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: None, operation_ids: [0] }, true, Some(PrimaryKeyMismatch { id: 1, reason: CannotInterfereWithPrimaryKeyGuessing { task_pk: "id" } })))"###);
+    debug_snapshot!(autobatch_from(true, None, [doc_imp(ReplaceDocuments, true, Some("id")), doc_imp(ReplaceDocuments, true, None)]), @r###"Some((DocumentOperation { allow_index_creation: true, primary_key: Some("id"), operation_ids: [0, 1] }, true, None))"###);
 }

crates/index-scheduler/src/scheduler/create_batch.rs

@@ -1,9 +1,10 @@
 use std::fmt;
+use std::io::ErrorKind;
 
 use meilisearch_types::heed::RoTxn;
 use meilisearch_types::milli::update::IndexDocumentsMethod;
 use meilisearch_types::settings::{Settings, Unchecked};
-use meilisearch_types::tasks::{Kind, KindWithContent, Status, Task};
+use meilisearch_types::tasks::{BatchStopReason, Kind, KindWithContent, Status, Task};
 use roaring::RoaringBitmap;
 use uuid::Uuid;
 
@@ -47,6 +48,9 @@ pub(crate) enum Batch {
     IndexSwap {
         task: Task,
     },
+    Export {
+        task: Task,
+    },
     UpgradeDatabase {
         tasks: Vec<Task>,
     },
@@ -54,7 +58,8 @@ pub(crate) enum Batch {
 
 #[derive(Debug)]
 pub(crate) enum DocumentOperation {
-    Add(Uuid),
+    Replace(Uuid),
+    Update(Uuid),
     Delete(Vec<String>),
 }
 
@@ -64,7 +69,6 @@ pub(crate) enum IndexOperation {
     DocumentOperation {
         index_uid: String,
         primary_key: Option<String>,
-        method: IndexDocumentsMethod,
         operations: Vec<DocumentOperation>,
         tasks: Vec<Task>,
     },
@@ -103,6 +107,7 @@ impl Batch {
             Batch::TaskCancelation { task, .. }
             | Batch::Dump(task)
             | Batch::IndexCreation { task, .. }
+            | Batch::Export { task }
             | Batch::IndexUpdate { task, .. } => {
                 RoaringBitmap::from_sorted_iter(std::iter::once(task.uid)).unwrap()
             }
@@ -142,6 +147,7 @@ impl Batch {
             | TaskDeletions(_)
             | SnapshotCreation(_)
             | Dump(_)
+            | Export { .. }
             | UpgradeDatabase { .. }
             | IndexSwap { .. } => None,
             IndexOperation { op, .. } => Some(op.index_uid()),
@@ -167,6 +173,7 @@ impl fmt::Display for Batch {
             Batch::IndexUpdate { .. } => f.write_str("IndexUpdate")?,
             Batch::IndexDeletion { .. } => f.write_str("IndexDeletion")?,
             Batch::IndexSwap { .. } => f.write_str("IndexSwap")?,
+            Batch::Export { .. } => f.write_str("Export")?,
             Batch::UpgradeDatabase { .. } => f.write_str("UpgradeDatabase")?,
         };
         match index_uid {
@@ -254,7 +261,7 @@ impl IndexScheduler {
                     _ => unreachable!(),
                 }
             }
-            BatchKind::DocumentOperation { method, operation_ids, .. } => {
+            BatchKind::DocumentOperation { operation_ids, .. } => {
                 let tasks = self.queue.get_existing_tasks_for_processing_batch(
                     rtxn,
                     current_batch,
@@ -276,9 +283,17 @@ impl IndexScheduler {
 
                 for task in tasks.iter() {
                     match task.kind {
-                        KindWithContent::DocumentAdditionOrUpdate { content_file, .. } => {
-                            operations.push(DocumentOperation::Add(content_file));
+                        KindWithContent::DocumentAdditionOrUpdate {
+                            content_file, method, ..
+                        } => match method {
+                            IndexDocumentsMethod::ReplaceDocuments => {
+                                operations.push(DocumentOperation::Replace(content_file))
                             }
+                            IndexDocumentsMethod::UpdateDocuments => {
+                                operations.push(DocumentOperation::Update(content_file))
+                            }
+                            _ => unreachable!("Unknown document merging method"),
+                        },
                         KindWithContent::DocumentDeletion { ref documents_ids, .. } => {
                             operations.push(DocumentOperation::Delete(documents_ids.clone()));
                         }
@@ -290,7 +305,6 @@ impl IndexScheduler {
                     op: IndexOperation::DocumentOperation {
                         index_uid,
                         primary_key,
-                        method,
                         operations,
                         tasks,
                     },
@@ -416,11 +430,13 @@ impl IndexScheduler {
     }
 
     /// Create the next batch to be processed;
-    /// 1. We get the *last* task to cancel.
+    /// 0. We get the *last* task to cancel.
+    /// 1. We get the tasks to upgrade.
     /// 2. We get the *next* task to delete.
-    /// 3. We get the *next* snapshot to process.
-    /// 4. We get the *next* dump to process.
-    /// 5. We get the *next* tasks to process for a specific index.
+    /// 3. We get the *next* export to process.
+    /// 4. We get the *next* snapshot to process.
+    /// 5. We get the *next* dump to process.
+    /// 6. We get the *next* tasks to process for a specific index.
     #[tracing::instrument(level = "trace", skip(self, rtxn), target = "indexing::scheduler")]
     pub(crate) fn create_next_batch(
         &self,
@@ -433,9 +449,23 @@ impl IndexScheduler {
         let mut current_batch = ProcessingBatch::new(batch_id);
 
         let enqueued = &self.queue.tasks.get_status(rtxn, Status::Enqueued)?;
+        let count_total_enqueued = enqueued.len();
         let failed = &self.queue.tasks.get_status(rtxn, Status::Failed)?;
 
-        // 0. The priority over everything is to upgrade the instance
+        // 0. we get the last task to cancel.
+        let to_cancel = self.queue.tasks.get_kind(rtxn, Kind::TaskCancelation)? & enqueued;
+        if let Some(task_id) = to_cancel.max() {
+            let mut task =
+                self.queue.tasks.get_task(rtxn, task_id)?.ok_or(Error::CorruptedTaskQueue)?;
+            current_batch.processing(Some(&mut task));
+            current_batch.reason(BatchStopReason::TaskCannotBeBatched {
+                kind: Kind::TaskCancelation,
+                id: task_id,
+            });
+            return Ok(Some((Batch::TaskCancelation { task }, current_batch)));
+        }
+
+        // 1. We upgrade the instance
         // There shouldn't be multiple upgrade tasks but just in case we're going to batch all of them at the same time
         let upgrade = self.queue.tasks.get_kind(rtxn, Kind::UpgradeDatabase)? & (enqueued | failed);
         if !upgrade.is_empty() {
@@ -446,16 +476,26 @@ impl IndexScheduler {
                 current_batch.uid = batch_uid;
             }
             current_batch.processing(&mut tasks);
+            current_batch
+                .reason(BatchStopReason::TaskKindCannotBeBatched { kind: Kind::UpgradeDatabase });
             return Ok(Some((Batch::UpgradeDatabase { tasks }, current_batch)));
         }
 
-        // 1. we get the last task to cancel.
-        let to_cancel = self.queue.tasks.get_kind(rtxn, Kind::TaskCancelation)? & enqueued;
-        if let Some(task_id) = to_cancel.max() {
-            let mut task =
-                self.queue.tasks.get_task(rtxn, task_id)?.ok_or(Error::CorruptedTaskQueue)?;
-            current_batch.processing(Some(&mut task));
-            return Ok(Some((Batch::TaskCancelation { task }, current_batch)));
+        // check the version of the scheduler here.
+        // if the version is not the current, refuse to batch any additional task.
+        let version = self.version.get_version(rtxn)?;
+        let package_version = (
+            meilisearch_types::versioning::VERSION_MAJOR,
+            meilisearch_types::versioning::VERSION_MINOR,
+            meilisearch_types::versioning::VERSION_PATCH,
+        );
+        if version != Some(package_version) {
+            return Err(Error::UnrecoverableError(Box::new(
+                Error::IndexSchedulerVersionMismatch {
+                    index_scheduler_version: version.unwrap_or((1, 12, 0)),
+                    package_version,
+                },
+            )));
         }
 
         // 2. we get the next task to delete
@@ -463,27 +503,45 @@ impl IndexScheduler {
         if !to_delete.is_empty() {
             let mut tasks = self.queue.tasks.get_existing_tasks(rtxn, to_delete)?;
             current_batch.processing(&mut tasks);
+            current_batch
+                .reason(BatchStopReason::TaskKindCannotBeBatched { kind: Kind::TaskDeletion });
             return Ok(Some((Batch::TaskDeletions(tasks), current_batch)));
         }
 
-        // 3. we batch the snapshot.
+        // 3. we batch the export.
+        let to_export = self.queue.tasks.get_kind(rtxn, Kind::Export)? & enqueued;
+        if !to_export.is_empty() {
+            let task_id = to_export.iter().next().expect("There must be at least one export task");
+            let mut task = self.queue.tasks.get_task(rtxn, task_id)?.unwrap();
+            current_batch.processing([&mut task]);
+            current_batch.reason(BatchStopReason::TaskKindCannotBeBatched { kind: Kind::Export });
+            return Ok(Some((Batch::Export { task }, current_batch)));
+        }
+
+        // 4. we batch the snapshot.
         let to_snapshot = self.queue.tasks.get_kind(rtxn, Kind::SnapshotCreation)? & enqueued;
         if !to_snapshot.is_empty() {
             let mut tasks = self.queue.tasks.get_existing_tasks(rtxn, to_snapshot)?;
             current_batch.processing(&mut tasks);
+            current_batch
+                .reason(BatchStopReason::TaskKindCannotBeBatched { kind: Kind::SnapshotCreation });
             return Ok(Some((Batch::SnapshotCreation(tasks), current_batch)));
         }
 
-        // 4. we batch the dumps.
+        // 5. we batch the dumps.
         let to_dump = self.queue.tasks.get_kind(rtxn, Kind::DumpCreation)? & enqueued;
         if let Some(to_dump) = to_dump.min() {
             let mut task =
                 self.queue.tasks.get_task(rtxn, to_dump)?.ok_or(Error::CorruptedTaskQueue)?;
             current_batch.processing(Some(&mut task));
+            current_batch.reason(BatchStopReason::TaskCannotBeBatched {
+                kind: Kind::DumpCreation,
+                id: task.uid,
+            });
             return Ok(Some((Batch::Dump(task), current_batch)));
         }
 
