Upgrade rustls to 0.21.10 and ring to 0.17

4391: Tracing r=dureuill a=irevoire

# Pull Request

- [ ] Hide the parameters of the process batch
- [x] Make actix-web trace every call on every route
- [x] Remove all `env_logger`/`logs` dependencies
- [x] Be able to enable or disable the memory measurement using the `/logs` route parameters

See the following product discussion: https://github.com/orgs/meilisearch/discussions/721

Supersedes https://github.com/meilisearch/meilisearch/pull/4338

## Related issue
Fixes https://github.com/meilisearch/meilisearch/issues/4317

## What does this PR do?

Update the format of the logs from:
```
[2024-02-06T14:54:11Z INFO  actix_server::builder] starting 10 workers
```

to

```
2024-02-06T13:58:14.710803Z  INFO actix_server::builder: 200: starting 10 workers
```

First, run meilisearch with the route enabled via the feature flag:
- `cargo run --experimental-enable-logs-route`
- Or at runtime by sending the following payload:
```
curl \
  -X PATCH 'http://localhost:7700/experimental-features/' \
  -H 'Content-Type: application/json'  \
--data-binary '{
    "logsRoute": true
  }'
```

Then gather data from meilisearch by calling for example:
```
curl \
	-X POST http://localhost:7700/logs \
	-H 'Content-Type: application/json' \
	--data-binary '{
	    "mode": "fmt",
            "target": "milli=trace"
    }'
```

Once your operation is over, tell meilisearch to stop the route:
```
curl \
	-X DELETE http://localhost:7700/logs
```

----

In the case you’re profiling code, you will be interested by the next command that converts the output of the route to a format that the firefox profiler can understand.

```bash
cargo run --release --bin trace-to-firefox -- 2024-01-17_17:07:55-indexing-trace.json
```

Then go to https://profiler.firefox.com and load it.
Note that we can also share the profiles using the https://share.firefox.dev website.


Co-authored-by: Louis Dureuil <louis@meilisearch.com>
Co-authored-by: Clément Renault <clement@meilisearch.com>
Co-authored-by: Tamo <tamo@meilisearch.com>

.cargo/config.toml

@@ -0,0 +1,2 @@
+[alias]
+xtask = "run --package xtask --"

.github/ISSUE_TEMPLATE/sprint_issue.md

@@ -27,6 +27,23 @@ Related spec: WIP
 - [ ] If prototype validated, merge changes into `main`
 - [ ] Update the spec
 
+### Reminders when modifying the Setting API
+
+<!--- Special steps to remind when adding a new index setting -->
+
+- [ ] Ensure the new setting route is at least tested by the [`test_setting_routes` macro](https://github.com/meilisearch/meilisearch/blob/5204c0b60b384cbc79621b6b2176fca086069e8e/meilisearch/tests/settings/get_settings.rs#L276)
+- [ ] Ensure Analytics are fully implemented
+  - [ ] `/settings/my-new-setting` configurated in the [`make_setting_routes` macro](https://github.com/meilisearch/meilisearch/blob/5204c0b60b384cbc79621b6b2176fca086069e8e/meilisearch/src/routes/indexes/settings.rs#L141-L165)
+  - [ ] global `/settings` route configurated in the [`update_all` function](https://github.com/meilisearch/meilisearch/blob/5204c0b60b384cbc79621b6b2176fca086069e8e/meilisearch/src/routes/indexes/settings.rs#L655-L751)
+- [ ] Ensure the dump serializing is consistent with the `/settings` route serializing, e.g., enums case can be different (`camelCase` in route and `PascalCase` in the dump)
+
+#### Special cases when adding a setting for an experimental feature
+
+- [ ] ⚠️ API stability: The setting does not appear on the main settings route when the feature has never been enabled (e.g. mark it `Unset` when returned from the index in this situation. See [an example](https://github.com/meilisearch/meilisearch/blob/7a89abd2a025606a42f8b219e539117eb2eb029f/meilisearch-types/src/settings.rs#L608))
+- [ ] The setting cannot be set when the feature is disabled, either by the main settings route or the subroute (see [`validate_settings` function](https://github.com/meilisearch/meilisearch/blob/7a89abd2a025606a42f8b219e539117eb2eb029f/meilisearch/src/routes/indexes/settings.rs#L811))
+- [ ] If possible, the setting is reset when the feature is disabled (hard if it requires reindexing)
+
 ## Impacted teams
 
 <!---Ping the related teams. Ask for the engine manager if any hesitation-->
+<!---@meilisearch/docs-team when there is any API change, e.g. settings addition-->

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

@@ -97,7 +97,7 @@ jobs:
       - name: Send CI information to Cloud team
         # Do not send if nightly build (i.e. 'schedule' or 'workflow_dispatch' event)
         if: github.event_name == 'push'
-        uses: peter-evans/repository-dispatch@v2
+        uses: peter-evans/repository-dispatch@v3
         with:
           token: ${{ secrets.MEILI_BOT_GH_PAT }}
           repository: meilisearch/meilisearch-cloud

.github/workflows/test-suite.yml

@@ -66,6 +66,10 @@ jobs:
       - uses: actions/checkout@v3
       - name: Cache dependencies
         uses: Swatinem/rust-cache@v2.7.1
+      - uses: actions-rs/toolchain@v1
+        with:
+          toolchain: stable
+          override: true
       - name: Run cargo check without any default features
         uses: actions-rs/cargo@v1
         with:
@@ -78,7 +82,7 @@ jobs:
           args: --locked --release --all
 
   test-all-features:
-    name: Tests all features
+    name: Tests almost all features
     runs-on: ubuntu-latest
     container:
       # Use ubuntu-18.04 to compile with glibc 2.27, which are the production expectations
@@ -94,16 +98,12 @@ jobs:
         with:
           toolchain: stable
           override: true
-      - name: Run cargo build with all features
-        uses: actions-rs/cargo@v1
-        with:
-          command: build
-          args: --workspace --locked --release --all-features
-      - name: Run cargo test with all features
-        uses: actions-rs/cargo@v1
-        with:
-          command: test
-          args: --workspace --locked --release --all-features
+      - name: Run cargo build with almost all features
+        run: |
+          cargo build --workspace --locked --release --features "$(cargo xtask list-features --exclude-feature cuda)"
+      - name: Run cargo test with almost all features
+        run: |
+          cargo test --workspace --locked --release --features "$(cargo xtask list-features --exclude-feature cuda)"
 
   test-disabled-tokenization:
     name: Test disabled tokenization
@@ -164,7 +164,7 @@ jobs:
       - uses: actions-rs/toolchain@v1
         with:
           profile: minimal
-          toolchain: 1.71.1
+          toolchain: 1.75.0
           override: true
           components: clippy
       - name: Cache dependencies