-        // 5. We make a batch from the unprioritised tasks. Start by taking the next enqueued task.
+        // 6. We make a batch from the unprioritised tasks. Start by taking the next enqueued task.
         let task_id = if let Some(task_id) = enqueued.min() { task_id } else { return Ok(None) };
         let mut task =
             self.queue.tasks.get_task(rtxn, task_id)?.ok_or(Error::CorruptedTaskQueue)?;
@@ -497,6 +555,10 @@ impl IndexScheduler {
         } else {
             assert!(matches!(&task.kind, KindWithContent::IndexSwap { swaps } if swaps.is_empty()));
             current_batch.processing(Some(&mut task));
+            current_batch.reason(BatchStopReason::TaskCannotBeBatched {
+                kind: Kind::IndexSwap,
+                id: task.uid,
+            });
             return Ok(Some((Batch::IndexSwap { task }, current_batch)));
         };
 
@@ -518,9 +580,14 @@ impl IndexScheduler {
             1
         };
 
+        let mut stop_reason = BatchStopReason::default();
         let mut enqueued = Vec::new();
         let mut total_size: u64 = 0;
-        for task_id in index_tasks.into_iter().take(tasks_limit) {
+        for task_id in index_tasks.into_iter() {
+            if enqueued.len() >= tasks_limit {
+                stop_reason = BatchStopReason::ReachedTaskLimit { task_limit: tasks_limit };
+                break;
+            }
             let task = self
                 .queue
                 .tasks
@@ -528,20 +595,35 @@ impl IndexScheduler {
                 .and_then(|task| task.ok_or(Error::CorruptedTaskQueue))?;
 
             if let Some(uuid) = task.content_uuid() {
-                let content_size = self.queue.file_store.compute_size(uuid)?;
+                let content_size = match self.queue.file_store.compute_size(uuid) {
+                    Ok(content_size) => content_size,
+                    Err(file_store::Error::IoError(err)) if err.kind() == ErrorKind::NotFound => 0,
+                    Err(otherwise) => return Err(otherwise.into()),
+                };
                 total_size = total_size.saturating_add(content_size);
             }
 
-            if total_size > self.scheduler.batched_tasks_size_limit && !enqueued.is_empty() {
+            let size_limit = self.scheduler.batched_tasks_size_limit;
+            if total_size > size_limit && !enqueued.is_empty() {
+                stop_reason = BatchStopReason::ReachedSizeLimit { size_limit, size: total_size };
                 break;
             }
 
             enqueued.push((task.uid, task.kind));
         }
 
-        if let Some((batchkind, create_index)) =
+        stop_reason.replace_unspecified({
+            if enqueued.len() == count_total_enqueued as usize {
+                BatchStopReason::ExhaustedEnqueuedTasks
+            } else {
+                BatchStopReason::ExhaustedEnqueuedTasksForIndex { index: index_name.to_owned() }
+            }
+        });
+
+        if let Some((batchkind, create_index, autobatch_stop_reason)) =
             autobatcher::autobatch(enqueued, index_already_exists, primary_key.as_deref())
         {
+            current_batch.reason(autobatch_stop_reason.unwrap_or(stop_reason));
             return Ok(self
                 .create_next_batch_index(
                     rtxn,

crates/index-scheduler/src/scheduler/mod.rs

@@ -4,6 +4,7 @@ mod autobatcher_test;
 mod create_batch;
 mod process_batch;
 mod process_dump_creation;
+mod process_export;
 mod process_index_operation;
 mod process_snapshot_creation;
 mod process_upgrade;
@@ -20,9 +21,12 @@ use std::path::PathBuf;
 use std::sync::atomic::{AtomicBool, AtomicU32, Ordering};
 use std::sync::Arc;
 
+use convert_case::{Case, Casing as _};
 use meilisearch_types::error::ResponseError;
+use meilisearch_types::heed::{Env, WithoutTls};
 use meilisearch_types::milli;
 use meilisearch_types::tasks::Status;
+use process_batch::ProcessBatchInfo;
 use rayon::current_num_threads;
 use rayon::iter::{IntoParallelIterator, ParallelIterator};
 use roaring::RoaringBitmap;
@@ -71,10 +75,18 @@ pub struct Scheduler {
     pub(crate) snapshots_path: PathBuf,
 
     /// The path to the folder containing the auth LMDB env.
-    pub(crate) auth_path: PathBuf,
+    pub(crate) auth_env: Env<WithoutTls>,
 
     /// The path to the version file of Meilisearch.
     pub(crate) version_file_path: PathBuf,
+
+    /// The maximal number of entries in the search query cache of an embedder.
+    ///
+    /// 0 disables the cache.
+    pub(crate) embedding_cache_cap: usize,
+
+    /// Snapshot compaction status.
+    pub(crate) experimental_no_snapshot_compaction: bool,
 }
 
 impl Scheduler {
@@ -87,12 +99,14 @@ impl Scheduler {
             batched_tasks_size_limit: self.batched_tasks_size_limit,
             dumps_path: self.dumps_path.clone(),
             snapshots_path: self.snapshots_path.clone(),
-            auth_path: self.auth_path.clone(),
+            auth_env: self.auth_env.clone(),
             version_file_path: self.version_file_path.clone(),
+            embedding_cache_cap: self.embedding_cache_cap,
+            experimental_no_snapshot_compaction: self.experimental_no_snapshot_compaction,
         }
     }
 
-    pub fn new(options: &IndexSchedulerOptions) -> Scheduler {
+    pub fn new(options: &IndexSchedulerOptions, auth_env: Env<WithoutTls>) -> Scheduler {
         Scheduler {
             must_stop_processing: MustStopProcessing::default(),
             // we want to start the loop right away in case meilisearch was ctrl+Ced while processing things
@@ -102,8 +116,10 @@ impl Scheduler {
             batched_tasks_size_limit: options.batched_tasks_size_limit,
             dumps_path: options.dumps_path.clone(),
             snapshots_path: options.snapshots_path.clone(),
-            auth_path: options.auth_path.clone(),
+            auth_env,
             version_file_path: options.version_file_path.clone(),
+            embedding_cache_cap: options.embedding_cache_cap,
+            experimental_no_snapshot_compaction: options.experimental_no_snapshot_compaction,
         }
     }
 }
@@ -215,14 +231,16 @@ impl IndexScheduler {
         let mut stop_scheduler_forever = false;
         let mut wtxn = self.env.write_txn().map_err(Error::HeedTransaction)?;
         let mut canceled = RoaringBitmap::new();
+        let mut process_batch_info = ProcessBatchInfo::default();
 
         match res {
-            Ok(tasks) => {
+            Ok((tasks, info)) => {
                 #[cfg(test)]
                 self.breakpoint(crate::test_utils::Breakpoint::ProcessBatchSucceeded);
 
                 let (task_progress, task_progress_obj) = AtomicTaskStep::new(tasks.len() as u32);
                 progress.update_progress(task_progress_obj);
+                process_batch_info = info;
                 let mut success = 0;
                 let mut failure = 0;
                 let mut canceled_by = None;
@@ -339,6 +357,60 @@ impl IndexScheduler {
 
         // We must re-add the canceled task so they're part of the same batch.
         ids |= canceled;
+
+        let ProcessBatchInfo { congestion, pre_commit_dabases_sizes, post_commit_dabases_sizes } =
+            process_batch_info;
+
+        processing_batch.stats.progress_trace =
+            progress.accumulated_durations().into_iter().map(|(k, v)| (k, v.into())).collect();
+        processing_batch.stats.write_channel_congestion = congestion.map(|congestion| {
+            let mut congestion_info = serde_json::Map::new();
+            congestion_info.insert("attempts".into(), congestion.attempts.into());
+            congestion_info.insert("blocking_attempts".into(), congestion.blocking_attempts.into());
+            congestion_info.insert("blocking_ratio".into(), congestion.congestion_ratio().into());
+            congestion_info
+        });
+        processing_batch.stats.internal_database_sizes = pre_commit_dabases_sizes
+            .iter()
+            .flat_map(|(dbname, pre_size)| {
+                post_commit_dabases_sizes
+                    .get(dbname)
+                    .map(|post_size| {
+                        use std::cmp::Ordering::{Equal, Greater, Less};
+
+                        use byte_unit::Byte;
+                        use byte_unit::UnitType::Binary;
+
+                        let post = Byte::from_u64(*post_size as u64).get_appropriate_unit(Binary);
+                        let diff_size = post_size.abs_diff(*pre_size) as u64;
+                        let diff = Byte::from_u64(diff_size).get_appropriate_unit(Binary);
+                        let sign = match post_size.cmp(pre_size) {
+                            Equal => return None,
+                            Greater => "+",
+                            Less => "-",
+                        };
+
+                        Some((
+                            dbname.to_case(Case::Camel),
+                            format!("{post:#.2} ({sign}{diff:#.2})").into(),
+                        ))
+                    })
+                    .into_iter()
+                    .flatten()
+            })
+            .collect();
+
+        if let Some(congestion) = congestion {
+            tracing::debug!(
+                "Channel congestion metrics - Attempts: {}, Blocked attempts: {}  ({:.1}% congestion)",
+                congestion.attempts,
+                congestion.blocking_attempts,
+                congestion.congestion_ratio(),
+            );
+        }
+
+        tracing::debug!("call trace: {:?}", progress.accumulated_durations());
+
         self.queue.write_batch(&mut wtxn, processing_batch, &ids)?;
 
         #[cfg(test)]

crates/index-scheduler/src/scheduler/process_batch.rs

@@ -5,14 +5,15 @@ use std::sync::atomic::Ordering;
 use meilisearch_types::batches::{BatchEnqueuedAt, BatchId};
 use meilisearch_types::heed::{RoTxn, RwTxn};
 use meilisearch_types::milli::progress::{Progress, VariableNameStep};
-use meilisearch_types::milli::{self};
-use meilisearch_types::tasks::{Details, IndexSwap, KindWithContent, Status, Task};
+use meilisearch_types::milli::{self, ChannelCongestion};
+use meilisearch_types::tasks::{Details, IndexSwap, Kind, KindWithContent, Status, Task};
+use meilisearch_types::versioning::{VERSION_MAJOR, VERSION_MINOR, VERSION_PATCH};
 use milli::update::Settings as MilliSettings;
 use roaring::RoaringBitmap;
 
 use super::create_batch::Batch;
 use crate::processing::{
-    AtomicBatchStep, AtomicTaskStep, CreateIndexProgress, DeleteIndexProgress,
+    AtomicBatchStep, AtomicTaskStep, CreateIndexProgress, DeleteIndexProgress, FinalizingIndexStep,
     InnerSwappingTwoIndexes, SwappingTheIndexes, TaskCancelationProgress, TaskDeletionProgress,
     UpdateIndexProgress,
 };
@@ -22,6 +23,16 @@ use crate::utils::{
 };
 use crate::{Error, IndexScheduler, Result, TaskId};
 
+#[derive(Debug, Default)]
+pub struct ProcessBatchInfo {
+    /// The write channel congestion. None when unavailable: settings update.
+    pub congestion: Option<ChannelCongestion>,
+    /// The sizes of the different databases before starting the indexation.
+    pub pre_commit_dabases_sizes: indexmap::IndexMap<&'static str, usize>,
+    /// The sizes of the different databases after commiting the indexation.
+    pub post_commit_dabases_sizes: indexmap::IndexMap<&'static str, usize>,
+}
+
 impl IndexScheduler {
     /// Apply the operation associated with the given batch.
     ///
@@ -35,7 +46,7 @@ impl IndexScheduler {
         batch: Batch,
         current_batch: &mut ProcessingBatch,
         progress: Progress,
-    ) -> Result<Vec<Task>> {
+    ) -> Result<(Vec<Task>, ProcessBatchInfo)> {
         #[cfg(test)]
         {
             self.maybe_fail(crate::test_utils::FailureLocation::InsideProcessBatch)?;
@@ -76,7 +87,7 @@ impl IndexScheduler {
 
                 canceled_tasks.push(task);
 
-                Ok(canceled_tasks)
+                Ok((canceled_tasks, ProcessBatchInfo::default()))
             }
             Batch::TaskDeletions(mut tasks) => {
                 // 1. Retrieve the tasks that matched the query at enqueue-time.
@@ -115,10 +126,14 @@ impl IndexScheduler {
                         _ => unreachable!(),
                     }
                 }
-                Ok(tasks)
+                Ok((tasks, ProcessBatchInfo::default()))
             }
-            Batch::SnapshotCreation(tasks) => self.process_snapshot(progress, tasks),
-            Batch::Dump(task) => self.process_dump_creation(progress, task),
+            Batch::SnapshotCreation(tasks) => self
+                .process_snapshot(progress, tasks)
+                .map(|tasks| (tasks, ProcessBatchInfo::default())),
+            Batch::Dump(task) => self
+                .process_dump_creation(progress, task)
+                .map(|tasks| (tasks, ProcessBatchInfo::default())),
             Batch::IndexOperation { op, must_create_index } => {
                 let index_uid = op.index_uid().to_string();
                 let index = if must_create_index {
@@ -130,14 +145,33 @@ impl IndexScheduler {
                     self.index_mapper.index(&rtxn, &index_uid)?
                 };
 
+                let mut index_wtxn = index.write_txn()?;
+
+                let index_version = index.get_version(&index_wtxn)?.unwrap_or((1, 12, 0));
+                let package_version = (VERSION_MAJOR, VERSION_MINOR, VERSION_PATCH);
+                if index_version != package_version {
+                    return Err(Error::IndexVersionMismatch {
+                        index: index_uid,
+                        index_version,
+                        package_version,
+                    });
+                }
+
                 // the index operation can take a long time, so save this handle to make it available to the search for the duration of the tick
                 self.index_mapper
                     .set_currently_updating_index(Some((index_uid.clone(), index.clone())));
 
-                let mut index_wtxn = index.write_txn()?;
-                let tasks = self.apply_index_operation(&mut index_wtxn, &index, op, progress)?;
+                let pre_commit_dabases_sizes = index.database_sizes(&index_wtxn)?;
+                let (tasks, congestion) = self.apply_index_operation(
+                    &mut index_wtxn,
+                    &index,
+                    op,
+                    &progress,
+                    current_batch.embedder_stats.clone(),
+                )?;
 
                 {
+                    progress.update_progress(FinalizingIndexStep::Committing);
                     let span = tracing::trace_span!(target: "indexing::scheduler", "commit");
                     let _entered = span.enter();
 
@@ -148,12 +182,15 @@ impl IndexScheduler {
                 // stats of the index. Since the tasks have already been processed and
                 // this is a non-critical operation. If it fails, we should not fail
                 // the entire batch.
+                let mut post_commit_dabases_sizes = None;
                 let res = || -> Result<()> {
+                    progress.update_progress(FinalizingIndexStep::ComputingStats);
                     let index_rtxn = index.read_txn()?;
                     let stats = crate::index_mapper::IndexStats::new(&index, &index_rtxn)
                         .map_err(|e| Error::from_milli(e, Some(index_uid.to_string())))?;
                     let mut wtxn = self.env.write_txn()?;
                     self.index_mapper.store_stats_of(&mut wtxn, &index_uid, &stats)?;
+                    post_commit_dabases_sizes = Some(index.database_sizes(&index_rtxn)?);
                     wtxn.commit()?;
                     Ok(())
                 }();
@@ -166,7 +203,16 @@ impl IndexScheduler {
                     ),
                 }
 
-                Ok(tasks)
+                let info = ProcessBatchInfo {
+                    congestion,
+                    // In case we fail to the get post-commit sizes we decide
+                    // that nothing changed and use the pre-commit sizes.
+                    post_commit_dabases_sizes: post_commit_dabases_sizes
+                        .unwrap_or_else(|| pre_commit_dabases_sizes.clone()),
+                    pre_commit_dabases_sizes,
+                };
+
+                Ok((tasks, info))
             }
             Batch::IndexCreation { index_uid, primary_key, task } => {
                 progress.update_progress(CreateIndexProgress::CreatingTheIndex);
@@ -197,10 +243,12 @@ impl IndexScheduler {
                     );
                     builder.set_primary_key(primary_key);
                     let must_stop_processing = self.scheduler.must_stop_processing.clone();
+
                     builder
                         .execute(
-                            |indexing_step| tracing::debug!(update = ?indexing_step),
-                            || must_stop_processing.get(),
+                            &|| must_stop_processing.get(),
+                            &progress,
+                            current_batch.embedder_stats.clone(),
                         )
                         .map_err(|e| Error::from_milli(e, Some(index_uid.to_string())))?;
                     index_wtxn.commit()?;
@@ -234,7 +282,7 @@ impl IndexScheduler {
                     ),
                 }
 
-                Ok(vec![task])
+                Ok((vec![task], ProcessBatchInfo::default()))
             }
             Batch::IndexDeletion { index_uid, index_has_been_created, mut tasks } => {
                 progress.update_progress(DeleteIndexProgress::DeletingTheIndex);
@@ -268,7 +316,9 @@ impl IndexScheduler {
                     };
                 }
 
-                Ok(tasks)
+                // Here we could also show that all the internal database sizes goes to 0
+                // but it would mean opening the index and that's costly.
+                Ok((tasks, ProcessBatchInfo::default()))
             }
             Batch::IndexSwap { mut task } => {
                 progress.update_progress(SwappingTheIndexes::EnsuringCorrectnessOfTheSwap);
@@ -316,15 +366,57 @@ impl IndexScheduler {
                 }
                 wtxn.commit()?;
                 task.status = Status::Succeeded;
-                Ok(vec![task])
+                Ok((vec![task], ProcessBatchInfo::default()))
+            }
+            Batch::Export { mut task } => {
+                let KindWithContent::Export { url, api_key, payload_size, indexes } = &task.kind
+                else {
+                    unreachable!()
+                };
+
+                let ret = catch_unwind(AssertUnwindSafe(|| {
+                    self.process_export(
+                        url,
+                        api_key.as_deref(),
+                        payload_size.as_ref(),
+                        indexes,
+                        progress,
+                    )
+                }));
+
+                let stats = match ret {
+                    Ok(Ok(stats)) => stats,
+                    Ok(Err(Error::AbortedTask)) => return Err(Error::AbortedTask),
+                    Ok(Err(e)) => return Err(Error::Export(Box::new(e))),
+                    Err(e) => {
+                        let msg = match e.downcast_ref::<&'static str>() {
+                            Some(s) => *s,
+                            None => match e.downcast_ref::<String>() {
+                                Some(s) => &s[..],
+                                None => "Box<dyn Any>",
+                            },
+                        };
+                        return Err(Error::Export(Box::new(Error::ProcessBatchPanicked(
+                            msg.to_string(),
+                        ))));
+                    }
+                };
+
+                task.status = Status::Succeeded;
+                if let Some(Details::Export { indexes, .. }) = task.details.as_mut() {
+                    *indexes = stats;
+                }
+                Ok((vec![task], ProcessBatchInfo::default()))
             }
             Batch::UpgradeDatabase { mut tasks } => {
                 let KindWithContent::UpgradeDatabase { from } = tasks.last().unwrap().kind else {
                     unreachable!();
                 };
+
                 let ret = catch_unwind(AssertUnwindSafe(|| self.process_upgrade(from, progress)));
                 match ret {
                     Ok(Ok(())) => (),
+                    Ok(Err(Error::AbortedTask)) => return Err(Error::AbortedTask),
                     Ok(Err(e)) => return Err(Error::DatabaseUpgrade(Box::new(e))),
                     Err(e) => {
                         let msg = match e.downcast_ref::<&'static str>() {
@@ -346,7 +438,7 @@ impl IndexScheduler {
                     task.error = None;
                 }
 
-                Ok(tasks)
+                Ok((tasks, ProcessBatchInfo::default()))
             }
         }
     }
@@ -622,17 +714,81 @@ impl IndexScheduler {
         progress: &Progress,
     ) -> Result<Vec<Task>> {
         progress.update_progress(TaskCancelationProgress::RetrievingTasks);
+        let mut tasks_to_cancel = RoaringBitmap::new();
 