CONTRIBUTING.md

@@ -75,6 +75,12 @@ If you get a "Too many open files" error you might want to increase the open fil
 ulimit -Sn 3000
 ```
 
+#### Build tools
+
+Meilisearch follows the [cargo xtask](https://github.com/matklad/cargo-xtask) workflow to provide some build tools.
+
+Run `cargo xtask --help` from the root of the repository to find out what is available.
+
 ## Git Guidelines
 
 ### Git Branches

Cargo.toml

@@ -16,11 +16,16 @@ members = [
     "json-depth-checker",
     "benchmarks",
     "fuzzers",
+    "tracing-trace",
+    "xtask",
 ]
 
 [workspace.package]
-version = "1.6.0"
-authors = ["Quentin de Quelen <quentin@dequelen.me>", "Clément Renault <clement@meilisearch.com>"]
+version = "1.6.1"
+authors = [
+    "Quentin de Quelen <quentin@dequelen.me>",
+    "Clément Renault <clement@meilisearch.com>",
+]
 description = "Meilisearch HTTP server"
 homepage = "https://meilisearch.com"
 readme = "README.md"

Dockerfile

@@ -1,5 +1,5 @@
 # Compile
-FROM    rust:1.71.1-alpine3.18 AS compiler
+FROM    rust:1.75.0-alpine3.18 AS compiler
 
 RUN     apk add -q --update-cache --no-cache build-base openssl-dev
 

README.md

@@ -41,10 +41,10 @@ Meilisearch helps you shape a delightful search experience in a snap, offering f
 ## ✨ Features
 
 - **Search-as-you-type:** find search results in less than 50 milliseconds
-- **[Typo tolerance](https://www.meilisearch.com/docs/learn/getting_started/customizing_relevancy?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=features#typo-tolerance):** get relevant matches even when queries contain typos and misspellings
+- **[Typo tolerance](https://www.meilisearch.com/docs/learn/configuration/typo_tolerance?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=features):** get relevant matches even when queries contain typos and misspellings
 - **[Filtering](https://www.meilisearch.com/docs/learn/fine_tuning_results/filtering?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=features) and [faceted search](https://www.meilisearch.com/docs/learn/fine_tuning_results/faceted_search?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=features):** enhance your users' search experience with custom filters and build a faceted search interface in a few lines of code
 - **[Sorting](https://www.meilisearch.com/docs/learn/fine_tuning_results/sorting?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=features):** sort results based on price, date, or pretty much anything else your users need
-- **[Synonym support](https://www.meilisearch.com/docs/learn/getting_started/customizing_relevancy?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=features#synonyms):** configure synonyms to include more relevant content in your search results
+- **[Synonym support](https://www.meilisearch.com/docs/learn/configuration/synonyms?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=features):** configure synonyms to include more relevant content in your search results
 - **[Geosearch](https://www.meilisearch.com/docs/learn/fine_tuning_results/geosearch?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=features):** filter and sort documents based on geographic data
 - **[Extensive language support](https://www.meilisearch.com/docs/learn/what_is_meilisearch/language?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=features):** search datasets in any language, with optimized support for Chinese, Japanese, Hebrew, and languages using the Latin alphabet
 - **[Security management](https://www.meilisearch.com/docs/learn/security/master_api_keys?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=features):** control which users can access what data with API keys that allow fine-grained permissions handling
@@ -61,8 +61,6 @@ You can consult Meilisearch's documentation at [https://www.meilisearch.com/docs
 
 For basic instructions on how to set up Meilisearch, add documents to an index, and search for documents, take a look at our [Quick Start](https://www.meilisearch.com/docs/learn/getting_started/quick_start?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=get-started) guide.
 
-You may also want to check out [Meilisearch 101](https://www.meilisearch.com/docs/learn/getting_started/filtering_and_sorting?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=get-started) for an introduction to some of Meilisearch's most popular features.
-
 ## ⚡ Supercharge your Meilisearch experience
 
 Say goodbye to server deployment and manual updates with [Meilisearch Cloud](https://www.meilisearch.com/cloud?utm_campaign=oss&utm_source=github&utm_medium=meilisearch). No credit card required.
@@ -101,7 +99,7 @@ Meilisearch is a search engine created by [Meili](https://www.welcometothejungle
 
 - For feature requests, please visit our [product repository](https://github.com/meilisearch/product/discussions)
 - Found a bug? Open an [issue](https://github.com/meilisearch/meilisearch/issues)!
-- Want to be part of our Discord community? [Join us!](https://discord.gg/meilisearch)
+- Want to be part of our Discord community? [Join us!](https://discord.meilisearch.com/?utm_campaign=oss&utm_source=github&utm_medium=meilisearch&utm_content=contact)
 
 Thank you for your support!
 

assets/grafana-dashboard.json

@@ -106,7 +106,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "meilisearch_index_count{job=\"meilisearch\", instance=\"$instance\"}",
+          "expr": "meilisearch_index_count{job=\"$job\", instance=\"$instance\"}",
           "interval": "",
           "legendFormat": "",
           "range": true,
@@ -165,7 +165,7 @@
             "type": "prometheus"
           },
           "editorMode": "builder",
-          "expr": "meilisearch_index_docs_count{job=\"meilisearch\", index=\"$Index\", instance=\"$instance\"}",
+          "expr": "meilisearch_index_docs_count{job=\"$job\", index=\"$Index\", instance=\"$instance\"}",
           "hide": false,
           "range": true,
           "refId": "A"
@@ -228,7 +228,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "round(increase(meilisearch_http_requests_total{method=\"POST\", path=\"/indexes/$Index/search\", job=\"meilisearch\"}[1h]))",
+          "expr": "round(increase(meilisearch_http_requests_total{method=\"POST\", path=\"/indexes/$Index/search\", job=\"$job\"}[1h]))",
           "interval": "",
           "legendFormat": "",
           "range": true,
@@ -288,7 +288,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "round(increase(meilisearch_http_requests_total{method=\"POST\", path=\"/indexes/$Index/search\", job=\"meilisearch\"}[24h]))",
+          "expr": "round(increase(meilisearch_http_requests_total{method=\"POST\", path=\"/indexes/$Index/search\", job=\"$job\"}[24h]))",
           "interval": "",
           "legendFormat": "",
           "range": true,
@@ -348,7 +348,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "round(increase(meilisearch_http_requests_total{method=\"POST\", path=\"/indexes/$Index/search\", job=\"meilisearch\"}[30d]))",
+          "expr": "round(increase(meilisearch_http_requests_total{method=\"POST\", path=\"/indexes/$Index/search\", job=\"$job\"}[30d]))",
           "interval": "",
           "legendFormat": "",
           "range": true,
@@ -447,7 +447,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "meilisearch_db_size_bytes{job=\"meilisearch\", instance=\"$instance\"}",
+          "expr": "meilisearch_db_size_bytes{job=\"$job\", instance=\"$instance\"}",
           "interval": "",
           "legendFormat": "Database size on disk",
           "range": true,
@@ -458,7 +458,7 @@
             "type": "prometheus"
           },
           "editorMode": "builder",
-          "expr": "meilisearch_used_db_size_bytes{job=\"meilisearch\", instance=\"$instance\"}",
+          "expr": "meilisearch_used_db_size_bytes{job=\"$job\", instance=\"$instance\"}",
           "hide": false,
           "legendFormat": "Used bytes",
           "range": true,
@@ -553,7 +553,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "rate(meilisearch_http_response_time_seconds_sum{instance=\"$instance\", job=\"meilisearch\"}[5m]) / rate(meilisearch_http_response_time_seconds_count[5m])",
+          "expr": "rate(meilisearch_http_response_time_seconds_sum{instance=\"$instance\", job=\"$job\"}[5m]) / rate(meilisearch_http_response_time_seconds_count[5m])",
           "interval": "",
           "legendFormat": "{{method}} {{path}}",
           "range": true,
@@ -646,7 +646,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "rate(meilisearch_http_requests_total{instance=\"$instance\", job=\"meilisearch\"}[5m])",
+          "expr": "rate(meilisearch_http_requests_total{instance=\"$instance\", job=\"$job\"}[5m])",
           "interval": "",
           "legendFormat": "{{method}} {{path}}",
           "range": true,
@@ -744,7 +744,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "sum by(le) (increase(meilisearch_http_response_time_seconds_bucket{path=\"/indexes/$Index/search\", instance=\"$instance\", job=\"meilisearch\"}[30s]))",
+          "expr": "sum by(le) (increase(meilisearch_http_response_time_seconds_bucket{path=\"/indexes/$Index/search\", instance=\"$instance\", job=\"$job\"}[30s]))",
           "format": "heatmap",
           "interval": "",
           "legendFormat": "{{le}}",
@@ -854,7 +854,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "meilisearch_nb_tasks{instance=\"$instance\", job=\"meilisearch\", kind=\"statuses\"}",
+          "expr": "meilisearch_nb_tasks{instance=\"$instance\", job=\"$job\", kind=\"statuses\"}",
           "interval": "",
           "legendFormat": "{{value}} ",
           "range": true,
@@ -947,7 +947,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "meilisearch_nb_tasks{instance=\"$instance\", job=\"meilisearch\", kind=\"types\"}",
+          "expr": "meilisearch_nb_tasks{instance=\"$instance\", job=\"$job\", kind=\"types\"}",
           "interval": "",
           "legendFormat": "{{value}} ",
           "range": true,
@@ -1040,7 +1040,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "meilisearch_nb_tasks{instance=\"$instance\", job=\"meilisearch\", kind=\"indexes\"}",
+          "expr": "meilisearch_nb_tasks{instance=\"$instance\", job=\"$job\", kind=\"indexes\"}",
           "interval": "",
           "legendFormat": "{{value}} ",
           "range": true,
@@ -1161,7 +1161,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "rate(process_cpu_seconds_total{job=\"meilisearch\", instance=\"$instance\"}[1m])",
+          "expr": "rate(process_cpu_seconds_total{job=\"$job\", instance=\"$instance\"}[1m])",
           "interval": "",
           "legendFormat": "process",
           "range": true,
@@ -1264,7 +1264,7 @@
           },
           "editorMode": "builder",
           "exemplar": true,
-          "expr": "process_resident_memory_bytes{job=\"meilisearch\", instance=\"$instance\"} / 1024 / 1024",
+          "expr": "process_resident_memory_bytes{job=\"$job\", instance=\"$instance\"} / 1024 / 1024",
           "interval": "",
           "legendFormat": "process",
           "range": true,
@@ -1342,6 +1342,33 @@
         "skipUrlSync": false,
         "sort": 0,
         "type": "query"
+      },
+      {
+        "current": {
+          "selected": true,
+          "text": "meilisearch",
+          "value": "meilisearch"
+        },
+        "datasource": {
+          "type": "prometheus"
+        },
+        "definition": "label_values(job)",
+        "description": "Prometheus job_name from scrape config (default is meilisearch)",
+        "hide": 0,
+        "includeAll": false,
+        "label": "Job",
+        "multi": false,
+        "name": "job",
+        "options": [],
+        "query": {
+          "query": "label_values(job)",
+          "refId": "StandardVariableQuery"
+        },
+        "refresh": 1,
+        "regex": "",
+        "skipUrlSync": false,
+        "sort": 0,
+        "type": "query"
       }
     ]
   },

benchmarks/Cargo.toml

@@ -11,24 +11,24 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-anyhow = "1.0.70"
-csv = "1.2.1"
+anyhow = "1.0.79"
+csv = "1.3.0"
 milli = { path = "../milli" }
-mimalloc = { version = "0.1.37", default-features = false }
-serde_json = { version = "1.0.95", features = ["preserve_order"] }
+mimalloc = { version = "0.1.39", default-features = false }
+serde_json = { version = "1.0.111", features = ["preserve_order"] }
 
 [dev-dependencies]
 criterion = { version = "0.5.1", features = ["html_reports"] }
 rand = "0.8.5"
 rand_chacha = "0.3.1"
-roaring = "0.10.1"
+roaring = "0.10.2"
 
 [build-dependencies]
-anyhow = "1.0.70"
-bytes = "1.4.0"
+anyhow = "1.0.79"
+bytes = "1.5.0"
 convert_case = "0.6.0"
-flate2 = "1.0.25"
-reqwest = { version = "0.11.16", features = ["blocking", "rustls-tls"], default-features = false }
+flate2 = "1.0.28"
+reqwest = { version = "0.11.23", features = ["blocking", "rustls-tls"], default-features = false }
 
 [features]
 default = ["milli/all-tokenizations"]

dump/Cargo.toml

@@ -11,22 +11,22 @@ readme.workspace = true
 license.workspace = true
 
 [dependencies]
-anyhow = "1.0.70"
-flate2 = "1.0.25"
-http = "0.2.9"
-log = "0.4.17"
+anyhow = "1.0.79"
+flate2 = "1.0.28"
+http = "0.2.11"
 meilisearch-auth = { path = "../meilisearch-auth" }
 meilisearch-types = { path = "../meilisearch-types" }
-once_cell = "1.17.1"
-regex = "1.7.3"
-roaring = { version = "0.10.1", features = ["serde"] }
-serde = { version = "1.0.160", features = ["derive"] }
-serde_json = { version = "1.0.95", features = ["preserve_order"] }
-tar = "0.4.38"
-tempfile = "3.5.0"
-thiserror = "1.0.40"
-time = { version = "0.3.20", features = ["serde-well-known", "formatting", "parsing", "macros"] }
-uuid = { version = "1.3.1", features = ["serde", "v4"] }
+once_cell = "1.19.0"
+regex = "1.10.2"
+roaring = { version = "0.10.2", features = ["serde"] }
+serde = { version = "1.0.195", features = ["derive"] }
+serde_json = { version = "1.0.111", features = ["preserve_order"] }
+tar = "0.4.40"
+tempfile = "3.9.0"
+thiserror = "1.0.56"
+time = { version = "0.3.31", features = ["serde-well-known", "formatting", "parsing", "macros"] }
+tracing = "0.1.40"
+uuid = { version = "1.6.1", features = ["serde", "v4"] }
 
 [dev-dependencies]
 big_s = "1.0.2"

dump/src/reader/compat/v1_to_v2.rs

@@ -120,7 +120,7 @@ impl From<v1::settings::Settings> for v2::Settings<v2::Unchecked> {
                                 criterion.as_ref().map(ToString::to_string)
                             }
                             Err(()) => {
-                                log::warn!(
+                                tracing::warn!(
                                     "Could not import the following ranking rule: `{}`.",
                                     ranking_rule
                                 );
@@ -152,11 +152,11 @@ impl From<v1::update::UpdateStatus> for Option<v2::updates::UpdateStatus> {
         use v2::updates::UpdateStatus as UpdateStatusV2;
         Some(match source {
             UpdateStatusV1::Enqueued { content } => {
-                log::warn!(
+                tracing::warn!(
                     "Cannot import task {} (importing enqueued tasks from v1 dumps is unsupported)",
                     content.update_id
                 );
-                log::warn!("Task will be skipped in the queue of imported tasks.");
+                tracing::warn!("Task will be skipped in the queue of imported tasks.");
 
                 return None;
             }
@@ -229,7 +229,7 @@ impl From<v1::update::UpdateType> for Option<v2::updates::UpdateMeta> {
         Some(match source {
             v1::update::UpdateType::ClearAll => v2::updates::UpdateMeta::ClearDocuments,
             v1::update::UpdateType::Customs => {
-                log::warn!("Ignoring task with type 'Customs' that is no longer supported");
+                tracing::warn!("Ignoring task with type 'Customs' that is no longer supported");
                 return None;
             }
             v1::update::UpdateType::DocumentsAddition { .. } => {