+        let enqueued_tasks = &self.queue.tasks.get_status(rtxn, Status::Enqueued)?;
+
+        // 0. Check if any upgrade task was matched.
+        //    If so, we cancel all the failed or enqueued upgrade tasks.
+        let upgrade_tasks = &self.queue.tasks.get_kind(rtxn, Kind::UpgradeDatabase)?;
+        let is_canceling_upgrade = !matched_tasks.is_disjoint(upgrade_tasks);
+        if is_canceling_upgrade {
+            let failed_tasks = self.queue.tasks.get_status(rtxn, Status::Failed)?;
+            tasks_to_cancel |= upgrade_tasks & (enqueued_tasks | failed_tasks);
+        }
         // 1. Remove from this list the tasks that we are not allowed to cancel
         //    Notice that only the _enqueued_ ones are cancelable and we should
         //    have already aborted the indexation of the _processing_ ones
-        let cancelable_tasks = self.queue.tasks.get_status(rtxn, Status::Enqueued)?;
-        let tasks_to_cancel = cancelable_tasks & matched_tasks;
+        tasks_to_cancel |= enqueued_tasks & matched_tasks;
 
+        // 2. If we're canceling an upgrade, attempt the rollback
+        if let Some(latest_upgrade_task) = (&tasks_to_cancel & upgrade_tasks).max() {
+            progress.update_progress(TaskCancelationProgress::CancelingUpgrade);
+
+            let task = self.queue.tasks.get_task(rtxn, latest_upgrade_task)?.unwrap();
+            let Some(Details::UpgradeDatabase { from, to }) = task.details else {
+                unreachable!("wrong details for upgrade task {latest_upgrade_task}")
+            };
+
+            // check that we are rollbacking an upgrade to the current Meilisearch
+            let bin_major: u32 = meilisearch_types::versioning::VERSION_MAJOR;
+            let bin_minor: u32 = meilisearch_types::versioning::VERSION_MINOR;
+            let bin_patch: u32 = meilisearch_types::versioning::VERSION_PATCH;
+
+            if to == (bin_major, bin_minor, bin_patch) {
+                tracing::warn!(
+                    "Rollbacking from v{}.{}.{} to v{}.{}.{}",
+                    to.0,
+                    to.1,
+                    to.2,
+                    from.0,
+                    from.1,
+                    from.2
+                );
+                let ret = catch_unwind(std::panic::AssertUnwindSafe(|| {
+                    self.process_rollback(from, progress)
+                }));
+
+                match ret {
+                    Ok(Ok(())) => {}
+                    Ok(Err(err)) => return Err(Error::DatabaseUpgrade(Box::new(err))),
+                    Err(e) => {
+                        let msg = match e.downcast_ref::<&'static str>() {
+                            Some(s) => *s,
+                            None => match e.downcast_ref::<String>() {
+                                Some(s) => &s[..],
+                                None => "Box<dyn Any>",
+                            },
+                        };
+                        return Err(Error::DatabaseUpgrade(Box::new(Error::ProcessBatchPanicked(
+                            msg.to_string(),
+                        ))));
+                    }
+                }
+            } else {
+                tracing::debug!(
+                    "Not rollbacking an upgrade targetting the earlier version v{}.{}.{}",
+                    bin_major,
+                    bin_minor,
+                    bin_patch
+                )
+            }
+        }
+
+        // 3. We now have a list of tasks to cancel, cancel them
         let (task_progress, progress_obj) = AtomicTaskStep::new(tasks_to_cancel.len() as u32);
         progress.update_progress(progress_obj);
 
-        // 2. We now have a list of tasks to cancel, cancel them
         let mut tasks = self.queue.tasks.get_existing_tasks(
             rtxn,
             tasks_to_cancel.iter().inspect(|_| {

crates/index-scheduler/src/scheduler/process_dump_creation.rs

@@ -43,7 +43,16 @@ impl IndexScheduler {
 
         let rtxn = self.env.read_txn()?;
 
-        // 2. dump the tasks
+        // 2. dump the chat completion settings
+        // TODO should I skip the export if the chat completion has been disabled?
+        progress.update_progress(DumpCreationProgress::DumpTheChatCompletionSettings);
+        let mut dump_chat_completion_settings = dump.create_chat_completions_settings()?;
+        for result in self.chat_settings.iter(&rtxn)? {
+            let (name, chat_settings) = result?;
+            dump_chat_completion_settings.push_settings(name, &chat_settings)?;
+        }
+
+        // 3. dump the tasks
         progress.update_progress(DumpCreationProgress::DumpTheTasks);
         let mut dump_tasks = dump.create_tasks_queue()?;
 
@@ -81,7 +90,7 @@ impl IndexScheduler {
 
             let mut dump_content_file = dump_tasks.push_task(&t.into())?;
 
-            // 2.1. Dump the `content_file` associated with the task if there is one and the task is not finished yet.
+            // 3.1. Dump the `content_file` associated with the task if there is one and the task is not finished yet.
             if let Some(content_file) = content_file {
                 if self.scheduler.must_stop_processing.get() {
                     return Err(Error::AbortedTask);
@@ -105,7 +114,7 @@ impl IndexScheduler {
         }
         dump_tasks.flush()?;
 
-        // 3. dump the batches
+        // 4. dump the batches
         progress.update_progress(DumpCreationProgress::DumpTheBatches);
         let mut dump_batches = dump.create_batches_queue()?;
 
@@ -138,7 +147,7 @@ impl IndexScheduler {
         }
         dump_batches.flush()?;
 
-        // 4. Dump the indexes
+        // 5. Dump the indexes
         progress.update_progress(DumpCreationProgress::DumpTheIndexes);
         let nb_indexes = self.index_mapper.index_mapping.len(&rtxn)? as u32;
         let mut count = 0;
@@ -165,9 +174,6 @@ impl IndexScheduler {
 
             let fields_ids_map = index.fields_ids_map(&rtxn)?;
             let all_fields: Vec<_> = fields_ids_map.iter().map(|(id, _)| id).collect();
-            let embedding_configs = index
-                .embedding_configs(&rtxn)
-                .map_err(|e| Error::from_milli(e, Some(uid.to_string())))?;
 
             let nb_documents = index
                 .number_of_documents(&rtxn)
@@ -178,7 +184,7 @@ impl IndexScheduler {
             let documents = index
                 .all_documents(&rtxn)
                 .map_err(|e| Error::from_milli(e, Some(uid.to_string())))?;
-            // 4.1. Dump the documents
+            // 5.1. Dump the documents
             for ret in documents {
                 if self.scheduler.must_stop_processing.get() {
                     return Err(Error::AbortedTask);
@@ -221,16 +227,12 @@ impl IndexScheduler {
                         return Err(Error::from_milli(user_err, Some(uid.to_string())));
                     };
 
-                    for (embedder_name, embeddings) in embeddings {
-                        let user_provided = embedding_configs
-                            .iter()
-                            .find(|conf| conf.name == embedder_name)
-                            .is_some_and(|conf| conf.user_provided.contains(id));
+                    for (embedder_name, (embeddings, regenerate)) in embeddings {
                         let embeddings = ExplicitVectors {
                             embeddings: Some(VectorOrArrayOfVectors::from_array_of_vectors(
                                 embeddings,
                             )),
-                            regenerate: !user_provided,
+                            regenerate,
                         };
                         vectors.insert(embedder_name, serde_json::to_value(embeddings).unwrap());
                     }
@@ -240,7 +242,7 @@ impl IndexScheduler {
                 atomic.fetch_add(1, Ordering::Relaxed);
             }
 
-            // 4.2. Dump the settings
+            // 5.2. Dump the settings
             let settings = meilisearch_types::settings::settings(
                 index,
                 &rtxn,
@@ -251,7 +253,7 @@ impl IndexScheduler {
             Ok(())
         })?;
 
-        // 5. Dump experimental feature settings
+        // 6. Dump experimental feature settings
         progress.update_progress(DumpCreationProgress::DumpTheExperimentalFeatures);
         let features = self.features().runtime_features();
         dump.create_experimental_features(features)?;

crates/index-scheduler/src/scheduler/process_export.rs

@@ -0,0 +1,365 @@
+use std::collections::BTreeMap;
+use std::io::{self, Write as _};
+use std::sync::atomic;
+use std::time::Duration;
+
+use backoff::ExponentialBackoff;
+use byte_unit::Byte;
+use flate2::write::GzEncoder;
+use flate2::Compression;
+use meilisearch_types::index_uid_pattern::IndexUidPattern;
+use meilisearch_types::milli::constants::RESERVED_VECTORS_FIELD_NAME;
+use meilisearch_types::milli::progress::{Progress, VariableNameStep};
+use meilisearch_types::milli::update::{request_threads, Setting};
+use meilisearch_types::milli::vector::parsed_vectors::{ExplicitVectors, VectorOrArrayOfVectors};
+use meilisearch_types::milli::{self, obkv_to_json, Filter, InternalError};
+use meilisearch_types::settings::{self, SecretPolicy};
+use meilisearch_types::tasks::{DetailsExportIndexSettings, ExportIndexSettings};
+use serde::Deserialize;
+use ureq::{json, Response};
+
+use super::MustStopProcessing;
+use crate::processing::AtomicDocumentStep;
+use crate::{Error, IndexScheduler, Result};
+
+impl IndexScheduler {
+    pub(super) fn process_export(
+        &self,
+        base_url: &str,
+        api_key: Option<&str>,
+        payload_size: Option<&Byte>,
+        indexes: &BTreeMap<IndexUidPattern, ExportIndexSettings>,
+        progress: Progress,
+    ) -> Result<BTreeMap<IndexUidPattern, DetailsExportIndexSettings>> {
+        #[cfg(test)]
+        self.maybe_fail(crate::test_utils::FailureLocation::ProcessExport)?;
+
+        let indexes: Vec<_> = self
+            .index_names()?
+            .into_iter()
+            .flat_map(|uid| {
+                indexes
+                    .iter()
+                    .find(|(pattern, _)| pattern.matches_str(&uid))
+                    .map(|(pattern, settings)| (pattern, uid, settings))
+            })
+            .collect();
+
+        let mut output = BTreeMap::new();
+        let agent = ureq::AgentBuilder::new().timeout(Duration::from_secs(5)).build();
+        let must_stop_processing = self.scheduler.must_stop_processing.clone();
+        for (i, (_pattern, uid, export_settings)) in indexes.iter().enumerate() {
+            if must_stop_processing.get() {
+                return Err(Error::AbortedTask);
+            }
+
+            progress.update_progress(VariableNameStep::<ExportIndex>::new(
+                format!("Exporting index `{uid}`"),
+                i as u32,
+                indexes.len() as u32,
+            ));
+
+            let ExportIndexSettings { filter, override_settings } = export_settings;
+            let index = self.index(uid)?;
+            let index_rtxn = index.read_txn()?;
+
+            // First, check if the index already exists
+            let url = format!("{base_url}/indexes/{uid}");
+            let response = retry(&must_stop_processing, || {
+                let mut request = agent.get(&url);
+                if let Some(api_key) = api_key {
+                    request = request.set("Authorization", &format!("Bearer {api_key}"));
+                }
+
+                request.send_bytes(Default::default()).map_err(into_backoff_error)
+            });
+            let index_exists = match response {
+                Ok(response) => response.status() == 200,
+                Err(Error::FromRemoteWhenExporting { code, .. }) if code == "index_not_found" => {
+                    false
+                }
+                Err(e) => return Err(e),
+            };
+
+            let primary_key = index
+                .primary_key(&index_rtxn)
+                .map_err(|e| Error::from_milli(e.into(), Some(uid.to_string())))?;
+
+            // Create the index
+            if !index_exists {
+                let url = format!("{base_url}/indexes");
+                retry(&must_stop_processing, || {
+                    let mut request = agent.post(&url);
+                    if let Some(api_key) = api_key {
+                        request = request.set("Authorization", &format!("Bearer {api_key}"));
+                    }
+                    let index_param = json!({ "uid": uid, "primaryKey": primary_key });
+                    request.send_json(&index_param).map_err(into_backoff_error)
+                })?;
+            }
+
+            // Patch the index primary key
+            if index_exists && *override_settings {
+                let url = format!("{base_url}/indexes/{uid}");
+                retry(&must_stop_processing, || {
+                    let mut request = agent.patch(&url);
+                    if let Some(api_key) = api_key {
+                        request = request.set("Authorization", &format!("Bearer {api_key}"));
+                    }
+                    let index_param = json!({ "primaryKey": primary_key });
+                    request.send_json(&index_param).map_err(into_backoff_error)
+                })?;
+            }
+
+            // Send the index settings
+            if !index_exists || *override_settings {
+                let mut settings =
+                    settings::settings(&index, &index_rtxn, SecretPolicy::RevealSecrets)
+                        .map_err(|e| Error::from_milli(e, Some(uid.to_string())))?;
+                // Remove the experimental chat setting if not enabled
+                if self.features().check_chat_completions("exporting chat settings").is_err() {
+                    settings.chat = Setting::NotSet;
+                }
+                // Retry logic for sending settings
+                let url = format!("{base_url}/indexes/{uid}/settings");
+                let bearer = api_key.map(|api_key| format!("Bearer {api_key}"));
+                retry(&must_stop_processing, || {
+                    let mut request = agent.patch(&url);
+                    if let Some(bearer) = bearer.as_ref() {
+                        request = request.set("Authorization", bearer);
+                    }
+                    request.send_json(settings.clone()).map_err(into_backoff_error)
+                })?;
+            }
+
+            let filter = filter
+                .as_ref()
+                .map(Filter::from_json)
+                .transpose()
+                .map_err(|e| Error::from_milli(e, Some(uid.to_string())))?
+                .flatten();
+
+            let filter_universe = filter
+                .map(|f| f.evaluate(&index_rtxn, &index))
+                .transpose()
+                .map_err(|e| Error::from_milli(e, Some(uid.to_string())))?;
+            let whole_universe = index
+                .documents_ids(&index_rtxn)
+                .map_err(|e| Error::from_milli(e.into(), Some(uid.to_string())))?;
+            let universe = filter_universe.unwrap_or(whole_universe);
+
+            let fields_ids_map = index.fields_ids_map(&index_rtxn)?;
+            let all_fields: Vec<_> = fields_ids_map.iter().map(|(id, _)| id).collect();
+
+            // We don't need to keep this one alive as we will
+            // spawn many threads to process the documents
+            drop(index_rtxn);
+
+            let total_documents = universe.len() as u32;
+            let (step, progress_step) = AtomicDocumentStep::new(total_documents);
+            progress.update_progress(progress_step);
+
+            output.insert(
+                IndexUidPattern::new_unchecked(uid.clone()),
+                DetailsExportIndexSettings {
+                    settings: (*export_settings).clone(),
+                    matched_documents: Some(total_documents as u64),
+                },
+            );
+
+            let limit = payload_size.map(|ps| ps.as_u64() as usize).unwrap_or(50 * 1024 * 1024); // defaults to 50 MiB
+            let documents_url = format!("{base_url}/indexes/{uid}/documents");
+
+            request_threads()
+                .broadcast(|ctx| {
+                    let index_rtxn = index
+                        .read_txn()
+                        .map_err(|e| Error::from_milli(e.into(), Some(uid.to_string())))?;
+
+                    let mut buffer = Vec::new();
+                    let mut tmp_buffer = Vec::new();
+                    let mut compressed_buffer = Vec::new();
+                    for (i, docid) in universe.iter().enumerate() {
+                        if i % ctx.num_threads() != ctx.index() {
+                            continue;
+                        }
+
+                        let document = index
+                            .document(&index_rtxn, docid)
+                            .map_err(|e| Error::from_milli(e, Some(uid.to_string())))?;
+
+                        let mut document = obkv_to_json(&all_fields, &fields_ids_map, document)
+                            .map_err(|e| Error::from_milli(e, Some(uid.to_string())))?;
+
+                        // TODO definitely factorize this code
+                        'inject_vectors: {
+                            let embeddings = index
+                                .embeddings(&index_rtxn, docid)
+                                .map_err(|e| Error::from_milli(e, Some(uid.to_string())))?;
+
+                            if embeddings.is_empty() {
+                                break 'inject_vectors;
+                            }
+
+                            let vectors = document
+                                .entry(RESERVED_VECTORS_FIELD_NAME)
+                                .or_insert(serde_json::Value::Object(Default::default()));
+
+                            let serde_json::Value::Object(vectors) = vectors else {
+                                return Err(Error::from_milli(
+                                    milli::Error::UserError(
+                                        milli::UserError::InvalidVectorsMapType {
+                                            document_id: {
+                                                if let Ok(Some(Ok(index))) = index
+                                                    .external_id_of(
+                                                        &index_rtxn,
+                                                        std::iter::once(docid),
+                                                    )
+                                                    .map(|it| it.into_iter().next())
+                                                {
+                                                    index
+                                                } else {
+                                                    format!("internal docid={docid}")
+                                                }
+                                            },
+                                            value: vectors.clone(),
+                                        },
+                                    ),
+                                    Some(uid.to_string()),
+                                ));
+                            };
+
+                            for (embedder_name, (embeddings, regenerate)) in embeddings {
+                                let embeddings = ExplicitVectors {
+                                    embeddings: Some(
+                                        VectorOrArrayOfVectors::from_array_of_vectors(embeddings),
+                                    ),
+                                    regenerate,
+                                };
+                                vectors.insert(
+                                    embedder_name,
+                                    serde_json::to_value(embeddings).unwrap(),
+                                );
+                            }
+                        }
+
+                        tmp_buffer.clear();
+                        serde_json::to_writer(&mut tmp_buffer, &document)
+                            .map_err(milli::InternalError::from)
+                            .map_err(|e| Error::from_milli(e.into(), Some(uid.to_string())))?;
+
+                        // Make sure we put at least one document in the buffer even
+                        // though we might go above the buffer limit before sending
+                        if !buffer.is_empty() && buffer.len() + tmp_buffer.len() > limit {
+                            // We compress the documents before sending them
+                            let mut encoder =
+                                GzEncoder::new(&mut compressed_buffer, Compression::default());
+                            encoder
+                                .write_all(&buffer)
+                                .map_err(|e| Error::from_milli(e.into(), Some(uid.clone())))?;
+                            encoder
+                                .finish()
+                                .map_err(|e| Error::from_milli(e.into(), Some(uid.clone())))?;
+
+                            retry(&must_stop_processing, || {
+                                let mut request = agent.post(&documents_url);
+                                request = request.set("Content-Type", "application/x-ndjson");
+                                request = request.set("Content-Encoding", "gzip");
+                                if let Some(api_key) = api_key {
+                                    request = request
+                                        .set("Authorization", &(format!("Bearer {api_key}")));
+                                }
+                                request.send_bytes(&compressed_buffer).map_err(into_backoff_error)
+                            })?;
+                            buffer.clear();
+                            compressed_buffer.clear();
+                        }
+                        buffer.extend_from_slice(&tmp_buffer);
+
+                        if i % 100 == 0 {
+                            step.fetch_add(100, atomic::Ordering::Relaxed);
+                        }
+                    }
+
+                    retry(&must_stop_processing, || {
+                        let mut request = agent.post(&documents_url);
+                        request = request.set("Content-Type", "application/x-ndjson");
+                        if let Some(api_key) = api_key {
+                            request = request.set("Authorization", &(format!("Bearer {api_key}")));
+                        }
+                        request.send_bytes(&buffer).map_err(into_backoff_error)
+                    })?;
+
+                    Ok(())
+                })
+                .map_err(|e| {
+                    Error::from_milli(
+                        milli::Error::InternalError(InternalError::PanicInThreadPool(e)),
+                        Some(uid.to_string()),
+                    )
+                })?;
+
+            step.store(total_documents, atomic::Ordering::Relaxed);
+        }
+
+        Ok(output)
+    }
+}
+
+fn retry<F>(must_stop_processing: &MustStopProcessing, send_request: F) -> Result<ureq::Response>
+where
+    F: Fn() -> Result<ureq::Response, backoff::Error<ureq::Error>>,
+{
+    match backoff::retry(ExponentialBackoff::default(), || {
+        if must_stop_processing.get() {
+            return Err(backoff::Error::Permanent(ureq::Error::Status(
+                u16::MAX,
+                // 444: Connection Closed Without Response
+                Response::new(444, "Abort", "Aborted task").unwrap(),
+            )));
+        }
+        send_request()
+    }) {
+        Ok(response) => Ok(response),
+        Err(backoff::Error::Permanent(e)) => Err(ureq_error_into_error(e)),
+        Err(backoff::Error::Transient { err, retry_after: _ }) => Err(ureq_error_into_error(err)),
+    }
+}
+
+fn into_backoff_error(err: ureq::Error) -> backoff::Error<ureq::Error> {
+    match err {
+        // Those code status must trigger an automatic retry
+        // <https://www.restapitutorial.com/advanced/responses/retries>
+        ureq::Error::Status(408 | 429 | 500 | 502 | 503 | 504, _) => {
+            backoff::Error::Transient { err, retry_after: None }
+        }
+        ureq::Error::Status(_, _) => backoff::Error::Permanent(err),
+        ureq::Error::Transport(_) => backoff::Error::Transient { err, retry_after: None },
+    }
+}
+
+/// Converts a `ureq::Error` into an `Error`.
+fn ureq_error_into_error(error: ureq::Error) -> Error {
+    #[derive(Deserialize)]
+    struct MeiliError {
+        message: String,
+        code: String,
+        r#type: String,
+        link: String,
+    }
+
+    match error {
+        // This is a workaround to handle task abortion - the error propagation path
+        // makes it difficult to cleanly surface the abortion at this level.
+        ureq::Error::Status(u16::MAX, _) => Error::AbortedTask,
+        ureq::Error::Status(_, response) => match response.into_json() {
+            Ok(MeiliError { message, code, r#type, link }) => {
+                Error::FromRemoteWhenExporting { message, code, r#type, link }
+            }
+            Err(e) => e.into(),
+        },
+        ureq::Error::Transport(transport) => io::Error::new(io::ErrorKind::Other, transport).into(),
+    }
+}
+
+enum ExportIndex {}