@@ -296,7 +296,7 @@ impl From<v1::settings::RankingRule> for Option<v2::settings::Criterion> {
             v1::settings::RankingRule::Proximity => Some(v2::settings::Criterion::Proximity),
             v1::settings::RankingRule::Attribute => Some(v2::settings::Criterion::Attribute),
             v1::settings::RankingRule::WordsPosition => {
-                log::warn!("Removing the 'WordsPosition' ranking rule that is no longer supported, please check the resulting ranking rules of your indexes");
+                tracing::warn!("Removing the 'WordsPosition' ranking rule that is no longer supported, please check the resulting ranking rules of your indexes");
                 None
             }
             v1::settings::RankingRule::Exactness => Some(v2::settings::Criterion::Exactness),

dump/src/reader/compat/v2_to_v3.rs

@@ -146,8 +146,8 @@ impl From<v2::updates::UpdateStatus> for v3::updates::UpdateStatus {
                         started_processing_at: processing.started_processing_at,
                     }),
                     Err(e) => {
-                        log::warn!("Error with task {}: {}", processing.from.update_id, e);
-                        log::warn!("Task will be marked as `Failed`.");
+                        tracing::warn!("Error with task {}: {}", processing.from.update_id, e);
+                        tracing::warn!("Task will be marked as `Failed`.");
                         v3::updates::UpdateStatus::Failed(v3::updates::Failed {
                             from: v3::updates::Processing {
                                 from: v3::updates::Enqueued {
@@ -172,8 +172,8 @@ impl From<v2::updates::UpdateStatus> for v3::updates::UpdateStatus {
                         enqueued_at: enqueued.enqueued_at,
                     }),
                     Err(e) => {
-                        log::warn!("Error with task {}: {}", enqueued.update_id, e);
-                        log::warn!("Task will be marked as `Failed`.");
+                        tracing::warn!("Error with task {}: {}", enqueued.update_id, e);
+                        tracing::warn!("Task will be marked as `Failed`.");
                         v3::updates::UpdateStatus::Failed(v3::updates::Failed {
                             from: v3::updates::Processing {
                                 from: v3::updates::Enqueued {
@@ -353,7 +353,7 @@ impl From<String> for v3::Code {
             "malformed_payload" => v3::Code::MalformedPayload,
             "missing_payload" => v3::Code::MissingPayload,
             other => {
-                log::warn!("Unknown error code {}", other);
+                tracing::warn!("Unknown error code {}", other);
                 v3::Code::UnretrievableErrorCode
             }
         }

dump/src/reader/compat/v3_to_v4.rs

@@ -76,20 +76,20 @@ impl CompatV3ToV4 {
                         let index_uid = match index_uid {
                             Some(uid) => uid,
                             None => {
-                                log::warn!(
+                                tracing::warn!(
                                     "Error while importing the update {}.",
                                     task.update.id()
                                 );
-                                log::warn!(
+                                tracing::warn!(
                                     "The index associated to the uuid `{}` could not be retrieved.",
                                     task.uuid.to_string()
                                 );
                                 if task.update.is_finished() {
                                     // we're fucking with his history but not his data, that's ok-ish.
-                                    log::warn!("The index-uuid will be set as `unknown`.");
+                                    tracing::warn!("The index-uuid will be set as `unknown`.");
                                     String::from("unknown")
                                 } else {
-                                    log::warn!("The task will be ignored.");
+                                    tracing::warn!("The task will be ignored.");
                                     return None;
                                 }
                             }

dump/src/reader/compat/v4_to_v5.rs

@@ -305,7 +305,7 @@ impl From<v4::ResponseError> for v5::ResponseError {
             "invalid_api_key_expires_at" => v5::Code::InvalidApiKeyExpiresAt,
             "invalid_api_key_description" => v5::Code::InvalidApiKeyDescription,
             other => {
-                log::warn!("Unknown error code {}", other);
+                tracing::warn!("Unknown error code {}", other);
                 v5::Code::UnretrievableErrorCode
             }
         };

dump/src/reader/compat/v5_to_v6.rs

@@ -304,7 +304,7 @@ impl From<v5::ResponseError> for v6::ResponseError {
             "immutable_field" => v6::Code::BadRequest,
             "api_key_already_exists" => v6::Code::ApiKeyAlreadyExists,
             other => {
-                log::warn!("Unknown error code {}", other);
+                tracing::warn!("Unknown error code {}", other);
                 v6::Code::UnretrievableErrorCode
             }
         };
@@ -329,7 +329,7 @@ impl<T> From<v5::Settings<T>> for v6::Settings<v6::Unchecked> {
                                     new_ranking_rules.push(new_rule);
                                 }
                                 Err(_) => {
-                                    log::warn!("Error while importing settings. The ranking rule `{rule}` does not exist anymore.")
+                                    tracing::warn!("Error while importing settings. The ranking rule `{rule}` does not exist anymore.")
                                 }
                             }
                         }

dump/src/reader/v3/errors.rs

@@ -1,5 +1,6 @@
 use serde::{Deserialize, Serialize};
 
+#[allow(clippy::enum_variant_names)]
 #[derive(Serialize, Deserialize, Debug, Clone, Copy)]
 pub enum Code {
     // index related error

dump/src/reader/v4/errors.rs

@@ -95,6 +95,7 @@ impl fmt::Display for ErrorType {
     }
 }
 
+#[allow(clippy::enum_variant_names)]
 #[derive(Serialize, Deserialize, Debug, Clone, Copy)]
 pub enum Code {
     // index related error

dump/src/reader/v5/errors.rs

@@ -31,6 +31,7 @@ impl ResponseError {
     }
 }
 
+#[allow(clippy::enum_variant_names)]
 #[derive(Deserialize, Debug, Clone, Copy)]
 #[cfg_attr(test, derive(serde::Serialize))]
 pub enum Code {

dump/src/reader/v6/mod.rs

@@ -2,10 +2,10 @@ use std::fs::{self, File};
 use std::io::{BufRead, BufReader, ErrorKind};
 use std::path::Path;
 
-use log::debug;
 pub use meilisearch_types::milli;
 use tempfile::TempDir;
 use time::OffsetDateTime;
+use tracing::debug;
 use uuid::Uuid;
 
 use super::Document;

file-store/Cargo.toml

@@ -11,9 +11,9 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-tempfile = "3.5.0"
-thiserror = "1.0.40"
-uuid = { version = "1.3.1", features = ["serde", "v4"] }
+tempfile = "3.9.0"
+thiserror = "1.0.56"
+uuid = { version = "1.6.1", features = ["serde", "v4"] }
 
 [dev-dependencies]
-faux = "0.1.9"
+faux = "0.1.10"

filter-parser/Cargo.toml

@@ -13,8 +13,8 @@ license.workspace = true
 
 [dependencies]
 nom = "7.1.3"
-nom_locate = "4.1.0"
-unescaper = "0.1.2"
+nom_locate = "4.2.0"
+unescaper = "0.1.3"
 
 [dev-dependencies]
-insta = "1.29.0"
+insta = "1.34.0"

fuzzers/Cargo.toml

@@ -11,10 +11,10 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-arbitrary = { version = "1.3.0", features = ["derive"] }
-clap = { version = "4.3.0", features = ["derive"] }
-fastrand = "2.0.0"
+arbitrary = { version = "1.3.2", features = ["derive"] }
+clap = { version = "4.4.17", features = ["derive"] }
+fastrand = "2.0.1"
 milli = { path = "../milli" }
-serde = { version = "1.0.160", features = ["derive"] }
-serde_json = { version = "1.0.95", features = ["preserve_order"] }
-tempfile = "3.5.0"
+serde = { version = "1.0.195", features = ["derive"] }
+serde_json = { version = "1.0.111", features = ["preserve_order"] }
+tempfile = "3.9.0"

index-scheduler/Cargo.toml

@@ -11,32 +11,36 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-anyhow = "1.0.70"
+anyhow = "1.0.79"
 bincode = "1.3.3"
-csv = "1.2.1"
+csv = "1.3.0"
 derive_builder = "0.12.0"
 dump = { path = "../dump" }
-enum-iterator = "1.4.0"
+enum-iterator = "1.5.0"
 file-store = { path = "../file-store" }
 flate2 = "1.0.28"
-log = "0.4.17"
 meilisearch-auth = { path = "../meilisearch-auth" }
 meilisearch-types = { path = "../meilisearch-types" }
 page_size = "0.5.0"
 puffin = { version = "0.16.0", features = ["serialization"] }
-roaring = { version = "0.10.1", features = ["serde"] }
-serde = { version = "1.0.160", features = ["derive"] }
-serde_json = { version = "1.0.95", features = ["preserve_order"] }
+roaring = { version = "0.10.2", features = ["serde"] }
+serde = { version = "1.0.195", features = ["derive"] }
+serde_json = { version = "1.0.111", features = ["preserve_order"] }
 synchronoise = "1.0.1"
-tempfile = "3.5.0"
-thiserror = "1.0.40"
-time = { version = "0.3.20", features = ["serde-well-known", "formatting", "parsing", "macros"] }
+tempfile = "3.9.0"
+thiserror = "1.0.56"
+time = { version = "0.3.31", features = [
+    "serde-well-known",
+    "formatting",
+    "parsing",
+    "macros",
+] }
+tracing = "0.1.40"
 ureq = "2.9.1"
-uuid = { version = "1.3.1", features = ["serde", "v4"] }
+uuid = { version = "1.6.1", features = ["serde", "v4"] }
 
 [dev-dependencies]
 big_s = "1.0.2"
-crossbeam = "0.8.2"
-insta = { version = "1.29.0", features = ["json", "redactions"] }
+crossbeam = "0.8.4"
+insta = { version = "1.34.0", features = ["json", "redactions"] }
 meili-snap = { path = "../meili-snap" }
-nelson = { git = "https://github.com/meilisearch/nelson.git", rev = "675f13885548fb415ead8fbb447e9e6d9314000a"}

index-scheduler/src/batch.rs

@@ -24,7 +24,6 @@ use std::fs::{self, File};
 use std::io::BufWriter;
 
 use dump::IndexMetadata;
-use log::{debug, error, info, trace};
 use meilisearch_types::error::Code;
 use meilisearch_types::heed::{RoTxn, RwTxn};
 use meilisearch_types::milli::documents::{obkv_to_object, DocumentsBatchReader};
@@ -514,6 +513,7 @@ impl IndexScheduler {
     /// 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.
+    #[tracing::instrument(level = "trace", skip(self, rtxn), target = "indexing::scheduler")]
     pub(crate) fn create_next_batch(&self, rtxn: &RoTxn) -> Result<Option<Batch>> {
         #[cfg(test)]
         self.maybe_fail(crate::tests::FailureLocation::InsideCreateBatch)?;
@@ -619,6 +619,7 @@ impl IndexScheduler {
     /// The list of tasks that were processed. The metadata of each task in the returned
     /// list is updated accordingly, with the exception of the its date fields
     /// [`finished_at`](meilisearch_types::tasks::Task::finished_at) and [`started_at`](meilisearch_types::tasks::Task::started_at).
+    #[tracing::instrument(level = "trace", skip(self, batch), target = "indexing::scheduler", fields(batch=batch.to_string()))]
     pub(crate) fn process_batch(&self, batch: Batch) -> Result<Vec<Task>> {
         #[cfg(test)]
         {
@@ -668,9 +669,10 @@ impl IndexScheduler {
                     Ok(()) => {
                         for content_uuid in canceled_tasks_content_uuids {
                             if let Err(error) = self.delete_update_file(content_uuid) {
-                                error!(
-                                    "We failed deleting the content file indentified as {}: {}",
-                                    content_uuid, error
+                                tracing::error!(
+                                    file_content_uuid = %content_uuid,
+                                    %error,
+                                    "Failed deleting content file"
                                 )
                             }
                         }
@@ -969,7 +971,10 @@ impl IndexScheduler {
 
                 match res {
                     Ok(_) => (),
-                    Err(e) => error!("Could not write the stats of the index {}", e),
+                    Err(e) => tracing::error!(
+                        error = &e as &dyn std::error::Error,
+                        "Could not write the stats of the index"
+                    ),
                 }
 
                 Ok(tasks)
@@ -997,7 +1002,7 @@ impl IndexScheduler {
                     builder.set_primary_key(primary_key);
                     let must_stop_processing = self.must_stop_processing.clone();
                     builder.execute(
-                        |indexing_step| debug!("update: {:?}", indexing_step),
+                        |indexing_step| tracing::debug!(update = ?indexing_step),
                         || must_stop_processing.get(),
                     )?;
                     index_wtxn.commit()?;
@@ -1024,7 +1029,10 @@ impl IndexScheduler {
 
                 match res {
                     Ok(_) => (),
-                    Err(e) => error!("Could not write the stats of the index {}", e),
+                    Err(e) => tracing::error!(
+                        error = &e as &dyn std::error::Error,
+                        "Could not write the stats of the index"
+                    ),
                 }
 
                 Ok(vec![task])
@@ -1143,6 +1151,11 @@ impl IndexScheduler {
     ///
     /// ## Return
     /// The list of processed tasks.
+    #[tracing::instrument(
+        level = "trace",
+        skip(self, index_wtxn, index),
+        target = "indexing::scheduler"
+    )]
     fn apply_index_operation<'i>(
         &self,
         index_wtxn: &mut RwTxn<'i>,
@@ -1203,7 +1216,7 @@ impl IndexScheduler {
                                 milli::update::Settings::new(index_wtxn, index, indexer_config);
                             builder.set_primary_key(primary_key);
                             builder.execute(
-                                |indexing_step| debug!("update: {:?}", indexing_step),
+                                |indexing_step| tracing::debug!(update = ?indexing_step),
                                 || must_stop_processing.clone().get(),
                             )?;
                             primary_key_has_been_set = true;
@@ -1222,7 +1235,7 @@ impl IndexScheduler {
                     index,
                     indexer_config,
                     config,
-                    |indexing_step| trace!("update: {:?}", indexing_step),
+                    |indexing_step| tracing::trace!(?indexing_step, "Update"),
                     || must_stop_processing.get(),
                 )?;
 
@@ -1294,7 +1307,7 @@ impl IndexScheduler {
 
                 if !tasks.iter().all(|res| res.error.is_some()) {
                     let addition = builder.execute()?;
-                    info!("document addition done: {:?}", addition);
+                    tracing::info!(indexing_result = ?addition, "document indexing done");
                 } else if primary_key_has_been_set {
                     // Everything failed but we've set a primary key.
                     // We need to remove it.
@@ -1302,7 +1315,7 @@ impl IndexScheduler {
                         milli::update::Settings::new(index_wtxn, index, indexer_config);
                     builder.reset_primary_key();
                     builder.execute(
-                        |indexing_step| trace!("update: {:?}", indexing_step),
+                        |indexing_step| tracing::trace!(update = ?indexing_step),
                         || must_stop_processing.clone().get(),
                     )?;
                 }
@@ -1372,7 +1385,7 @@ impl IndexScheduler {
 
                 let must_stop_processing = self.must_stop_processing.clone();
                 builder.execute(
-                    |indexing_step| debug!("update: {:?}", indexing_step),
+                    |indexing_step| tracing::debug!(update = ?indexing_step),
                     || must_stop_processing.get(),
                 )?;
 
@@ -1584,7 +1597,7 @@ fn delete_document_by_filter<'a>(
             index,
             indexer_config,
             config,
-            |indexing_step| debug!("update: {:?}", indexing_step),
+            |indexing_step| tracing::debug!(update = ?indexing_step),
             || must_stop_processing.get(),
         )?;
 

index-scheduler/src/features.rs

@@ -30,19 +30,6 @@ impl RoFeatures {
         self.runtime
     }
 
-    pub fn check_score_details(&self) -> Result<()> {
-        if self.runtime.score_details {
-            Ok(())
-        } else {
-            Err(FeatureNotEnabledError {
-                disabled_action: "Computing score details",
-                feature: "score details",
-                issue_link: "https://github.com/meilisearch/product/discussions/674",
-            }
-            .into())
-        }
-    }
-
     pub fn check_metrics(&self) -> Result<()> {
         if self.runtime.metrics {
             Ok(())
@@ -56,6 +43,19 @@ impl RoFeatures {
         }
     }
 
+    pub fn check_logs_route(&self) -> Result<()> {
+        if self.runtime.logs_route {
+            Ok(())
+        } else {
+            Err(FeatureNotEnabledError {
+                disabled_action: "getting logs through the `/logs/stream` route",
+                feature: "logs route",
+                issue_link: "https://github.com/orgs/meilisearch/discussions/721",
+            }
+            .into())
+        }
+    }
+
     pub fn check_vector(&self, disabled_action: &'static str) -> Result<()> {
         if self.runtime.vector_store {
             Ok(())
@@ -94,6 +94,7 @@ impl FeatureData {
             runtime_features_db.get(&txn, EXPERIMENTAL_FEATURES)?.unwrap_or_default();
         let runtime = Arc::new(RwLock::new(RuntimeTogglableFeatures {
             metrics: instance_features.metrics || persisted_features.metrics,
+            logs_route: instance_features.logs_route || persisted_features.logs_route,
             ..persisted_features
         }));
 

index-scheduler/src/index_mapper/mod.rs

@@ -3,13 +3,13 @@ use std::sync::{Arc, RwLock};
 use std::time::Duration;
 use std::{fs, thread};
 
-use log::error;
 use meilisearch_types::heed::types::{SerdeJson, Str};
 use meilisearch_types::heed::{Database, Env, RoTxn, RwTxn};
 use meilisearch_types::milli::update::IndexerConfig;
 use meilisearch_types::milli::{FieldDistribution, Index};
 use serde::{Deserialize, Serialize};
 use time::OffsetDateTime;
+use tracing::error;
 use uuid::Uuid;
 
 use self::index_map::IndexMap;

index-scheduler/src/lib.rs

@@ -535,17 +535,17 @@ impl IndexScheduler {
         let budget = if Self::is_good_heed(tasks_path, DEFAULT_BUDGET) {
             DEFAULT_BUDGET
         } else {
-            log::debug!("determining budget with dichotomic search");
+            tracing::debug!("determining budget with dichotomic search");
             utils::dichotomic_search(DEFAULT_BUDGET / 2, |map_size| {
                 Self::is_good_heed(tasks_path, map_size)
             })
         };
 
-        log::debug!("memmap budget: {budget}B");
+        tracing::debug!("memmap budget: {budget}B");
         let mut budget = budget / 2;
         if task_db_size > (budget / 2) {
             task_db_size = clamp_to_page_size(budget * 2 / 5);
-            log::debug!(
+            tracing::debug!(
                 "Decreasing max size of task DB to {task_db_size}B due to constrained memory space"
             );
         }
@@ -555,13 +555,13 @@ impl IndexScheduler {
         let budget = budget;
         let task_db_size = task_db_size;
 
-        log::debug!("index budget: {budget}B");
+        tracing::debug!("index budget: {budget}B");
         let mut index_count = budget / base_map_size;
         if index_count < 2 {
             // take a bit less than half than the budget to make sure we can always afford to open an index
             let map_size = (budget * 2) / 5;
             // single index of max budget
-            log::debug!("1 index of {map_size}B can be opened simultaneously.");
+            tracing::debug!("1 index of {map_size}B can be opened simultaneously.");
             return IndexBudget { map_size, index_count: 1, task_db_size };
         }
         // give us some space for an additional index when the cache is already full
@@ -570,7 +570,7 @@ impl IndexScheduler {
         if index_count > max_index_count {
             index_count = max_index_count;
         }
-        log::debug!("Up to {index_count} indexes of {base_map_size}B opened simultaneously.");
+        tracing::debug!("Up to {index_count} indexes of {base_map_size}B opened simultaneously.");
         IndexBudget { map_size: base_map_size, index_count, task_db_size }
     }
 
@@ -617,7 +617,7 @@ impl IndexScheduler {
                         Ok(TickOutcome::TickAgain(_)) => (),
                         Ok(TickOutcome::WaitForSignal) => run.wake_up.wait(),
                         Err(e) => {
-                            log::error!("{e}");
+                            tracing::error!("{e}");
                             // Wait one second when an irrecoverable error occurs.
                             if !e.is_recoverable() {
                                 std::thread::sleep(Duration::from_secs(1));
@@ -634,15 +634,15 @@ impl IndexScheduler {
                             let mut file = match File::create(format!("{}.puffin", now)) {
                                 Ok(file) => file,
                                 Err(e) => {
-                                    log::error!("{e}");
+                                    tracing::error!("{e}");
                                     continue;
                                 }
                             };
                             if let Err(e) = frame_view.save_to_writer(&mut file) {
-                                log::error!("{e}");
+                                tracing::error!("{e}");
                             }
                             if let Err(e) = file.sync_all() {
-                                log::error!("{e}");
+                                tracing::error!("{e}");
                             }
                             // We erase this frame view as it is no more useful. We want to
                             // measure the new frames now that we exported the previous ones.
@@ -1190,10 +1190,10 @@ impl IndexScheduler {
                     self.update_task(&mut wtxn, &task)
                         .map_err(|e| Error::TaskDatabaseUpdate(Box::new(e)))?;
                     if let Err(e) = self.delete_persisted_task_data(&task) {
-                        log::error!("Failure to delete the content files associated with task {}. Error: {e}", task.uid);
+                        tracing::error!("Failure to delete the content files associated with task {}. Error: {e}", task.uid);
                     }
                 }
-                log::info!("A batch of tasks was successfully completed.");
+                tracing::info!("A batch of tasks was successfully completed.");
             }
             // If we have an abortion error we must stop the tick here and re-schedule tasks.
             Err(Error::Milli(milli::Error::InternalError(
@@ -1247,7 +1247,7 @@ impl IndexScheduler {
                     self.maybe_fail(tests::FailureLocation::UpdatingTaskAfterProcessBatchFailure)?;
 
                     if let Err(e) = self.delete_persisted_task_data(&task) {
-                        log::error!("Failure to delete the content files associated with task {}. Error: {e}", task.uid);
+                        tracing::error!("Failure to delete the content files associated with task {}. Error: {e}", task.uid);
                     }
                     self.update_task(&mut wtxn, &task)
                         .map_err(|e| Error::TaskDatabaseUpdate(Box::new(e)))?;
@@ -1341,7 +1341,7 @@ impl IndexScheduler {
             };
 
             if let Err(e) = request.send(reader) {
-                log::error!("While sending data to the webhook: {e}");
+                tracing::error!("While sending data to the webhook: {e}");
             }
         }
 
@@ -1367,12 +1367,12 @@ impl IndexScheduler {
         // /!\ the len must be at least 2 or else we might enter an infinite loop where we only delete
         //     the deletion tasks we enqueued ourselves.
         if to_delete.len() < 2 {
-            log::warn!("The task queue is almost full, but no task can be deleted yet.");
+            tracing::warn!("The task queue is almost full, but no task can be deleted yet.");
             // the only thing we can do is hope that the user tasks are going to finish
             return Ok(());
         }
 
-        log::info!(
+        tracing::info!(
             "The task queue is almost full. Deleting the oldest {} finished tasks.",
             to_delete.len()
         );

meili-snap/Cargo.toml

@@ -11,6 +11,6 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-insta = { version = "^1.29.0", features = ["json", "redactions"] }
+insta = { version = "^1.34.0", features = ["json", "redactions"] }
 md5 = "0.7.0"
-once_cell = "1.17"
+once_cell = "1.19"

meilisearch-auth/Cargo.toml

@@ -11,16 +11,16 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-base64 = "0.21.0"
-enum-iterator = "1.4.0"
+base64 = "0.21.7"
+enum-iterator = "1.5.0"
 hmac = "0.12.1"
 maplit = "1.0.2"
 meilisearch-types = { path = "../meilisearch-types" }
 rand = "0.8.5"
-roaring = { version = "0.10.1", features = ["serde"] }
-serde = { version = "1.0.160", features = ["derive"] }
-serde_json = { version = "1.0.95", features = ["preserve_order"] }
-sha2 = "0.10.6"
-thiserror = "1.0.40"
-time = { version = "0.3.20", features = ["serde-well-known", "formatting", "parsing", "macros"] }
-uuid = { version = "1.3.1", features = ["serde", "v4"] }
+roaring = { version = "0.10.2", features = ["serde"] }
+serde = { version = "1.0.195", features = ["derive"] }
+serde_json = { version = "1.0.111", features = ["preserve_order"] }
+sha2 = "0.10.8"
+thiserror = "1.0.56"
+time = { version = "0.3.31", features = ["serde-well-known", "formatting", "parsing", "macros"] }
+uuid = { version = "1.6.1", features = ["serde", "v4"] }

meilisearch-types/Cargo.toml

@@ -11,31 +11,31 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-actix-web = { version = "4.3.1", default-features = false }
-anyhow = "1.0.70"
+actix-web = { version = "4.5.1", default-features = false }
+anyhow = "1.0.79"
 convert_case = "0.6.0"
-csv = "1.2.1"
-deserr = { version = "0.6.0", features = ["actix-web"] }
-either = { version = "1.8.1", features = ["serde"] }
-enum-iterator = "1.4.0"
+csv = "1.3.0"
+deserr = { version = "0.6.1", features = ["actix-web"] }
+either = { version = "1.9.0", features = ["serde"] }
+enum-iterator = "1.5.0"
 file-store = { path = "../file-store" }
-flate2 = "1.0.25"
+flate2 = "1.0.28"
 fst = "0.4.7"
 memmap2 = "0.7.1"
 milli = { path = "../milli" }
-roaring = { version = "0.10.1", features = ["serde"] }
-serde = { version = "1.0.160", features = ["derive"] }
+roaring = { version = "0.10.2", features = ["serde"] }
+serde = { version = "1.0.195", features = ["derive"] }
 serde-cs = "0.2.4"
-serde_json = "1.0.95"
-tar = "0.4.38"
-tempfile = "3.5.0"
-thiserror = "1.0.40"
-time = { version = "0.3.20", features = ["serde-well-known", "formatting", "parsing", "macros"] }
-tokio = "1.27"
-uuid = { version = "1.3.1", features = ["serde", "v4"] }
+serde_json = "1.0.111"
+tar = "0.4.40"
+tempfile = "3.9.0"
+thiserror = "1.0.56"
+time = { version = "0.3.31", features = ["serde-well-known", "formatting", "parsing", "macros"] }
+tokio = "1.35"
+uuid = { version = "1.6.1", features = ["serde", "v4"] }
 
 [dev-dependencies]
-insta = "1.29.0"
+insta = "1.34.0"
 meili-snap = { path = "../meili-snap" }
 
 [features]

meilisearch-types/src/error.rs

@@ -310,6 +310,8 @@ TooManyVectors                        , InvalidRequest       , BAD_REQUEST ;
 UnretrievableDocument                 , Internal             , BAD_REQUEST ;
 UnretrievableErrorCode                , InvalidRequest       , BAD_REQUEST ;
 UnsupportedMediaType                  , InvalidRequest       , UNSUPPORTED_MEDIA_TYPE ;
+
+// Experimental features
 VectorEmbeddingError                  , InvalidRequest       , BAD_REQUEST
 }
 
@@ -347,6 +349,9 @@ impl ErrorCode for milli::Error {
                     UserError::InvalidFieldForSource { .. }
                     | UserError::MissingFieldForSource { .. }
                     | UserError::InvalidOpenAiModel { .. }
+                    | UserError::InvalidOpenAiModelDimensions { .. }
+                    | UserError::InvalidOpenAiModelDimensionsMax { .. }
+                    | UserError::InvalidSettingsDimensions { .. }
                     | UserError::InvalidPrompt(_) => Code::InvalidSettingsEmbedders,
                     UserError::TooManyEmbedders(_) => Code::InvalidSettingsEmbedders,
                     UserError::InvalidPromptForEmbeddings(..) => Code::InvalidSettingsEmbedders,

meilisearch-types/src/features.rs

@@ -3,13 +3,14 @@ use serde::{Deserialize, Serialize};
 #[derive(Serialize, Deserialize, Debug, Clone, Copy, Default, PartialEq, Eq)]
 #[serde(rename_all = "camelCase", default)]
 pub struct RuntimeTogglableFeatures {
-    pub score_details: bool,
     pub vector_store: bool,
     pub metrics: bool,
+    pub logs_route: bool,
     pub export_puffin_reports: bool,
 }
 
 #[derive(Default, Debug, Clone, Copy)]
 pub struct InstanceTogglableFeatures {
     pub metrics: bool,
+    pub logs_route: bool,
 }

meilisearch/Cargo.toml

@@ -13,124 +13,126 @@ license.workspace = true
 default-run = "meilisearch"
 
 [dependencies]
-actix-cors = "0.6.4"
-actix-http = { version = "3.3.1", default-features = false, features = [
+actix-cors = "0.7.0"
+actix-http = { version = "3.6.0", default-features = false, features = [
     "compress-brotli",
     "compress-gzip",
-    "rustls",
+    "rustls-0_21",
 ] }
 actix-utils = "3.0.1"
-actix-web = { version = "4.3.1", default-features = false, features = [
+actix-web = { version = "4.5.1", default-features = false, features = [
     "macros",
     "compress-brotli",
     "compress-gzip",
     "cookies",
-    "rustls",
+    "rustls-0_21",
 ] }
 actix-web-static-files = { git = "https://github.com/kilork/actix-web-static-files.git", rev = "2d3b6160", optional = true }
-anyhow = { version = "1.0.70", features = ["backtrace"] }
+anyhow = { version = "1.0.79", features = ["backtrace"] }
 async-stream = "0.3.5"
-async-trait = "0.1.68"
-bstr = "1.4.0"
+async-trait = "0.1.77"
+bstr = "1.9.0"
 byte-unit = { version = "4.0.19", default-features = false, features = [
     "std",
     "serde",
 ] }
-bytes = "1.4.0"
-clap = { version = "4.2.1", features = ["derive", "env"] }
-crossbeam-channel = "0.5.8"
-deserr = { version = "0.6.0", features = ["actix-web"] }
+bytes = "1.5.0"
+clap = { version = "4.4.17", features = ["derive", "env"] }
+crossbeam-channel = "0.5.11"
+deserr = { version = "0.6.1", features = ["actix-web"] }
 dump = { path = "../dump" }
-either = "1.8.1"
-env_logger = "0.10.0"
+either = "1.9.0"
 file-store = { path = "../file-store" }
-flate2 = "1.0.25"
+flate2 = "1.0.28"
 fst = "0.4.7"
-futures = "0.3.28"
-futures-util = "0.3.28"
-http = "0.2.9"
+futures = "0.3.30"
+futures-util = "0.3.30"
+http = "0.2.11"
 index-scheduler = { path = "../index-scheduler" }
-indexmap = { version = "2.0.0", features = ["serde"] }
-is-terminal = "0.4.8"
+indexmap = { version = "2.1.0", features = ["serde"] }
+is-terminal = "0.4.10"
 itertools = "0.11.0"
-jsonwebtoken = "8.3.0"
+jsonwebtoken = "9.2.0"
 lazy_static = "1.4.0"
-log = "0.4.17"
 meilisearch-auth = { path = "../meilisearch-auth" }
 meilisearch-types = { path = "../meilisearch-types" }
-mimalloc = { version = "0.1.37", default-features = false }
+mimalloc = { version = "0.1.39", default-features = false }
 mime = "0.3.17"
-num_cpus = "1.15.0"
-obkv = "0.2.0"
-once_cell = "1.17.1"