crates/index-scheduler/src/scheduler/process_index_operation.rs

@@ -1,11 +1,13 @@
+use std::sync::Arc;
+
 use bumpalo::collections::CollectIn;
 use bumpalo::Bump;
 use meilisearch_types::heed::RwTxn;
 use meilisearch_types::milli::documents::PrimaryKey;
-use meilisearch_types::milli::progress::Progress;
+use meilisearch_types::milli::progress::{EmbedderStats, Progress};
 use meilisearch_types::milli::update::new::indexer::{self, UpdateByFunction};
 use meilisearch_types::milli::update::DocumentAdditionResult;
-use meilisearch_types::milli::{self, Filter, ThreadPoolNoAbortBuilder};
+use meilisearch_types::milli::{self, ChannelCongestion, Filter};
 use meilisearch_types::settings::apply_settings_to_builder;
 use meilisearch_types::tasks::{Details, KindWithContent, Status, Task};
 use meilisearch_types::Index;
@@ -24,7 +26,7 @@ impl IndexScheduler {
     /// The list of processed tasks.
     #[tracing::instrument(
         level = "trace",
-        skip(self, index_wtxn, index, progress),
+        skip(self, index_wtxn, index, progress, embedder_stats),
         target = "indexing::scheduler"
     )]
     pub(crate) fn apply_index_operation<'i>(
@@ -32,10 +34,10 @@ impl IndexScheduler {
         index_wtxn: &mut RwTxn<'i>,
         index: &'i Index,
         operation: IndexOperation,
-        progress: Progress,
-    ) -> Result<Vec<Task>> {
+        progress: &Progress,
+        embedder_stats: Arc<EmbedderStats>,
+    ) -> Result<(Vec<Task>, Option<ChannelCongestion>)> {
         let indexer_alloc = Bump::new();
-
         let started_processing_at = std::time::Instant::now();
         let must_stop_processing = self.scheduler.must_stop_processing.clone();
 
@@ -60,26 +62,24 @@ impl IndexScheduler {
                     };
                 }
 
-                Ok(tasks)
+                Ok((tasks, None))
             }
-            IndexOperation::DocumentOperation {
-                index_uid,
-                primary_key,
-                method,
-                operations,
-                mut tasks,
-            } => {
+            IndexOperation::DocumentOperation { index_uid, primary_key, operations, mut tasks } => {
                 progress.update_progress(DocumentOperationProgress::RetrievingConfig);
                 // TODO: at some point, for better efficiency we might want to reuse the bumpalo for successive batches.
                 // this is made difficult by the fact we're doing private clones of the index scheduler and sending it
                 // to a fresh thread.
                 let mut content_files = Vec::new();
                 for operation in &operations {
-                    if let DocumentOperation::Add(content_uuid) = operation {
+                    match operation {
+                        DocumentOperation::Replace(content_uuid)
+                        | DocumentOperation::Update(content_uuid) => {
                             let content_file = self.queue.file_store.get_update(*content_uuid)?;
                             let mmap = unsafe { memmap2::Mmap::map(&content_file)? };
                             content_files.push(mmap);
                         }
+                        _ => (),
+                    }
                 }
 
                 let rtxn = index.read_txn()?;
@@ -87,17 +87,24 @@ impl IndexScheduler {
                 let mut new_fields_ids_map = db_fields_ids_map.clone();
 
                 let mut content_files_iter = content_files.iter();
-                let mut indexer = indexer::DocumentOperation::new(method);
+                let mut indexer = indexer::DocumentOperation::new();
                 let embedders = index
+                    .embedding_configs()
                     .embedding_configs(index_wtxn)
-                    .map_err(|e| Error::from_milli(e, Some(index_uid.clone())))?;
+                    .map_err(|e| Error::from_milli(e.into(), Some(index_uid.clone())))?;
                 let embedders = self.embedders(index_uid.clone(), embedders)?;
                 for operation in operations {
                     match operation {
-                        DocumentOperation::Add(_content_uuid) => {
+                        DocumentOperation::Replace(_content_uuid) => {
                             let mmap = content_files_iter.next().unwrap();
                             indexer
-                                .add_documents(mmap)
+                                .replace_documents(mmap)
+                                .map_err(|e| Error::from_milli(e, Some(index_uid.clone())))?;
+                        }
+                        DocumentOperation::Update(_content_uuid) => {
+                            let mmap = content_files_iter.next().unwrap();
+                            indexer
+                                .update_documents(mmap)
                                 .map_err(|e| Error::from_milli(e, Some(index_uid.clone())))?;
                         }
                         DocumentOperation::Delete(document_ids) => {
@@ -110,18 +117,8 @@ impl IndexScheduler {
                     }
                 }
 
-                let local_pool;
                 let indexer_config = self.index_mapper.indexer_config();
-                let pool = match &indexer_config.thread_pool {
-                    Some(pool) => pool,
-                    None => {
-                        local_pool = ThreadPoolNoAbortBuilder::new()
-                            .thread_name(|i| format!("indexing-thread-{i}"))
-                            .build()
-                            .unwrap();
-                        &local_pool
-                    }
-                };
+                let pool = &indexer_config.thread_pool;
 
                 progress.update_progress(DocumentOperationProgress::ComputingDocumentChanges);
                 let (document_changes, operation_stats, primary_key) = indexer
@@ -169,7 +166,9 @@ impl IndexScheduler {
                 }
 
                 progress.update_progress(DocumentOperationProgress::Indexing);
+                let mut congestion = None;
                 if tasks.iter().any(|res| res.error.is_none()) {
+                    congestion = Some(
                         indexer::index(
                             index_wtxn,
                             index,
@@ -181,9 +180,11 @@ impl IndexScheduler {
                             &document_changes,
                             embedders,
                             &|| must_stop_processing.get(),
-                        &progress,
+                            progress,
+                            &embedder_stats,
                         )
-                    .map_err(|e| Error::from_milli(e, Some(index_uid.clone())))?;
+                        .map_err(|e| Error::from_milli(e, Some(index_uid.clone())))?,
+                    );
 
                     let addition = DocumentAdditionResult {
                         indexed_documents: candidates_count,
@@ -195,7 +196,7 @@ impl IndexScheduler {
                     tracing::info!(indexing_result = ?addition, processed_in = ?started_processing_at.elapsed(), "document indexing done");
                 }
 
-                Ok(tasks)
+                Ok((tasks, congestion))
             }
             IndexOperation::DocumentEdition { index_uid, mut task } => {
                 progress.update_progress(DocumentEditionProgress::RetrievingConfig);
@@ -243,7 +244,7 @@ impl IndexScheduler {
                         edited_documents: Some(0),
                     });
 
-                    return Ok(vec![task]);
+                    return Ok((vec![task], None));
                 }
 
                 let rtxn = index.read_txn()?;
@@ -258,19 +259,10 @@ impl IndexScheduler {
 
                 let result_count = Ok((candidates.len(), candidates.len())) as Result<_>;
 
+                let mut congestion = None;
                 if task.error.is_none() {
-                    let local_pool;
                     let indexer_config = self.index_mapper.indexer_config();
-                    let pool = match &indexer_config.thread_pool {
-                        Some(pool) => pool,
-                        None => {
-                            local_pool = ThreadPoolNoAbortBuilder::new()
-                                .thread_name(|i| format!("indexing-thread-{i}"))
-                                .build()
-                                .unwrap();
-                            &local_pool
-                        }
-                    };
+                    let pool = &indexer_config.thread_pool;
 
                     let candidates_count = candidates.len();
                     progress.update_progress(DocumentEditionProgress::ComputingDocumentChanges);
@@ -283,11 +275,13 @@ impl IndexScheduler {
                         })
                         .unwrap()?;
                     let embedders = index
+                        .embedding_configs()
                         .embedding_configs(index_wtxn)
-                        .map_err(|err| Error::from_milli(err, Some(index_uid.clone())))?;
+                        .map_err(|err| Error::from_milli(err.into(), Some(index_uid.clone())))?;
                     let embedders = self.embedders(index_uid.clone(), embedders)?;
 
                     progress.update_progress(DocumentEditionProgress::Indexing);
+                    congestion = Some(
                         indexer::index(
                             index_wtxn,
                             index,
@@ -299,9 +293,11 @@ impl IndexScheduler {
                             &document_changes,
                             embedders,
                             &|| must_stop_processing.get(),
-                        &progress,
+                            progress,
+                            &embedder_stats,
                         )
-                    .map_err(|err| Error::from_milli(err, Some(index_uid.clone())))?;
+                        .map_err(|err| Error::from_milli(err, Some(index_uid.clone())))?,
+                    );
 
                     let addition = DocumentAdditionResult {
                         indexed_documents: candidates_count,
@@ -337,7 +333,7 @@ impl IndexScheduler {
                     }
                 }
 
-                Ok(vec![task])
+                Ok((vec![task], congestion))
             }
             IndexOperation::DocumentDeletion { mut tasks, index_uid } => {
                 progress.update_progress(DocumentDeletionProgress::RetrievingConfig);
@@ -404,7 +400,7 @@ impl IndexScheduler {
                 }
 
                 if to_delete.is_empty() {
-                    return Ok(tasks);
+                    return Ok((tasks, None));
                 }
 
                 let rtxn = index.read_txn()?;
@@ -418,19 +414,10 @@ impl IndexScheduler {
                     PrimaryKey::new_or_insert(primary_key, &mut new_fields_ids_map)
                         .map_err(|err| Error::from_milli(err.into(), Some(index_uid.clone())))?;
 