-ordered-float = "3.7.0"
+num_cpus = "1.16.0"
+obkv = "0.2.1"
+once_cell = "1.19.0"
+ordered-float = "4.2.0"
 parking_lot = "0.12.1"
 permissive-json-pointer = { path = "../permissive-json-pointer" }
-pin-project-lite = "0.2.9"
+pin-project-lite = "0.2.13"
 platform-dirs = "0.3.0"
 prometheus = { version = "0.13.3", features = ["process"] }
 puffin = { version = "0.16.0", features = ["serialization"] }
 rand = "0.8.5"
-rayon = "1.7.0"
-regex = "1.7.3"
-reqwest = { version = "0.11.16", features = [
+rayon = "1.8.0"
+regex = "1.10.2"
+reqwest = { version = "0.11.23", features = [
     "rustls-tls",
     "json",
 ], default-features = false }
-rustls = "0.20.8"
+rustls = "0.21.6"
 rustls-pemfile = "1.0.2"
-segment = { version = "0.2.2", optional = true }
-serde = { version = "1.0.160", features = ["derive"] }
-serde_json = { version = "1.0.95", features = ["preserve_order"] }
-sha2 = "0.10.6"
-siphasher = "0.3.10"
-slice-group-by = "0.3.0"
+segment = { version = "0.2.3", optional = true }
+serde = { version = "1.0.195", features = ["derive"] }
+serde_json = { version = "1.0.111", features = ["preserve_order"] }
+sha2 = "0.10.8"
+siphasher = "1.0.0"
+slice-group-by = "0.3.1"
 static-files = { version = "0.2.3", optional = true }
-sysinfo = "0.29.7"
-tar = "0.4.38"
-tempfile = "3.5.0"
-thiserror = "1.0.40"
-time = { version = "0.3.20", features = [
+sysinfo = "0.30.5"
+tar = "0.4.40"
+tempfile = "3.9.0"
+thiserror = "1.0.56"
+time = { version = "0.3.31", features = [
     "serde-well-known",
     "formatting",
     "parsing",
     "macros",
 ] }
-tokio = { version = "1.27.0", features = ["full"] }
-tokio-stream = "0.1.12"
-toml = "0.7.3"
-uuid = { version = "1.3.1", features = ["serde", "v4"] }
-walkdir = "2.3.3"
+tokio = { version = "1.35.1", features = ["full"] }
+tokio-stream = "0.1.14"
+toml = "0.8.8"
+uuid = { version = "1.6.1", features = ["serde", "v4"] }
+walkdir = "2.4.0"
 yaup = "0.2.1"
 serde_urlencoded = "0.7.1"
-termcolor = "1.2.0"
+termcolor = "1.4.1"
 url = { version = "2.5.0", features = ["serde"] }
+tracing = "0.1.40"
+tracing-subscriber = "0.3.18"
+tracing-trace = { version = "0.1.0", path = "../tracing-trace" }
+tracing-actix-web = "0.7.9"
 
 [dev-dependencies]
-actix-rt = "2.8.0"
+actix-rt = "2.9.0"
 assert-json-diff = "2.0.2"
-brotli = "3.3.4"
-insta = "1.29.0"
-manifest-dir-macros = "0.1.16"
+brotli = "3.4.0"
+insta = "1.34.0"
+manifest-dir-macros = "0.1.18"
 maplit = "1.0.2"
 meili-snap = { path = "../meili-snap" }
-temp-env = "0.3.3"
-urlencoding = "2.1.2"
+temp-env = "0.3.6"
+urlencoding = "2.1.3"
 yaup = "0.2.1"
 
 [build-dependencies]
-anyhow = { version = "1.0.70", optional = true }
-cargo_toml = { version = "0.15.2", optional = true }
+anyhow = { version = "1.0.79", optional = true }
+cargo_toml = { version = "0.18.0", optional = true }
 hex = { version = "0.4.3", optional = true }
-reqwest = { version = "0.11.16", features = [
+reqwest = { version = "0.11.23", features = [
     "blocking",
     "rustls-tls",
 ], default-features = false, optional = true }
 sha-1 = { version = "0.10.1", optional = true }
 static-files = { version = "0.2.3", optional = true }
-tempfile = { version = "3.5.0", optional = true }
+tempfile = { version = "3.9.0", optional = true }
 vergen = { version = "7.5.1", default-features = false, features = ["git"] }
-zip = { version = "0.6.4", optional = true }
+zip = { version = "0.6.6", optional = true }
 
 [features]
 default = ["analytics", "meilisearch-types/all-tokenizations", "mini-dashboard"]
@@ -154,5 +156,5 @@ greek = ["meilisearch-types/greek"]
 khmer = ["meilisearch-types/khmer"]
 
 [package.metadata.mini-dashboard]
-assets-url = "https://github.com/meilisearch/mini-dashboard/releases/download/v0.2.12/build.zip"
-sha1 = "acfe9a018c93eb0604ea87ee87bff7df5474e18e"
+assets-url = "https://github.com/meilisearch/mini-dashboard/releases/download/v0.2.13/build.zip"
+sha1 = "e20cc9b390003c6c844f4b8bcc5c5013191a77ff"

meilisearch/src/analytics/segment_analytics.rs

@@ -18,7 +18,7 @@ use segment::message::{Identify, Track, User};
 use segment::{AutoBatcher, Batcher, HttpClient};
 use serde::Serialize;
 use serde_json::{json, Value};
-use sysinfo::{DiskExt, System, SystemExt};
+use sysinfo::{Disks, System};
 use time::OffsetDateTime;
 use tokio::select;
 use tokio::sync::mpsc::{self, Receiver, Sender};
@@ -250,6 +250,7 @@ impl super::Analytics for SegmentAnalytics {
 struct Infos {
     env: String,
     experimental_enable_metrics: bool,
+    experimental_enable_logs_route: bool,
     experimental_reduce_indexing_memory_usage: bool,
     experimental_max_number_of_batched_tasks: usize,
     db_path: bool,
@@ -287,6 +288,7 @@ impl From<Opt> for Infos {
         let Opt {
             db_path,
             experimental_enable_metrics,
+            experimental_enable_logs_route,
             experimental_reduce_indexing_memory_usage,
             experimental_max_number_of_batched_tasks,
             http_addr,
@@ -333,6 +335,7 @@ impl From<Opt> for Infos {
         Self {
             env,
             experimental_enable_metrics,
+            experimental_enable_logs_route,
             experimental_reduce_indexing_memory_usage,
             db_path: db_path != PathBuf::from("./data.ms"),
             import_dump: import_dump.is_some(),
@@ -386,16 +389,17 @@ impl Segment {
     fn compute_traits(opt: &Opt, stats: Stats) -> Value {
         static FIRST_START_TIMESTAMP: Lazy<Instant> = Lazy::new(Instant::now);
         static SYSTEM: Lazy<Value> = Lazy::new(|| {
+            let disks = Disks::new_with_refreshed_list();
             let mut sys = System::new_all();
             sys.refresh_all();
-            let kernel_version =
-                sys.kernel_version().and_then(|k| k.split_once('-').map(|(k, _)| k.to_string()));
+            let kernel_version = System::kernel_version()
+                .and_then(|k| k.split_once('-').map(|(k, _)| k.to_string()));
             json!({
-                    "distribution": sys.name(),
+                    "distribution": System::name(),
                     "kernel_version": kernel_version,
                     "cores": sys.cpus().len(),
                     "ram_size": sys.total_memory(),
-                    "disk_size": sys.disks().iter().map(|disk| disk.total_space()).max(),
+                    "disk_size": disks.iter().map(|disk| disk.total_space()).max(),
                     "server_provider": std::env::var("MEILI_SERVER_PROVIDER").ok(),
             })
         });

meilisearch/src/error.rs

@@ -12,6 +12,8 @@ pub enum MeilisearchHttpError {
     #[error("A Content-Type header is missing. Accepted values for the Content-Type header are: {}",
             .0.iter().map(|s| format!("`{}`", s)).collect::<Vec<_>>().join(", "))]
     MissingContentType(Vec<String>),
+    #[error("The `/logs/stream` route is currently in use by someone else.")]
+    AlreadyUsedLogRoute,
     #[error("The Content-Type `{0}` does not support the use of a csv delimiter. The csv delimiter can only be used with the Content-Type `text/csv`.")]
     CsvDelimiterWithWrongContentType(String),
     #[error(
@@ -59,6 +61,7 @@ impl ErrorCode for MeilisearchHttpError {
     fn error_code(&self) -> Code {
         match self {
             MeilisearchHttpError::MissingContentType(_) => Code::MissingContentType,
+            MeilisearchHttpError::AlreadyUsedLogRoute => Code::BadRequest,
             MeilisearchHttpError::CsvDelimiterWithWrongContentType(_) => Code::InvalidContentType,
             MeilisearchHttpError::MissingPayload(_) => Code::MissingPayload,
             MeilisearchHttpError::InvalidContentType(_, _) => Code::InvalidContentType,

meilisearch/src/lib.rs

@@ -29,7 +29,6 @@ use error::PayloadError;
 use extractors::payload::PayloadConfig;
 use http::header::CONTENT_TYPE;
 use index_scheduler::{IndexScheduler, IndexSchedulerOptions};
-use log::error;
 use meilisearch_auth::AuthController;
 use meilisearch_types::milli::documents::{DocumentsBatchBuilder, DocumentsBatchReader};
 use meilisearch_types::milli::update::{IndexDocumentsConfig, IndexDocumentsMethod};
@@ -39,6 +38,8 @@ use meilisearch_types::versioning::{check_version_file, create_version_file};
 use meilisearch_types::{compression, milli, VERSION_FILE_NAME};
 pub use option::Opt;
 use option::ScheduleSnapshot;
+use tracing::{error, info_span};
+use tracing_subscriber::filter::Targets;
 
 use crate::error::MeilisearchHttpError;
 
@@ -86,10 +87,21 @@ fn is_empty_db(db_path: impl AsRef<Path>) -> bool {
     }
 }
 
+/// The handle used to update the logs at runtime. Must be accessible from the `main.rs` and the `route/logs.rs`.
+pub type LogRouteHandle =
+    tracing_subscriber::reload::Handle<LogRouteType, tracing_subscriber::Registry>;
+
+pub type LogRouteType = tracing_subscriber::filter::Filtered<
+    Option<Box<dyn tracing_subscriber::Layer<tracing_subscriber::Registry> + Send + Sync>>,
+    Targets,
+    tracing_subscriber::Registry,
+>;
+
 pub fn create_app(
     index_scheduler: Data<IndexScheduler>,
     auth_controller: Data<AuthController>,
     opt: Opt,
+    logs: LogRouteHandle,
     analytics: Arc<dyn Analytics>,
     enable_dashboard: bool,
 ) -> actix_web::App<
@@ -108,6 +120,7 @@ pub fn create_app(
                 index_scheduler.clone(),
                 auth_controller.clone(),
                 &opt,
+                logs,
                 analytics.clone(),
             )
         })
@@ -123,11 +136,49 @@ pub fn create_app(
             .allow_any_method()
             .max_age(86_400), // 24h
     )
-    .wrap(actix_web::middleware::Logger::default())
+    .wrap(tracing_actix_web::TracingLogger::<AwebTracingLogger>::new())
     .wrap(actix_web::middleware::Compress::default())
     .wrap(actix_web::middleware::NormalizePath::new(actix_web::middleware::TrailingSlash::Trim))
 }
 
+struct AwebTracingLogger;
+
+impl tracing_actix_web::RootSpanBuilder for AwebTracingLogger {
+    fn on_request_start(request: &actix_web::dev::ServiceRequest) -> tracing::Span {
+        use tracing::field::Empty;
+
+        let conn_info = request.connection_info();
+        let headers = request.headers();
+        let user_agent = headers
+            .get(http::header::USER_AGENT)
+            .map(|value| String::from_utf8_lossy(value.as_bytes()).into_owned())
+            .unwrap_or_default();
+        info_span!("HTTP request", method = %request.method(), host = conn_info.host(), route = %request.path(), query_parameters = %request.query_string(), %user_agent, status_code = Empty, error = Empty)
+    }
+
+    fn on_request_end<B: MessageBody>(
+        span: tracing::Span,
+        outcome: &Result<ServiceResponse<B>, actix_web::Error>,
+    ) {
+        match &outcome {
+            Ok(response) => {
+                let code: i32 = response.response().status().as_u16().into();
+                span.record("status_code", code);
+
+                if let Some(error) = response.response().error() {
+                    // use the status code already constructed for the outgoing HTTP response
+                    span.record("error", &tracing::field::display(error.as_response_error()));
+                }
+            }
+            Err(error) => {
+                let code: i32 = error.error_response().status().as_u16().into();
+                span.record("status_code", code);
+                span.record("error", &tracing::field::display(error.as_response_error()));
+            }
+        };
+    }
+}
+
 enum OnFailure {
     RemoveDb,
     KeepDb,
@@ -280,15 +331,15 @@ fn import_dump(
     let mut dump_reader = dump::DumpReader::open(reader)?;
 
     if let Some(date) = dump_reader.date() {
-        log::info!(
-            "Importing a dump of meilisearch `{:?}` from the {}",
-            dump_reader.version(), // TODO: get the meilisearch version instead of the dump version
-            date
+        tracing::info!(
+            version = ?dump_reader.version(), // TODO: get the meilisearch version instead of the dump version
+            %date,
+            "Importing a dump of meilisearch"
         );
     } else {
-        log::info!(
-            "Importing a dump of meilisearch `{:?}`",
-            dump_reader.version(), // TODO: get the meilisearch version instead of the dump version
+        tracing::info!(
+            version = ?dump_reader.version(), // TODO: get the meilisearch version instead of the dump version
+            "Importing a dump of meilisearch",
         );
     }
 
@@ -322,7 +373,7 @@ fn import_dump(
     for index_reader in dump_reader.indexes()? {
         let mut index_reader = index_reader?;
         let metadata = index_reader.metadata();
-        log::info!("Importing index `{}`.", metadata.uid);
+        tracing::info!("Importing index `{}`.", metadata.uid);
 
         let date = Some((metadata.created_at, metadata.updated_at));
         let index = index_scheduler.create_raw_index(&metadata.uid, date)?;
@@ -336,14 +387,15 @@ fn import_dump(
         }
 
         // 4.2 Import the settings.
-        log::info!("Importing the settings.");
+        tracing::info!("Importing the settings.");
         let settings = index_reader.settings()?;
         apply_settings_to_builder(&settings, &mut builder);
-        builder.execute(|indexing_step| log::debug!("update: {:?}", indexing_step), || false)?;
+        builder
+            .execute(|indexing_step| tracing::debug!("update: {:?}", indexing_step), || false)?;
 
         // 4.3 Import the documents.
         // 4.3.1 We need to recreate the grenad+obkv format accepted by the index.
-        log::info!("Importing the documents.");
+        tracing::info!("Importing the documents.");
         let file = tempfile::tempfile()?;
         let mut builder = DocumentsBatchBuilder::new(BufWriter::new(file));
         for document in index_reader.documents()? {
@@ -365,15 +417,16 @@ fn import_dump(
                 update_method: IndexDocumentsMethod::ReplaceDocuments,
                 ..Default::default()
             },
-            |indexing_step| log::trace!("update: {:?}", indexing_step),
+            |indexing_step| tracing::trace!("update: {:?}", indexing_step),
             || false,
         )?;
 
         let (builder, user_result) = builder.add_documents(reader)?;
-        log::info!("{} documents found.", user_result?);
+        let user_result = user_result?;
+        tracing::info!(documents_found = user_result, "{} documents found.", user_result);
         builder.execute()?;
         wtxn.commit()?;
-        log::info!("All documents successfully imported.");
+        tracing::info!("All documents successfully imported.");
     }
 
     let mut index_scheduler_dump = index_scheduler.register_dumped_task()?;
@@ -391,6 +444,7 @@ pub fn configure_data(
     index_scheduler: Data<IndexScheduler>,
     auth: Data<AuthController>,
     opt: &Opt,
+    logs: LogRouteHandle,
     analytics: Arc<dyn Analytics>,
 ) {
     let http_payload_size_limit = opt.http_payload_size_limit.get_bytes() as usize;
@@ -398,6 +452,7 @@ pub fn configure_data(
         .app_data(index_scheduler)
         .app_data(auth)
         .app_data(web::Data::from(analytics))
+        .app_data(web::Data::new(logs))
         .app_data(
             web::JsonConfig::default()
                 .limit(http_payload_size_limit)

meilisearch/src/main.rs

@@ -1,6 +1,7 @@
 use std::env;
 use std::io::{stderr, Write};
 use std::path::PathBuf;
+use std::str::FromStr;
 use std::sync::Arc;
 
 use actix_web::http::KeepAlive;
@@ -9,37 +10,60 @@ use actix_web::HttpServer;
 use index_scheduler::IndexScheduler;
 use is_terminal::IsTerminal;
 use meilisearch::analytics::Analytics;
-use meilisearch::{analytics, create_app, prototype_name, setup_meilisearch, Opt};
+use meilisearch::{
+    analytics, create_app, prototype_name, setup_meilisearch, LogRouteHandle, LogRouteType, Opt,
+};
 use meilisearch_auth::{generate_master_key, AuthController, MASTER_KEY_MIN_SIZE};
+use mimalloc::MiMalloc;
 use termcolor::{Color, ColorChoice, ColorSpec, StandardStream, WriteColor};
+use tracing::level_filters::LevelFilter;
+use tracing_subscriber::layer::SubscriberExt as _;
+use tracing_subscriber::Layer;
 
 #[global_allocator]
-static ALLOC: mimalloc::MiMalloc = mimalloc::MiMalloc;
+static ALLOC: MiMalloc = MiMalloc;
+
+fn default_layer() -> LogRouteType {
+    None.with_filter(tracing_subscriber::filter::Targets::new().with_target("", LevelFilter::OFF))
+}
 
 /// does all the setup before meilisearch is launched
-fn setup(opt: &Opt) -> anyhow::Result<()> {
-    let mut log_builder = env_logger::Builder::new();
-    let log_filters = format!(
-        "{},h2=warn,hyper=warn,tokio_util=warn,tracing=warn,rustls=warn,mio=warn,reqwest=warn",
-        opt.log_level
+fn setup(opt: &Opt) -> anyhow::Result<LogRouteHandle> {
+    let (route_layer, route_layer_handle) = tracing_subscriber::reload::Layer::new(default_layer());
+    let route_layer: tracing_subscriber::reload::Layer<_, _> = route_layer;
+
+    let subscriber = tracing_subscriber::registry().with(route_layer).with(
+        tracing_subscriber::fmt::layer()
+            .with_span_events(tracing_subscriber::fmt::format::FmtSpan::CLOSE)
+            .with_filter(
+                tracing_subscriber::filter::LevelFilter::from_str(&opt.log_level.to_string())
+                    .unwrap(),
+            ),
     );
-    log_builder.parse_filters(&log_filters);
 
-    log_builder.init();
+    // set the subscriber as the default for the application
+    tracing::subscriber::set_global_default(subscriber).unwrap();
 
-    Ok(())
+    Ok(route_layer_handle)
+}
+
+fn on_panic(info: &std::panic::PanicInfo) {
+    let info = info.to_string().replace('\n', " ");
+    tracing::error!(%info);
 }
 
 #[actix_web::main]
 async fn main() -> anyhow::Result<()> {
     let (opt, config_read_from) = Opt::try_build()?;
 
+    std::panic::set_hook(Box::new(on_panic));
+
     anyhow::ensure!(
         !(cfg!(windows) && opt.experimental_reduce_indexing_memory_usage),
         "The `experimental-reduce-indexing-memory-usage` flag is not supported on Windows"
     );
 
-    setup(&opt)?;
+    let log_handle = setup(&opt)?;
 
     match (opt.env.as_ref(), &opt.master_key) {
         ("production", Some(master_key)) if master_key.len() < MASTER_KEY_MIN_SIZE => {
@@ -77,7 +101,7 @@ async fn main() -> anyhow::Result<()> {
 
     print_launch_resume(&opt, analytics.clone(), config_read_from);
 
-    run_http(index_scheduler, auth_controller, opt, analytics).await?;
+    run_http(index_scheduler, auth_controller, opt, log_handle, analytics).await?;
 
     Ok(())
 }
@@ -86,6 +110,7 @@ async fn run_http(
     index_scheduler: Arc<IndexScheduler>,
     auth_controller: Arc<AuthController>,
     opt: Opt,
+    logs: LogRouteHandle,
     analytics: Arc<dyn Analytics>,
 ) -> anyhow::Result<()> {
     let enable_dashboard = &opt.env == "development";
@@ -98,6 +123,7 @@ async fn run_http(
             index_scheduler.clone(),
             auth_controller.clone(),
             opt.clone(),
+            logs.clone(),
             analytics.clone(),
             enable_dashboard,
         )
@@ -107,7 +133,7 @@ async fn run_http(
     .keep_alive(KeepAlive::Os);
 
     if let Some(config) = opt_clone.get_ssl_config()? {
-        http_server.bind_rustls(opt_clone.http_addr, config)?.run().await?;
+        http_server.bind_rustls_021(opt_clone.http_addr, config)?.run().await?;
     } else {
         http_server.bind(&opt_clone.http_addr)?.run().await?;
     }

meilisearch/src/option.rs

@@ -20,7 +20,7 @@ use rustls::server::{
 use rustls::RootCertStore;
 use rustls_pemfile::{certs, pkcs8_private_keys, rsa_private_keys};
 use serde::{Deserialize, Serialize};
-use sysinfo::{RefreshKind, System, SystemExt};
+use sysinfo::{MemoryRefreshKind, RefreshKind, System};
 use url::Url;
 
 const POSSIBLE_ENV: [&str; 2] = ["development", "production"];
@@ -51,6 +51,7 @@ const MEILI_IGNORE_MISSING_DUMP: &str = "MEILI_IGNORE_MISSING_DUMP";
 const MEILI_IGNORE_DUMP_IF_DB_EXISTS: &str = "MEILI_IGNORE_DUMP_IF_DB_EXISTS";
 const MEILI_DUMP_DIR: &str = "MEILI_DUMP_DIR";
 const MEILI_LOG_LEVEL: &str = "MEILI_LOG_LEVEL";
+const MEILI_EXPERIMENTAL_ENABLE_LOGS_ROUTE: &str = "MEILI_EXPERIMENTAL_ENABLE_LOGS_ROUTE";
 const MEILI_EXPERIMENTAL_ENABLE_METRICS: &str = "MEILI_EXPERIMENTAL_ENABLE_METRICS";
 const MEILI_EXPERIMENTAL_REDUCE_INDEXING_MEMORY_USAGE: &str =
     "MEILI_EXPERIMENTAL_REDUCE_INDEXING_MEMORY_USAGE";
@@ -309,6 +310,13 @@ pub struct Opt {
     #[serde(default)]
     pub experimental_enable_metrics: bool,
 
+    /// Experimental logs route feature. For more information, see: <https://github.com/orgs/meilisearch/discussions/721>
+    ///
+    /// Enables the log route on the `POST /logs/stream` endpoint and the `DELETE /logs/stream` to stop receiving logs.
+    #[clap(long, env = MEILI_EXPERIMENTAL_ENABLE_LOGS_ROUTE)]
+    #[serde(default)]
+    pub experimental_enable_logs_route: bool,
+
     /// Experimental RAM reduction during indexing, do not use in production, see: <https://github.com/meilisearch/product/discussions/652>
     #[clap(long, env = MEILI_EXPERIMENTAL_REDUCE_INDEXING_MEMORY_USAGE)]
     #[serde(default)]
@@ -414,6 +422,7 @@ impl Opt {
             #[cfg(feature = "analytics")]
             no_analytics,
             experimental_enable_metrics,
+            experimental_enable_logs_route,
             experimental_reduce_indexing_memory_usage,
         } = self;
         export_to_env_if_not_present(MEILI_DB_PATH, db_path);
@@ -470,6 +479,10 @@ impl Opt {
             MEILI_EXPERIMENTAL_ENABLE_METRICS,
             experimental_enable_metrics.to_string(),
         );
+        export_to_env_if_not_present(
+            MEILI_EXPERIMENTAL_ENABLE_LOGS_ROUTE,
+            experimental_enable_logs_route.to_string(),
+        );
         export_to_env_if_not_present(
             MEILI_EXPERIMENTAL_REDUCE_INDEXING_MEMORY_USAGE,
             experimental_reduce_indexing_memory_usage.to_string(),
@@ -490,11 +503,11 @@ impl Opt {
                     }
                     if self.ssl_require_auth {
                         let verifier = AllowAnyAuthenticatedClient::new(client_auth_roots);
-                        config.with_client_cert_verifier(verifier)
+                        config.with_client_cert_verifier(Arc::from(verifier))
                     } else {
                         let verifier =
                             AllowAnyAnonymousOrAuthenticatedClient::new(client_auth_roots);
-                        config.with_client_cert_verifier(verifier)
+                        config.with_client_cert_verifier(Arc::from(verifier))
                     }
                 }
                 None => config.with_no_client_auth(),
@@ -524,7 +537,10 @@ impl Opt {
     }
 
     pub(crate) fn to_instance_features(&self) -> InstanceTogglableFeatures {
-        InstanceTogglableFeatures { metrics: self.experimental_enable_metrics }
+        InstanceTogglableFeatures {
+            metrics: self.experimental_enable_metrics,
+            logs_route: self.experimental_enable_logs_route,
+        }
     }
 }
 
@@ -633,8 +649,8 @@ impl MaxMemory {
 
 /// Returns the total amount of bytes available or `None` if this system isn't supported.
 fn total_memory_bytes() -> Option<u64> {
-    if System::IS_SUPPORTED {
-        let memory_kind = RefreshKind::new().with_memory();
+    if sysinfo::IS_SUPPORTED_SYSTEM {
+        let memory_kind = RefreshKind::new().with_memory(MemoryRefreshKind::new().with_ram());
         let mut system = System::new_with_specifics(memory_kind);
         system.refresh_memory();
         Some(system.total_memory())

meilisearch/src/routes/dump.rs

@@ -1,11 +1,11 @@
 use actix_web::web::Data;
 use actix_web::{web, HttpRequest, HttpResponse};
 use index_scheduler::IndexScheduler;
-use log::debug;
 use meilisearch_auth::AuthController;
 use meilisearch_types::error::ResponseError;
 use meilisearch_types::tasks::KindWithContent;
 use serde_json::json;
+use tracing::debug;
 
 use crate::analytics::Analytics;
 use crate::extractors::authentication::policies::*;
@@ -32,6 +32,6 @@ pub async fn create_dump(
     let task: SummarizedTaskView =
         tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
 
-    debug!("returns: {:?}", task);
+    debug!(returns = ?task, "Create dump");
     Ok(HttpResponse::Accepted().json(task))
 }

meilisearch/src/routes/features.rs

@@ -3,11 +3,11 @@ use actix_web::{HttpRequest, HttpResponse};
 use deserr::actix_web::AwebJson;
 use deserr::Deserr;
 use index_scheduler::IndexScheduler;
-use log::debug;
 use meilisearch_types::deserr::DeserrJsonError;
 use meilisearch_types::error::ResponseError;
 use meilisearch_types::keys::actions;
 use serde_json::json;
+use tracing::debug;
 
 use crate::analytics::Analytics;
 use crate::extractors::authentication::policies::ActionPolicy;
@@ -33,20 +33,21 @@ async fn get_features(
     let features = index_scheduler.features();
 
     analytics.publish("Experimental features Seen".to_string(), json!(null), Some(&req));
-    debug!("returns: {:?}", features.runtime_features());
-    HttpResponse::Ok().json(features.runtime_features())
+    let features = features.runtime_features();
+    debug!(returns = ?features, "Get features");
+    HttpResponse::Ok().json(features)
 }
 
 #[derive(Debug, Deserr)]
 #[deserr(error = DeserrJsonError, rename_all = camelCase, deny_unknown_fields)]
 pub struct RuntimeTogglableFeatures {
-    #[deserr(default)]
-    pub score_details: Option<bool>,
     #[deserr(default)]
     pub vector_store: Option<bool>,
     #[deserr(default)]
     pub metrics: Option<bool>,
     #[deserr(default)]
+    pub logs_route: Option<bool>,
+    #[deserr(default)]
     pub export_puffin_reports: Option<bool>,
 }
 
@@ -60,12 +61,13 @@ async fn patch_features(
     analytics: Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
     let features = index_scheduler.features();
+    debug!(parameters = ?new_features, "Patch features");
 
     let old_features = features.runtime_features();
     let new_features = meilisearch_types::features::RuntimeTogglableFeatures {
-        score_details: new_features.0.score_details.unwrap_or(old_features.score_details),
         vector_store: new_features.0.vector_store.unwrap_or(old_features.vector_store),
         metrics: new_features.0.metrics.unwrap_or(old_features.metrics),
+        logs_route: new_features.0.logs_route.unwrap_or(old_features.logs_route),
         export_puffin_reports: new_features
             .0
             .export_puffin_reports
@@ -76,22 +78,23 @@ async fn patch_features(
     // the it renames to camelCase, which we don't want for analytics.
     // **Do not** ignore fields with `..` or `_` here, because we want to add them in the future.
     let meilisearch_types::features::RuntimeTogglableFeatures {
-        score_details,
         vector_store,
         metrics,
+        logs_route,
         export_puffin_reports,
     } = new_features;
 
     analytics.publish(
         "Experimental features Updated".to_string(),
         json!({
-            "score_details": score_details,
             "vector_store": vector_store,
             "metrics": metrics,
+            "logs_route": logs_route,
             "export_puffin_reports": export_puffin_reports,
         }),
         Some(&req),
     );
     index_scheduler.put_runtime_features(new_features)?;
+    debug!(returns = ?new_features, "Patch features");
     Ok(HttpResponse::Ok().json(new_features))
 }

meilisearch/src/routes/indexes/documents.rs

@@ -8,7 +8,6 @@ use deserr::actix_web::{AwebJson, AwebQueryParameter};
 use deserr::Deserr;
 use futures::StreamExt;
 use index_scheduler::IndexScheduler;
-use log::debug;
 use meilisearch_types::deserr::query_params::Param;
 use meilisearch_types::deserr::{DeserrJsonError, DeserrQueryParamError};
 use meilisearch_types::document_formats::{read_csv, read_json, read_ndjson, PayloadType};
@@ -28,6 +27,7 @@ use serde_json::Value;
 use tempfile::tempfile;
 use tokio::fs::File;
 use tokio::io::{AsyncSeekExt, AsyncWriteExt, BufWriter};
+use tracing::debug;
 
 use crate::analytics::{Analytics, DocumentDeletionKind, DocumentFetchKind};
 use crate::error::MeilisearchHttpError;
@@ -101,6 +101,7 @@ pub async fn get_document(
     analytics: web::Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
     let DocumentParam { index_uid, document_id } = document_param.into_inner();
+    debug!(parameters = ?params, "Get document");
     let index_uid = IndexUid::try_from(index_uid)?;
 
     analytics.get_fetch_documents(&DocumentFetchKind::PerDocumentId, &req);
@@ -110,7 +111,7 @@ pub async fn get_document(
 
     let index = index_scheduler.index(&index_uid)?;
     let document = retrieve_document(&index, &document_id, attributes_to_retrieve)?;
-    debug!("returns: {:?}", document);
+    debug!(returns = ?document, "Get document");
     Ok(HttpResponse::Ok().json(document))
 }
 
@@ -131,7 +132,7 @@ pub async fn delete_document(
     };
     let task: SummarizedTaskView =
         tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
-    debug!("returns: {:?}", task);
+    debug!(returns = ?task, "Delete document");
     Ok(HttpResponse::Accepted().json(task))
 }
 
@@ -168,9 +169,8 @@ pub async fn documents_by_query_post(
     req: HttpRequest,
     analytics: web::Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
-    debug!("called with body: {:?}", body);
-
     let body = body.into_inner();
+    debug!(parameters = ?body, "Get documents POST");
 
     analytics.post_fetch_documents(
         &DocumentFetchKind::Normal {
@@ -191,7 +191,7 @@ pub async fn get_documents(
     req: HttpRequest,
     analytics: web::Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
-    debug!("called with params: {:?}", params);
+    debug!(parameters = ?params, "Get documents GET");
 
     let BrowseQueryGet { limit, offset, fields, filter } = params.into_inner();
 
@@ -235,7 +235,7 @@ fn documents_by_query(
 
     let ret = PaginationView::new(offset, limit, total as usize, documents);
 
-    debug!("returns: {:?}", ret);
+    debug!(returns = ?ret, "Get documents");
     Ok(HttpResponse::Ok().json(ret))
 }
 
@@ -271,7 +271,7 @@ pub async fn replace_documents(
 ) -> Result<HttpResponse, ResponseError> {
     let index_uid = IndexUid::try_from(index_uid.into_inner())?;
 
-    debug!("called with params: {:?}", params);
+    debug!(parameters = ?params, "Replace documents");
     let params = params.into_inner();
 
     analytics.add_documents(&params, index_scheduler.index(&index_uid).is_err(), &req);
@@ -288,6 +288,7 @@ pub async fn replace_documents(
         allow_index_creation,
     )
     .await?;
+    debug!(returns = ?task, "Replace documents");
 
     Ok(HttpResponse::Accepted().json(task))
 }
@@ -302,8 +303,8 @@ pub async fn update_documents(
 ) -> Result<HttpResponse, ResponseError> {
     let index_uid = IndexUid::try_from(index_uid.into_inner())?;
 
-    debug!("called with params: {:?}", params);
     let params = params.into_inner();
+    debug!(parameters = ?params, "Update documents");
 
     analytics.update_documents(&params, index_scheduler.index(&index_uid).is_err(), &req);
 
@@ -319,6 +320,7 @@ pub async fn update_documents(
         allow_index_creation,
     )
     .await?;
+    debug!(returns = ?task, "Update documents");
 
     Ok(HttpResponse::Accepted().json(task))
 }
@@ -427,7 +429,10 @@ async fn document_addition(
                 Err(index_scheduler::Error::FileStore(file_store::Error::IoError(e)))
                     if e.kind() == ErrorKind::NotFound => {}
                 Err(e) => {
-                    log::warn!("Unknown error happened while deleting a malformed update file with uuid {uuid}: {e}");
+                    tracing::warn!(
+                        index_uuid = %uuid,
+                        "Unknown error happened while deleting a malformed update file: {e}"
+                    );
                 }
             }