+                let mut congestion = None;
                 if !tasks.iter().all(|res| res.error.is_some()) {
-                    let local_pool;
                     let indexer_config = self.index_mapper.indexer_config();
-                    let pool = match &indexer_config.thread_pool {
-                        Some(pool) => pool,
-                        None => {
-                            local_pool = ThreadPoolNoAbortBuilder::new()
-                                .thread_name(|i| format!("indexing-thread-{i}"))
-                                .build()
-                                .unwrap();
-                            &local_pool
-                        }
-                    };
+                    let pool = &indexer_config.thread_pool;
 
                     progress.update_progress(DocumentDeletionProgress::DeleteDocuments);
                     let mut indexer = indexer::DocumentDeletion::new();
@@ -438,11 +425,13 @@ impl IndexScheduler {
                     indexer.delete_documents_by_docids(to_delete);
                     let document_changes = indexer.into_changes(&indexer_alloc, primary_key);
                     let embedders = index
+                        .embedding_configs()
                         .embedding_configs(index_wtxn)
-                        .map_err(|err| Error::from_milli(err, Some(index_uid.clone())))?;
+                        .map_err(|err| Error::from_milli(err.into(), Some(index_uid.clone())))?;
                     let embedders = self.embedders(index_uid.clone(), embedders)?;
 
                     progress.update_progress(DocumentDeletionProgress::Indexing);
+                    congestion = Some(
                         indexer::index(
                             index_wtxn,
                             index,
@@ -454,9 +443,11 @@ impl IndexScheduler {
                             &document_changes,
                             embedders,
                             &|| must_stop_processing.get(),
-                        &progress,
+                            progress,
+                            &embedder_stats,
                         )
-                    .map_err(|err| Error::from_milli(err, Some(index_uid.clone())))?;
+                        .map_err(|err| Error::from_milli(err, Some(index_uid.clone())))?,
+                    );
 
                     let addition = DocumentAdditionResult {
                         indexed_documents: candidates_count,
@@ -468,7 +459,7 @@ impl IndexScheduler {
                     tracing::info!(indexing_result = ?addition, processed_in = ?started_processing_at.elapsed(), "document indexing done");
                 }
 
-                Ok(tasks)
+                Ok((tasks, congestion))
             }
             IndexOperation::Settings { index_uid, settings, mut tasks } => {
                 progress.update_progress(SettingsProgress::RetrievingAndMergingTheSettings);
@@ -486,14 +477,11 @@ impl IndexScheduler {
                 }
 
                 progress.update_progress(SettingsProgress::ApplyTheSettings);
-                builder
-                    .execute(
-                        |indexing_step| tracing::debug!(update = ?indexing_step),
-                        || must_stop_processing.get(),
-                    )
+                let congestion = builder
+                    .execute(&|| must_stop_processing.get(), progress, embedder_stats)
                     .map_err(|err| Error::from_milli(err, Some(index_uid.clone())))?;
 
-                Ok(tasks)
+                Ok((tasks, congestion))
             }
             IndexOperation::DocumentClearAndSetting {
                 index_uid,
@@ -501,26 +489,28 @@ impl IndexScheduler {
                 settings,
                 settings_tasks,
             } => {
-                let mut import_tasks = self.apply_index_operation(
+                let (mut import_tasks, _congestion) = self.apply_index_operation(
                     index_wtxn,
                     index,
                     IndexOperation::DocumentClear {
                         index_uid: index_uid.clone(),
                         tasks: cleared_tasks,
                     },
-                    progress.clone(),
+                    progress,
+                    embedder_stats.clone(),
                 )?;
 
-                let settings_tasks = self.apply_index_operation(
+                let (settings_tasks, _congestion) = self.apply_index_operation(
                     index_wtxn,
                     index,
                     IndexOperation::Settings { index_uid, settings, tasks: settings_tasks },
                     progress,
+                    embedder_stats,
                 )?;
 
                 let mut tasks = settings_tasks;
                 tasks.append(&mut import_tasks);
-                Ok(tasks)
+                Ok((tasks, None))
             }
         }
     }

crates/index-scheduler/src/scheduler/process_snapshot_creation.rs

@@ -4,7 +4,6 @@ use std::sync::atomic::Ordering;
 
 use meilisearch_types::heed::CompactionOption;
 use meilisearch_types::milli::progress::{Progress, VariableNameStep};
-use meilisearch_types::milli::{self};
 use meilisearch_types::tasks::{Status, Task};
 use meilisearch_types::{compression, VERSION_FILE_NAME};
 
@@ -28,7 +27,7 @@ impl IndexScheduler {
 
         // 2. Snapshot the index-scheduler LMDB env
         //
-        // When we call copy_to_file, LMDB opens a read transaction by itself,
+        // When we call copy_to_path, LMDB opens a read transaction by itself,
         // we can't provide our own. It is an issue as we would like to know
         // the update files to copy but new ones can be enqueued between the copy
         // of the env and the new transaction we open to retrieve the enqueued tasks.
@@ -42,7 +41,12 @@ impl IndexScheduler {
         progress.update_progress(SnapshotCreationProgress::SnapshotTheIndexScheduler);
         let dst = temp_snapshot_dir.path().join("tasks");
         fs::create_dir_all(&dst)?;
-        self.env.copy_to_file(dst.join("data.mdb"), CompactionOption::Enabled)?;
+        let compaction_option = if self.scheduler.experimental_no_snapshot_compaction {
+            CompactionOption::Disabled
+        } else {
+            CompactionOption::Enabled
+        };
+        self.env.copy_to_path(dst.join("data.mdb"), compaction_option)?;
 
         // 2.2 Create a read transaction on the index-scheduler
         let rtxn = self.env.read_txn()?;
@@ -81,7 +85,7 @@ impl IndexScheduler {
             let dst = temp_snapshot_dir.path().join("indexes").join(uuid.to_string());
             fs::create_dir_all(&dst)?;
             index
-                .copy_to_file(dst.join("data.mdb"), CompactionOption::Enabled)
+                .copy_to_path(dst.join("data.mdb"), compaction_option)
                 .map_err(|e| Error::from_milli(e, Some(name.to_string())))?;
         }
 
@@ -91,14 +95,7 @@ impl IndexScheduler {
         progress.update_progress(SnapshotCreationProgress::SnapshotTheApiKeys);
         let dst = temp_snapshot_dir.path().join("auth");
         fs::create_dir_all(&dst)?;
-        // TODO We can't use the open_auth_store_env function here but we should
-        let auth = unsafe {
-            milli::heed::EnvOpenOptions::new()
-                .map_size(1024 * 1024 * 1024) // 1 GiB
-                .max_dbs(2)
-                .open(&self.scheduler.auth_path)
-        }?;
-        auth.copy_to_file(dst.join("data.mdb"), CompactionOption::Enabled)?;
+        self.scheduler.auth_env.copy_to_path(dst.join("data.mdb"), compaction_option)?;
 
         // 5. Copy and tarball the flat snapshot
         progress.update_progress(SnapshotCreationProgress::CreateTheTarball);

crates/index-scheduler/src/scheduler/process_upgrade.rs

@@ -12,10 +12,14 @@ impl IndexScheduler {
         #[cfg(test)]
         self.maybe_fail(crate::test_utils::FailureLocation::ProcessUpgrade)?;
 
-        enum UpgradeIndex {}
         let indexes = self.index_names()?;
 
         for (i, uid) in indexes.iter().enumerate() {
+            let must_stop_processing = self.scheduler.must_stop_processing.clone();
+
+            if must_stop_processing.get() {
+                return Err(Error::AbortedTask);
+            }
             progress.update_progress(VariableNameStep::<UpgradeIndex>::new(
                 format!("Upgrading index `{uid}`"),
                 i as u32,
@@ -27,6 +31,7 @@ impl IndexScheduler {
                 &mut index_wtxn,
                 &index,
                 db_version,
+                || must_stop_processing.get(),
                 progress.clone(),
             )
             .map_err(|e| Error::from_milli(e, Some(uid.to_string())))?;
@@ -46,4 +51,42 @@ impl IndexScheduler {
 
         Ok(())
     }
+
+    pub fn process_rollback(&self, db_version: (u32, u32, u32), progress: &Progress) -> Result<()> {
+        let mut wtxn = self.env.write_txn()?;
+        tracing::info!(?db_version, "roll back index scheduler version");
+        self.version.set_version(&mut wtxn, db_version)?;
+        let db_path = self.scheduler.version_file_path.parent().unwrap();
+        wtxn.commit()?;
+
+        let indexes = self.index_names()?;
+
+        tracing::info!("roll backing all indexes");
+        for (i, uid) in indexes.iter().enumerate() {
+            progress.update_progress(VariableNameStep::<UpgradeIndex>::new(
+                format!("Rollbacking index `{uid}`"),
+                i as u32,
+                indexes.len() as u32,
+            ));
+            let index_schd_rtxn = self.env.read_txn()?;
+
+            let rollback_outcome =
+                self.index_mapper.rollback_index(&index_schd_rtxn, uid, db_version)?;
+            if !rollback_outcome.succeeded() {
+                return Err(crate::Error::RollbackFailed { index: uid.clone(), rollback_outcome });
+            }
+        }
+
+        tracing::info!(?db_path, ?db_version, "roll back version file");
+        meilisearch_types::versioning::create_version_file(
+            db_path,
+            db_version.0,
+            db_version.1,
+            db_version.2,
+        )?;
+
+        Ok(())
+    }
 }
+
+enum UpgradeIndex {}

crates/index-scheduler/src/scheduler/snapshots/index_scheduler__scheduler__test__settings_update-7.snap

@@ -0,0 +1,17 @@
+---
+source: crates/index-scheduler/src/scheduler/test.rs
+expression: config.embedder_options
+---
+{
+  "Rest": {
+    "api_key": "My super secret",
+    "distribution": null,
+    "dimensions": 4,
+    "url": "http://localhost:7777",
+    "request": "{{text}}",
+    "search_fragments": {},
+    "indexing_fragments": {},
+    "response": "{{embedding}}",
+    "headers": {}
+  }
+}

crates/index-scheduler/src/scheduler/snapshots/index_scheduler__scheduler__test_embedders__import_vectors-14.snap

@@ -0,0 +1,12 @@
+---
+source: crates/index-scheduler/src/scheduler/test_embedders.rs
+expression: simple_hf_config.embedder_options
+---
+{
+  "HuggingFace": {
+    "model": "sentence-transformers/all-MiniLM-L6-v2",
+    "revision": "e4ce9877abf3edfe10b0d82785e83bdcb973e22e",
+    "distribution": null,
+    "pooling": "useModel"
+  }
+}

crates/index-scheduler/src/scheduler/snapshots/index_scheduler__scheduler__test_embedders__import_vectors-27.snap