             // We still want to return the original error to the end user.
@@ -453,7 +458,6 @@ async fn document_addition(
         }
     };
 
-    debug!("returns: {:?}", task);
     Ok(task.into())
 }
 
@@ -464,7 +468,7 @@ pub async fn delete_documents_batch(
     req: HttpRequest,
     analytics: web::Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
-    debug!("called with params: {:?}", body);
+    debug!(parameters = ?body, "Delete documents by batch");
     let index_uid = IndexUid::try_from(index_uid.into_inner())?;
 
     analytics.delete_documents(DocumentDeletionKind::PerBatch, &req);
@@ -479,7 +483,7 @@ pub async fn delete_documents_batch(
     let task: SummarizedTaskView =
         tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
 
-    debug!("returns: {:?}", task);
+    debug!(returns = ?task, "Delete documents by batch");
     Ok(HttpResponse::Accepted().json(task))
 }
 
@@ -497,7 +501,7 @@ pub async fn delete_documents_by_filter(
     req: HttpRequest,
     analytics: web::Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
-    debug!("called with params: {:?}", body);
+    debug!(parameters = ?body, "Delete documents by filter");
     let index_uid = IndexUid::try_from(index_uid.into_inner())?;
     let index_uid = index_uid.into_inner();
     let filter = body.into_inner().filter;
@@ -515,7 +519,7 @@ pub async fn delete_documents_by_filter(
     let task: SummarizedTaskView =
         tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
 
-    debug!("returns: {:?}", task);
+    debug!(returns = ?task, "Delete documents by filter");
     Ok(HttpResponse::Accepted().json(task))
 }
 
@@ -532,7 +536,7 @@ pub async fn clear_all_documents(
     let task: SummarizedTaskView =
         tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
 
-    debug!("returns: {:?}", task);
+    debug!(returns = ?task, "Delete all documents");
     Ok(HttpResponse::Accepted().json(task))
 }
 

meilisearch/src/routes/indexes/facet_search.rs

@@ -2,12 +2,12 @@ use actix_web::web::Data;
 use actix_web::{web, HttpRequest, HttpResponse};
 use deserr::actix_web::AwebJson;
 use index_scheduler::IndexScheduler;
-use log::debug;
 use meilisearch_types::deserr::DeserrJsonError;
 use meilisearch_types::error::deserr_codes::*;
 use meilisearch_types::error::ResponseError;
 use meilisearch_types::index_uid::IndexUid;
 use serde_json::Value;
+use tracing::debug;
 
 use crate::analytics::{Analytics, FacetSearchAggregator};
 use crate::extractors::authentication::policies::*;
@@ -56,7 +56,7 @@ pub async fn search(
     let index_uid = IndexUid::try_from(index_uid.into_inner())?;
 
     let query = params.into_inner();
-    debug!("facet search called with params: {:?}", query);
+    debug!(parameters = ?query, "Facet search");
 
     let mut aggregate = FacetSearchAggregator::from_query(&query, &req);
 
@@ -83,7 +83,7 @@ pub async fn search(
 
     let search_result = search_result?;
 
-    debug!("returns: {:?}", search_result);
+    debug!(returns = ?search_result, "Facet search");
     Ok(HttpResponse::Ok().json(search_result))
 }
 

meilisearch/src/routes/indexes/mod.rs

@@ -5,7 +5,6 @@ use actix_web::{web, HttpRequest, HttpResponse};
 use deserr::actix_web::{AwebJson, AwebQueryParameter};
 use deserr::{DeserializeError, Deserr, ValuePointerRef};
 use index_scheduler::IndexScheduler;
-use log::debug;
 use meilisearch_types::deserr::query_params::Param;
 use meilisearch_types::deserr::{immutable_field_error, DeserrJsonError, DeserrQueryParamError};
 use meilisearch_types::error::deserr_codes::*;
@@ -16,6 +15,7 @@ use meilisearch_types::tasks::KindWithContent;
 use serde::Serialize;
 use serde_json::json;
 use time::OffsetDateTime;
+use tracing::debug;
 
 use super::{Pagination, SummarizedTaskView, PAGINATION_DEFAULT_LIMIT};
 use crate::analytics::Analytics;
@@ -93,6 +93,7 @@ pub async fn list_indexes(
     index_scheduler: GuardedData<ActionPolicy<{ actions::INDEXES_GET }>, Data<IndexScheduler>>,
     paginate: AwebQueryParameter<ListIndexes, DeserrQueryParamError>,
 ) -> Result<HttpResponse, ResponseError> {
+    debug!(parameters = ?paginate, "List indexes");
     let filters = index_scheduler.filters();
     let indexes: Vec<Option<IndexView>> =
         index_scheduler.try_for_each_index(|uid, index| -> Result<Option<IndexView>, _> {
@@ -105,7 +106,7 @@ pub async fn list_indexes(
     let indexes: Vec<IndexView> = indexes.into_iter().flatten().collect();
     let ret = paginate.as_pagination().auto_paginate_sized(indexes.into_iter());
 
-    debug!("returns: {:?}", ret);
+    debug!(returns = ?ret, "List indexes");
     Ok(HttpResponse::Ok().json(ret))
 }
 
@@ -124,6 +125,7 @@ pub async fn create_index(
     req: HttpRequest,
     analytics: web::Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
+    debug!(parameters = ?body, "Create index");
     let IndexCreateRequest { primary_key, uid } = body.into_inner();
 
     let allow_index_creation = index_scheduler.filters().allow_index_creation(&uid);
@@ -137,6 +139,7 @@ pub async fn create_index(
         let task = KindWithContent::IndexCreation { index_uid: uid.to_string(), primary_key };
         let task: SummarizedTaskView =
             tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
+        debug!(returns = ?task, "Create index");
 
         Ok(HttpResponse::Accepted().json(task))
     } else {
@@ -177,7 +180,7 @@ pub async fn get_index(
     let index = index_scheduler.index(&index_uid)?;
     let index_view = IndexView::new(index_uid.into_inner(), &index)?;
 
-    debug!("returns: {:?}", index_view);
+    debug!(returns = ?index_view, "Get index");
 
     Ok(HttpResponse::Ok().json(index_view))
 }
@@ -189,7 +192,7 @@ pub async fn update_index(
     req: HttpRequest,
     analytics: web::Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
-    debug!("called with params: {:?}", body);
+    debug!(parameters = ?body, "Update index");
     let index_uid = IndexUid::try_from(index_uid.into_inner())?;
     let body = body.into_inner();
     analytics.publish(
@@ -206,7 +209,7 @@ pub async fn update_index(
     let task: SummarizedTaskView =
         tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
 
-    debug!("returns: {:?}", task);
+    debug!(returns = ?task, "Update index");
     Ok(HttpResponse::Accepted().json(task))
 }
 
@@ -218,6 +221,7 @@ pub async fn delete_index(
     let task = KindWithContent::IndexDeletion { index_uid: index_uid.into_inner() };
     let task: SummarizedTaskView =
         tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
+    debug!(returns = ?task, "Delete index");
 
     Ok(HttpResponse::Accepted().json(task))
 }
@@ -255,6 +259,6 @@ pub async fn get_index_stats(
 
     let stats = IndexStats::from(index_scheduler.index_stats(&index_uid)?);
 
-    debug!("returns: {:?}", stats);
+    debug!(returns = ?stats, "Get index stats");
     Ok(HttpResponse::Ok().json(stats))
 }

meilisearch/src/routes/indexes/search.rs

@@ -2,7 +2,6 @@ use actix_web::web::Data;
 use actix_web::{web, HttpRequest, HttpResponse};
 use deserr::actix_web::{AwebJson, AwebQueryParameter};
 use index_scheduler::IndexScheduler;
-use log::{debug, warn};
 use meilisearch_types::deserr::query_params::Param;
 use meilisearch_types::deserr::{DeserrJsonError, DeserrQueryParamError};
 use meilisearch_types::error::deserr_codes::*;
@@ -12,6 +11,7 @@ use meilisearch_types::milli;
 use meilisearch_types::milli::vector::DistributionShift;
 use meilisearch_types::serde_cs::vec::CS;
 use serde_json::Value;
+use tracing::{debug, warn};
 
 use crate::analytics::{Analytics, SearchAggregator};
 use crate::extractors::authentication::policies::*;
@@ -186,7 +186,7 @@ pub async fn search_with_url_query(
     req: HttpRequest,
     analytics: web::Data<dyn Analytics>,
 ) -> Result<HttpResponse, ResponseError> {
-    debug!("called with params: {:?}", params);
+    debug!(parameters = ?params, "Search get");
     let index_uid = IndexUid::try_from(index_uid.into_inner())?;
 
     let mut query: SearchQuery = params.into_inner().into();
@@ -213,7 +213,7 @@ pub async fn search_with_url_query(
 
     let search_result = search_result?;
 
-    debug!("returns: {:?}", search_result);
+    debug!(returns = ?search_result, "Search get");
     Ok(HttpResponse::Ok().json(search_result))
 }
 
@@ -227,7 +227,7 @@ pub async fn search_with_post(
     let index_uid = IndexUid::try_from(index_uid.into_inner())?;
 
     let mut query = params.into_inner();
-    debug!("search called with params: {:?}", query);
+    debug!(parameters = ?query, "Search post");
 
     // Tenant token search_rules.
     if let Some(search_rules) = index_scheduler.filters().get_index_search_rules(&index_uid) {
@@ -252,7 +252,7 @@ pub async fn search_with_post(
 
     let search_result = search_result?;
 
-    debug!("returns: {:?}", search_result);
+    debug!(returns = ?search_result, "Search post");
     Ok(HttpResponse::Ok().json(search_result))
 }
 

meilisearch/src/routes/indexes/settings.rs

@@ -2,7 +2,6 @@ use actix_web::web::Data;
 use actix_web::{web, HttpRequest, HttpResponse};
 use deserr::actix_web::AwebJson;
 use index_scheduler::IndexScheduler;
-use log::debug;
 use meilisearch_types::deserr::DeserrJsonError;
 use meilisearch_types::error::ResponseError;
 use meilisearch_types::facet_values_sort::FacetValuesSort;
@@ -11,6 +10,7 @@ use meilisearch_types::milli::update::Setting;
 use meilisearch_types::settings::{settings, RankingRuleView, Settings, Unchecked};
 use meilisearch_types::tasks::KindWithContent;
 use serde_json::json;
+use tracing::debug;
 
 use crate::analytics::Analytics;
 use crate::extractors::authentication::policies::*;
@@ -24,12 +24,12 @@ macro_rules! make_setting_route {
             use actix_web::web::Data;
             use actix_web::{web, HttpRequest, HttpResponse, Resource};
             use index_scheduler::IndexScheduler;
-            use log::debug;
             use meilisearch_types::error::ResponseError;
             use meilisearch_types::index_uid::IndexUid;
             use meilisearch_types::milli::update::Setting;
             use meilisearch_types::settings::{settings, Settings};
             use meilisearch_types::tasks::KindWithContent;
+            use tracing::debug;
             use $crate::analytics::Analytics;
             use $crate::extractors::authentication::policies::*;
             use $crate::extractors::authentication::GuardedData;
@@ -61,7 +61,7 @@ macro_rules! make_setting_route {
                         .await??
                         .into();
 
-                debug!("returns: {:?}", task);
+                debug!(returns = ?task, "Delete settings");
                 Ok(HttpResponse::Accepted().json(task))
             }
 
@@ -78,6 +78,7 @@ macro_rules! make_setting_route {
                 let index_uid = IndexUid::try_from(index_uid.into_inner())?;
 
                 let body = body.into_inner();
+                debug!(parameters = ?body, "Update settings");
 
                 #[allow(clippy::redundant_closure_call)]
                 $analytics(&body, &req);
@@ -109,7 +110,7 @@ macro_rules! make_setting_route {
                         .await??
                         .into();
 
-                debug!("returns: {:?}", task);
+                debug!(returns = ?task, "Update settings");
                 Ok(HttpResponse::Accepted().json(task))
             }
 
@@ -126,7 +127,7 @@ macro_rules! make_setting_route {
                 let rtxn = index.read_txn()?;
                 let settings = settings(&index, &rtxn)?;
 
-                debug!("returns: {:?}", settings);
+                debug!(returns = ?settings, "Update settings");
                 let mut json = serde_json::json!(&settings);
                 let val = json[$camelcase_attr].take();
 
@@ -656,6 +657,7 @@ pub async fn update_all(
     let index_uid = IndexUid::try_from(index_uid.into_inner())?;
 
     let new_settings = body.into_inner();
+    debug!(parameters = ?new_settings, "Update all settings");
     let new_settings = validate_settings(new_settings, &index_scheduler)?;
 
     analytics.publish(
@@ -768,7 +770,7 @@ pub async fn update_all(
     let task: SummarizedTaskView =
         tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
 
-    debug!("returns: {:?}", task);
+    debug!(returns = ?task, "Update all settings");
     Ok(HttpResponse::Accepted().json(task))
 }
 
@@ -781,7 +783,7 @@ pub async fn get_all(
     let index = index_scheduler.index(&index_uid)?;
     let rtxn = index.read_txn()?;
     let new_settings = settings(&index, &rtxn)?;
-    debug!("returns: {:?}", new_settings);
+    debug!(returns = ?new_settings, "Get all settings");
     Ok(HttpResponse::Ok().json(new_settings))
 }
 
@@ -804,7 +806,7 @@ pub async fn delete_all(
     let task: SummarizedTaskView =
         tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
 
-    debug!("returns: {:?}", task);
+    debug!(returns = ?task, "Delete all settings");
     Ok(HttpResponse::Accepted().json(task))
 }
 

meilisearch/src/routes/logs.rs

@@ -0,0 +1,281 @@
+use std::convert::Infallible;
+use std::io::Write;
+use std::ops::ControlFlow;
+use std::pin::Pin;
+use std::str::FromStr;
+use std::sync::Arc;
+
+use actix_web::web::{Bytes, Data};
+use actix_web::{web, HttpResponse};
+use deserr::actix_web::AwebJson;
+use deserr::{DeserializeError, Deserr, ErrorKind, MergeWithError, ValuePointerRef};
+use futures_util::Stream;
+use index_scheduler::IndexScheduler;
+use meilisearch_types::deserr::DeserrJsonError;
+use meilisearch_types::error::deserr_codes::*;
+use meilisearch_types::error::{Code, ResponseError};
+use tokio::sync::mpsc;
+use tracing_subscriber::filter::Targets;
+use tracing_subscriber::Layer;
+
+use crate::error::MeilisearchHttpError;
+use crate::extractors::authentication::policies::*;
+use crate::extractors::authentication::GuardedData;
+use crate::extractors::sequential_extractor::SeqHandler;
+use crate::LogRouteHandle;
+
+pub fn configure(cfg: &mut web::ServiceConfig) {
+    cfg.service(
+        web::resource("stream")
+            .route(web::post().to(SeqHandler(get_logs)))
+            .route(web::delete().to(SeqHandler(cancel_logs))),
+    );
+}
+
+#[derive(Debug, Default, Clone, Copy, Deserr, PartialEq, Eq)]
+#[deserr(rename_all = camelCase)]
+pub enum LogMode {
+    #[default]
+    Human,
+    Profile,
+}
+
+/// Simple wrapper around the `Targets` from `tracing_subscriber` to implement `MergeWithError` on it.
+#[derive(Clone, Debug)]
+struct MyTargets(Targets);
+
+/// Simple wrapper around the `ParseError` from `tracing_subscriber` to implement `MergeWithError` on it.
+#[derive(Debug, thiserror::Error)]
+enum MyParseError {
+    #[error(transparent)]
+    ParseError(#[from] tracing_subscriber::filter::ParseError),
+    #[error(
+        "Empty string is not a valid target. If you want to get no logs use `OFF`. Usage: `info`, `meilisearch=info`, or you can write multiple filters in one target: `index_scheduler=info,milli=trace`"
+    )]
+    Example,
+}
+
+impl FromStr for MyTargets {
+    type Err = MyParseError;
+
+    fn from_str(s: &str) -> Result<Self, Self::Err> {
+        if s.is_empty() {
+            Err(MyParseError::Example)
+        } else {
+            Ok(MyTargets(Targets::from_str(s).map_err(MyParseError::ParseError)?))
+        }
+    }
+}
+
+impl MergeWithError<MyParseError> for DeserrJsonError<BadRequest> {
+    fn merge(
+        _self_: Option<Self>,
+        other: MyParseError,
+        merge_location: ValuePointerRef,
+    ) -> ControlFlow<Self, Self> {
+        Self::error::<Infallible>(
+            None,
+            ErrorKind::Unexpected { msg: other.to_string() },
+            merge_location,
+        )
+    }
+}
+
+#[derive(Debug, Deserr)]
+#[deserr(error = DeserrJsonError, rename_all = camelCase, deny_unknown_fields, validate = validate_get_logs -> DeserrJsonError<InvalidSettingsTypoTolerance>)]
+pub struct GetLogs {
+    #[deserr(default = "info".parse().unwrap(), try_from(&String) = MyTargets::from_str -> DeserrJsonError<BadRequest>)]