@@ -0,0 +1,15 @@
+---
+source: crates/index-scheduler/src/scheduler/test_embedders.rs
+expression: doc
+---
+{
+  "doggo": "Intel",
+  "breed": "beagle",
+  "_vectors": {
+    "noise": [
+      0.1,
+      0.2,
+      0.3
+    ]
+  }
+}

crates/index-scheduler/src/scheduler/snapshots/index_scheduler__scheduler__test_embedders__import_vectors-40.snap

@@ -0,0 +1,15 @@
+---
+source: crates/index-scheduler/src/scheduler/test_embedders.rs
+expression: doc
+---
+{
+  "doggo": "kefir",
+  "breed": "patou",
+  "_vectors": {
+    "noise": [
+      0.1,
+      0.2,
+      0.3
+    ]
+  }
+}

crates/index-scheduler/src/scheduler/snapshots/index_scheduler__scheduler__test_embedders__import_vectors-8.snap

@@ -1,12 +1,17 @@
 ---
 source: crates/index-scheduler/src/scheduler/test_embedders.rs
-expression: simple_hf_config.embedder_options
-snapshot_kind: text
+expression: fakerest_config.embedder_options
 ---
 {
-  "HuggingFace": {
-    "model": "sentence-transformers/all-MiniLM-L6-v2",
-    "revision": "e4ce9877abf3edfe10b0d82785e83bdcb973e22e",
-    "distribution": null
+  "Rest": {
+    "api_key": "My super secret",
+    "distribution": null,
+    "dimensions": 384,
+    "url": "http://localhost:7777",
+    "request": "{{text}}",
+    "search_fragments": {},
+    "indexing_fragments": {},
+    "response": "{{embedding}}",
+    "headers": {}
   }
 }

crates/index-scheduler/src/scheduler/snapshots/test.rs/cancel_enqueued_task/cancel_processed.snap

@@ -1,6 +1,5 @@
 ---
 source: crates/index-scheduler/src/scheduler/test.rs
-snapshot_kind: text
 ---
 ### Autobatching Enabled = true
 ### Processing batch None:
@@ -40,7 +39,7 @@ catto [0,]
 [timestamp] [0,1,]
 ----------------------------------------------------------------------
 ### All Batches:
-0 {uid: 0, details: {"receivedDocuments":1,"indexedDocuments":0,"matchedTasks":1,"canceledTasks":1,"originalFilter":"test_query"}, stats: {"totalNbTasks":2,"status":{"succeeded":1,"canceled":1},"types":{"documentAdditionOrUpdate":1,"taskCancelation":1},"indexUids":{"catto":1}}, }
+0 {uid: 0, details: {"receivedDocuments":1,"indexedDocuments":0,"matchedTasks":1,"canceledTasks":1,"originalFilter":"test_query"}, stats: {"totalNbTasks":2,"status":{"succeeded":1,"canceled":1},"types":{"documentAdditionOrUpdate":1,"taskCancelation":1},"indexUids":{"catto":1}}, stop reason: "created batch containing only task with id 1 of type `taskCancelation` that cannot be batched with any other task.", }
 ----------------------------------------------------------------------
 ### Batch to tasks mapping:
 0 [0,1,]

crates/index-scheduler/src/scheduler/snapshots/test.rs/cancel_mix_of_tasks/aborted_indexation.snap

@@ -1,11 +1,10 @@
 ---
 source: crates/index-scheduler/src/scheduler/test.rs
-snapshot_kind: text
 ---
 ### Autobatching Enabled = true
 ### Processing batch Some(1):
 [1,]
-{uid: 1, details: {"receivedDocuments":1,"indexedDocuments":null}, stats: {"totalNbTasks":1,"status":{"processing":1},"types":{"documentAdditionOrUpdate":1},"indexUids":{"beavero":1}}, }
+{uid: 1, details: {"receivedDocuments":1,"indexedDocuments":null}, stats: {"totalNbTasks":1,"status":{"processing":1},"types":{"documentAdditionOrUpdate":1},"indexUids":{"beavero":1}}, stop reason: "batched all enqueued tasks for index `beavero`", }
 ----------------------------------------------------------------------
 ### All Tasks:
 0 {uid: 0, batch_uid: 0, status: succeeded, details: { received_documents: 1, indexed_documents: Some(1) }, kind: DocumentAdditionOrUpdate { index_uid: "catto", primary_key: None, method: ReplaceDocuments, content_file: 00000000-0000-0000-0000-000000000000, documents_count: 1, allow_index_creation: true }}
@@ -47,7 +46,7 @@ catto: { number_of_documents: 1, field_distribution: {"id": 1} }
 [timestamp] [0,]
 ----------------------------------------------------------------------
 ### All Batches:
-0 {uid: 0, details: {"receivedDocuments":1,"indexedDocuments":1}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"documentAdditionOrUpdate":1},"indexUids":{"catto":1}}, }
+0 {uid: 0, details: {"receivedDocuments":1,"indexedDocuments":1}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"documentAdditionOrUpdate":1},"indexUids":{"catto":1}}, stop reason: "batched all enqueued tasks for index `catto`", }
 ----------------------------------------------------------------------
 ### Batch to tasks mapping:
 0 [0,]

crates/index-scheduler/src/scheduler/snapshots/test.rs/cancel_mix_of_tasks/cancel_processed.snap

@@ -1,6 +1,5 @@
 ---
 source: crates/index-scheduler/src/scheduler/test.rs
-snapshot_kind: text
 ---
 ### Autobatching Enabled = true
 ### Processing batch None:
@@ -50,8 +49,8 @@ catto: { number_of_documents: 1, field_distribution: {"id": 1} }
 [timestamp] [1,2,3,]
 ----------------------------------------------------------------------
 ### All Batches:
-0 {uid: 0, details: {"receivedDocuments":1,"indexedDocuments":1}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"documentAdditionOrUpdate":1},"indexUids":{"catto":1}}, }
-1 {uid: 1, details: {"receivedDocuments":2,"indexedDocuments":0,"matchedTasks":3,"canceledTasks":2,"originalFilter":"test_query"}, stats: {"totalNbTasks":3,"status":{"succeeded":1,"canceled":2},"types":{"documentAdditionOrUpdate":2,"taskCancelation":1},"indexUids":{"beavero":1,"wolfo":1}}, }
+0 {uid: 0, details: {"receivedDocuments":1,"indexedDocuments":1}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"documentAdditionOrUpdate":1},"indexUids":{"catto":1}}, stop reason: "batched all enqueued tasks for index `catto`", }
+1 {uid: 1, details: {"receivedDocuments":2,"indexedDocuments":0,"matchedTasks":3,"canceledTasks":2,"originalFilter":"test_query"}, stats: {"totalNbTasks":3,"status":{"succeeded":1,"canceled":2},"types":{"documentAdditionOrUpdate":2,"taskCancelation":1},"indexUids":{"beavero":1,"wolfo":1}}, stop reason: "created batch containing only task with id 3 of type `taskCancelation` that cannot be batched with any other task.", }
 ----------------------------------------------------------------------
 ### Batch to tasks mapping:
 0 [0,]

crates/index-scheduler/src/scheduler/snapshots/test.rs/cancel_mix_of_tasks/first_task_processed.snap

@@ -1,6 +1,5 @@
 ---
 source: crates/index-scheduler/src/scheduler/test.rs
-snapshot_kind: text
 ---
 ### Autobatching Enabled = true
 ### Processing batch None:
@@ -42,7 +41,7 @@ catto: { number_of_documents: 1, field_distribution: {"id": 1} }
 [timestamp] [0,]
 ----------------------------------------------------------------------
 ### All Batches:
-0 {uid: 0, details: {"receivedDocuments":1,"indexedDocuments":1}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"documentAdditionOrUpdate":1},"indexUids":{"catto":1}}, }
+0 {uid: 0, details: {"receivedDocuments":1,"indexedDocuments":1}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"documentAdditionOrUpdate":1},"indexUids":{"catto":1}}, stop reason: "batched all enqueued tasks for index `catto`", }
 ----------------------------------------------------------------------
 ### Batch to tasks mapping:
 0 [0,]

crates/index-scheduler/src/scheduler/snapshots/test.rs/cancel_mix_of_tasks/processing_second_task_cancel_enqueued.snap

@@ -1,11 +1,10 @@
 ---
 source: crates/index-scheduler/src/scheduler/test.rs
-snapshot_kind: text
 ---
 ### Autobatching Enabled = true
 ### Processing batch Some(1):
 [1,]
-{uid: 1, details: {"receivedDocuments":1,"indexedDocuments":null}, stats: {"totalNbTasks":1,"status":{"processing":1},"types":{"documentAdditionOrUpdate":1},"indexUids":{"beavero":1}}, }
+{uid: 1, details: {"receivedDocuments":1,"indexedDocuments":null}, stats: {"totalNbTasks":1,"status":{"processing":1},"types":{"documentAdditionOrUpdate":1},"indexUids":{"beavero":1}}, stop reason: "batched all enqueued tasks for index `beavero`", }
 ----------------------------------------------------------------------
 ### All Tasks:
 0 {uid: 0, batch_uid: 0, status: succeeded, details: { received_documents: 1, indexed_documents: Some(1) }, kind: DocumentAdditionOrUpdate { index_uid: "catto", primary_key: None, method: ReplaceDocuments, content_file: 00000000-0000-0000-0000-000000000000, documents_count: 1, allow_index_creation: true }}
@@ -46,7 +45,7 @@ catto: { number_of_documents: 1, field_distribution: {"id": 1} }
 [timestamp] [0,]
 ----------------------------------------------------------------------
 ### All Batches:
-0 {uid: 0, details: {"receivedDocuments":1,"indexedDocuments":1}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"documentAdditionOrUpdate":1},"indexUids":{"catto":1}}, }
+0 {uid: 0, details: {"receivedDocuments":1,"indexedDocuments":1}, stats: {"totalNbTasks":1,"status":{"succeeded":1},"types":{"documentAdditionOrUpdate":1},"indexUids":{"catto":1}}, stop reason: "batched all enqueued tasks for index `catto`", }
 ----------------------------------------------------------------------
 ### Batch to tasks mapping:
 0 [0,]