+    target: MyTargets,
+
+    #[deserr(default, error = DeserrJsonError<BadRequest>)]
+    mode: LogMode,
+
+    #[deserr(default = false, error = DeserrJsonError<BadRequest>)]
+    profile_memory: bool,
+}
+
+fn validate_get_logs<E: DeserializeError>(
+    logs: GetLogs,
+    location: ValuePointerRef,
+) -> Result<GetLogs, E> {
+    if logs.profile_memory && logs.mode != LogMode::Profile {
+        Err(deserr::take_cf_content(E::error::<Infallible>(
+            None,
+            ErrorKind::Unexpected {
+                msg: format!("`profile_memory` can only be used while profiling code and is not compatible with the {:?} mode.", logs.mode),
+            },
+            location,
+        )))
+    } else {
+        Ok(logs)
+    }
+}
+
+struct LogWriter {
+    sender: mpsc::UnboundedSender<Vec<u8>>,
+}
+
+impl Write for LogWriter {
+    fn write(&mut self, buf: &[u8]) -> std::io::Result<usize> {
+        self.sender.send(buf.to_vec()).map_err(std::io::Error::other)?;
+        Ok(buf.len())
+    }
+
+    fn flush(&mut self) -> std::io::Result<()> {
+        Ok(())
+    }
+}
+
+struct HandleGuard {
+    /// We need to keep an handle on the logs to make it available again when the streamer is dropped
+    logs: Arc<LogRouteHandle>,
+}
+
+impl Drop for HandleGuard {
+    fn drop(&mut self) {
+        if let Err(e) = self.logs.modify(|layer| *layer.inner_mut() = None) {
+            tracing::error!("Could not free the logs route: {e}");
+        }
+    }
+}
+
+fn byte_stream(
+    receiver: mpsc::UnboundedReceiver<Vec<u8>>,
+    guard: HandleGuard,
+) -> impl futures_util::Stream<Item = Result<Bytes, ResponseError>> {
+    futures_util::stream::unfold((receiver, guard), move |(mut receiver, guard)| async move {
+        let vec = receiver.recv().await;
+
+        vec.map(From::from).map(Ok).map(|a| (a, (receiver, guard)))
+    })
+}
+
+type PinnedByteStream = Pin<Box<dyn Stream<Item = Result<Bytes, ResponseError>>>>;
+
+fn make_layer<
+    S: tracing::Subscriber + for<'span> tracing_subscriber::registry::LookupSpan<'span>,
+>(
+    opt: &GetLogs,
+    logs: Data<LogRouteHandle>,
+) -> (Box<dyn Layer<S> + Send + Sync>, PinnedByteStream) {
+    let guard = HandleGuard { logs: logs.into_inner() };
+    match opt.mode {
+        LogMode::Human => {
+            let (sender, receiver) = tokio::sync::mpsc::unbounded_channel();
+
+            let fmt_layer = tracing_subscriber::fmt::layer()
+                .with_writer(move || LogWriter { sender: sender.clone() })
+                .with_span_events(tracing_subscriber::fmt::format::FmtSpan::ACTIVE);
+
+            let stream = byte_stream(receiver, guard);
+            (Box::new(fmt_layer) as Box<dyn Layer<S> + Send + Sync>, Box::pin(stream))
+        }
+        LogMode::Profile => {
+            let (trace, layer) = tracing_trace::Trace::new(opt.profile_memory);
+
+            let stream = entry_stream(trace, guard);
+
+            (Box::new(layer) as Box<dyn Layer<S> + Send + Sync>, Box::pin(stream))
+        }
+    }
+}
+
+fn entry_stream(
+    trace: tracing_trace::Trace,
+    guard: HandleGuard,
+) -> impl Stream<Item = Result<Bytes, ResponseError>> {
+    let receiver = trace.into_receiver();
+    let entry_buf = Vec::new();
+
+    futures_util::stream::unfold(
+        (receiver, entry_buf, guard),
+        move |(mut receiver, mut entry_buf, guard)| async move {
+            let mut bytes = Vec::new();
+
+            while bytes.len() < 8192 {
+                entry_buf.clear();
+
+                let Ok(count) = tokio::time::timeout(
+                    std::time::Duration::from_secs(1),
+                    receiver.recv_many(&mut entry_buf, 100),
+                )
+                .await
+                else {
+                    break;
+                };
+
+                if count == 0 {
+                    if !bytes.is_empty() {
+                        break;
+                    }
+
+                    // channel closed, exit
+                    return None;
+                }
+
+                for entry in &entry_buf {
+                    if let Err(error) = serde_json::to_writer(&mut bytes, entry) {
+                        tracing::error!(
+                            error = &error as &dyn std::error::Error,
+                            "deserializing entry"
+                        );
+                        return Some((
+                            Err(ResponseError::from_msg(
+                                format!("error deserializing entry: {error}"),
+                                Code::Internal,
+                            )),
+                            (receiver, entry_buf, guard),
+                        ));
+                    }
+                }
+            }
+
+            Some((Ok(bytes.into()), (receiver, entry_buf, guard)))
+        },
+    )
+}
+
+pub async fn get_logs(
+    index_scheduler: GuardedData<ActionPolicy<{ actions::METRICS_GET }>, Data<IndexScheduler>>,
+    logs: Data<LogRouteHandle>,
+    body: AwebJson<GetLogs, DeserrJsonError>,
+) -> Result<HttpResponse, ResponseError> {
+    index_scheduler.features().check_logs_route()?;
+
+    let opt = body.into_inner();
+    let mut stream = None;
+
+    logs.modify(|layer| match layer.inner_mut() {
+        None => {
+            // there is no one getting logs
+            *layer.filter_mut() = opt.target.0.clone();
+            let (new_layer, new_stream) = make_layer(&opt, logs.clone());
+
+            *layer.inner_mut() = Some(new_layer);
+            stream = Some(new_stream);
+        }
+        Some(_) => {
+            // there is already someone getting logs
+        }
+    })
+    .unwrap();
+
+    if let Some(stream) = stream {
+        Ok(HttpResponse::Ok().streaming(stream))
+    } else {
+        Err(MeilisearchHttpError::AlreadyUsedLogRoute.into())
+    }
+}
+
+pub async fn cancel_logs(
+    index_scheduler: GuardedData<ActionPolicy<{ actions::METRICS_GET }>, Data<IndexScheduler>>,
+    logs: Data<LogRouteHandle>,
+) -> Result<HttpResponse, ResponseError> {
+    index_scheduler.features().check_logs_route()?;
+
+    if let Err(e) = logs.modify(|layer| *layer.inner_mut() = None) {
+        tracing::error!("Could not free the logs route: {e}");
+    }
+
+    Ok(HttpResponse::NoContent().finish())
+}

meilisearch/src/routes/mod.rs

@@ -3,7 +3,6 @@ use std::collections::BTreeMap;
 use actix_web::web::Data;
 use actix_web::{web, HttpRequest, HttpResponse};
 use index_scheduler::IndexScheduler;
-use log::debug;
 use meilisearch_auth::AuthController;
 use meilisearch_types::error::ResponseError;
 use meilisearch_types::settings::{Settings, Unchecked};
@@ -11,6 +10,7 @@ use meilisearch_types::tasks::{Kind, Status, Task, TaskId};
 use serde::{Deserialize, Serialize};
 use serde_json::json;
 use time::OffsetDateTime;
+use tracing::debug;
 
 use crate::analytics::Analytics;
 use crate::extractors::authentication::policies::*;
@@ -22,6 +22,7 @@ mod api_key;
 mod dump;
 pub mod features;
 pub mod indexes;
+mod logs;
 mod metrics;
 mod multi_search;
 mod snapshot;
@@ -31,6 +32,7 @@ pub mod tasks;
 pub fn configure(cfg: &mut web::ServiceConfig) {
     cfg.service(web::scope("/tasks").configure(tasks::configure))
         .service(web::resource("/health").route(web::get().to(get_health)))
+        .service(web::scope("/logs").configure(logs::configure))
         .service(web::scope("/keys").configure(api_key::configure))
         .service(web::scope("/dumps").configure(dump::configure))
         .service(web::scope("/snapshots").configure(snapshot::configure))
@@ -250,7 +252,7 @@ async fn get_stats(
 
     let stats = create_all_stats((*index_scheduler).clone(), (*auth_controller).clone(), filters)?;
 
-    debug!("returns: {:?}", stats);
+    debug!(returns = ?stats, "Get stats");
     Ok(HttpResponse::Ok().json(stats))
 }
 

meilisearch/src/routes/multi_search.rs

@@ -3,11 +3,11 @@ use actix_web::web::{self, Data};
 use actix_web::{HttpRequest, HttpResponse};
 use deserr::actix_web::AwebJson;
 use index_scheduler::IndexScheduler;
-use log::debug;
 use meilisearch_types::deserr::DeserrJsonError;
 use meilisearch_types::error::ResponseError;
 use meilisearch_types::keys::actions;
 use serde::Serialize;
+use tracing::debug;
 
 use crate::analytics::{Analytics, MultiSearchAggregator};
 use crate::extractors::authentication::policies::ActionPolicy;
@@ -52,7 +52,7 @@ pub async fn multi_search_with_post(
         for (query_index, (index_uid, mut query)) in
             queries.into_iter().map(SearchQueryWithIndex::into_index_query).enumerate()
         {
-            debug!("multi-search #{query_index}: called with params: {:?}", query);
+            debug!(on_index = query_index, parameters = ?query, "Multi-search");
 
             // Check index from API key
             if !index_scheduler.filters().is_index_authorized(&index_uid) {
@@ -107,7 +107,7 @@ pub async fn multi_search_with_post(
         err
     })?;
 
-    debug!("returns: {:?}", search_results);
+    debug!(returns = ?search_results, "Multi-search");
 
     Ok(HttpResponse::Ok().json(SearchResults { results: search_results }))
 }

meilisearch/src/routes/snapshot.rs

@@ -1,10 +1,10 @@
 use actix_web::web::Data;
 use actix_web::{web, HttpRequest, HttpResponse};
 use index_scheduler::IndexScheduler;
-use log::debug;
 use meilisearch_types::error::ResponseError;
 use meilisearch_types::tasks::KindWithContent;
 use serde_json::json;
+use tracing::debug;
 
 use crate::analytics::Analytics;
 use crate::extractors::authentication::policies::*;
@@ -27,6 +27,6 @@ pub async fn create_snapshot(
     let task: SummarizedTaskView =
         tokio::task::spawn_blocking(move || index_scheduler.register(task)).await??.into();
 
-    debug!("returns: {:?}", task);
+    debug!(returns = ?task, "Create snapshot");
     Ok(HttpResponse::Accepted().json(task))
 }

meilisearch/src/search.rs

@@ -441,10 +441,6 @@ fn prepare_search<'t>(
         ScoringStrategy::Skip
     });
 
-    if query.show_ranking_score_details {
-        features.check_score_details()?;
-    }
-
     if let Some(HybridQuery { embedder: Some(embedder), .. }) = &query.hybrid {
         search.embedder_name(embedder);
     }

meilisearch/tests/auth/authorization.rs

@@ -59,6 +59,8 @@ pub static AUTHORIZATIONS: Lazy<HashMap<(&'static str, &'static str), HashSet<&'
             ("POST",    "/snapshots") =>                                       hashset!{"snapshots.create", "snapshots.*", "*"},
             ("GET",     "/version") =>                                         hashset!{"version", "*"},
             ("GET",     "/metrics") =>                                         hashset!{"metrics.get", "metrics.*", "*"},
+            ("POST",    "/logs/stream") =>                                     hashset!{"metrics.get", "metrics.*", "*"},
+            ("DELETE",  "/logs/stream") =>                                     hashset!{"metrics.get", "metrics.*", "*"},
             ("PATCH",   "/keys/mykey/") =>                                     hashset!{"keys.update", "*"},
             ("GET",     "/keys/mykey/") =>                                     hashset!{"keys.get", "*"},
             ("DELETE",  "/keys/mykey/") =>                                     hashset!{"keys.delete", "*"},

meilisearch/tests/common/mod.rs

@@ -64,7 +64,7 @@ impl Display for Value {
         write!(
             f,
             "{}",
-            json_string!(self, { ".enqueuedAt" => "[date]", ".processedAt" => "[date]", ".finishedAt" => "[date]", ".duration" => "[duration]" })
+            json_string!(self, { ".enqueuedAt" => "[date]", ".startedAt" => "[date]", ".finishedAt" => "[date]", ".duration" => "[duration]" })
         )
     }
 }

meilisearch/tests/common/server.rs

@@ -13,6 +13,8 @@ use meilisearch::{analytics, create_app, setup_meilisearch};
 use once_cell::sync::Lazy;
 use tempfile::TempDir;
 use tokio::time::sleep;
+use tracing::level_filters::LevelFilter;
+use tracing_subscriber::Layer;
 
 use super::index::Index;
 use super::service::Service;
@@ -81,10 +83,16 @@ impl Server {
         Response = ServiceResponse<impl MessageBody>,
         Error = actix_web::Error,
     > {
+        let (_route_layer, route_layer_handle) =
+            tracing_subscriber::reload::Layer::new(None.with_filter(
+                tracing_subscriber::filter::Targets::new().with_target("", LevelFilter::OFF),
+            ));
+
         actix_web::test::init_service(create_app(
             self.service.index_scheduler.clone().into(),
             self.service.auth.clone().into(),
             self.service.options.clone(),
+            route_layer_handle,
             analytics::MockAnalytics::new(&self.service.options),
             true,
         ))

meilisearch/tests/common/service.rs

@@ -7,6 +7,8 @@ use actix_web::test::TestRequest;
 use index_scheduler::IndexScheduler;
 use meilisearch::{analytics, create_app, Opt};
 use meilisearch_auth::AuthController;
+use tracing::level_filters::LevelFilter;
+use tracing_subscriber::Layer;
 
 use crate::common::encoder::Encoder;
 use crate::common::Value;
@@ -105,10 +107,16 @@ impl Service {
     }
 
     pub async fn request(&self, mut req: test::TestRequest) -> (Value, StatusCode) {
+        let (_route_layer, route_layer_handle) =
+            tracing_subscriber::reload::Layer::new(None.with_filter(
+                tracing_subscriber::filter::Targets::new().with_target("", LevelFilter::OFF),
+            ));
+
         let app = test::init_service(create_app(
             self.index_scheduler.clone().into(),
             self.auth.clone().into(),
             self.options.clone(),
+            route_layer_handle,
             analytics::MockAnalytics::new(&self.options),
             true,
         ))

meilisearch/tests/documents/add_documents.rs

@@ -1760,6 +1760,181 @@ async fn add_documents_invalid_geo_field() {
       "finishedAt": "[date]"
     }
     "###);
+
+    // The three next tests are related to #4333
+
+    // _geo has a lat and lng but set to `null`
+    let documents = json!([
+        {
+            "id": "12",
+            "_geo": { "lng": null, "lat": 67}
+        }
+    ]);
+
+    let (response, code) = index.add_documents(documents, None).await;
+    snapshot!(code, @"202 Accepted");
+    let response = index.wait_task(response.uid()).await;
+    snapshot!(json_string!(response, { ".duration" => "[duration]", ".enqueuedAt" => "[date]", ".startedAt" => "[date]", ".finishedAt" => "[date]" }),
+        @r###"
+    {
+      "uid": 14,
+      "indexUid": "test",
+      "status": "failed",
+      "type": "documentAdditionOrUpdate",
+      "canceledBy": null,
+      "details": {
+        "receivedDocuments": 1,
+        "indexedDocuments": 0
+      },
+      "error": {
+        "message": "Could not parse longitude in the document with the id: `12`. Was expecting a finite number but instead got `null`.",
+        "code": "invalid_document_geo_field",
+        "type": "invalid_request",
+        "link": "https://docs.meilisearch.com/errors#invalid_document_geo_field"
+      },
+      "duration": "[duration]",
+      "enqueuedAt": "[date]",
+      "startedAt": "[date]",
+      "finishedAt": "[date]"
+    }
+    "###);
+
+    // _geo has a lat and lng but set to `null`
+    let documents = json!([
+        {
+            "id": "12",
+            "_geo": { "lng": 35, "lat": null }
+        }
+    ]);
+
+    let (response, code) = index.add_documents(documents, None).await;
+    snapshot!(code, @"202 Accepted");
+    let response = index.wait_task(response.uid()).await;
+    snapshot!(json_string!(response, { ".duration" => "[duration]", ".enqueuedAt" => "[date]", ".startedAt" => "[date]", ".finishedAt" => "[date]" }),
+        @r###"
+    {
+      "uid": 15,
+      "indexUid": "test",
+      "status": "failed",
+      "type": "documentAdditionOrUpdate",
+      "canceledBy": null,
+      "details": {
+        "receivedDocuments": 1,
+        "indexedDocuments": 0
+      },
+      "error": {
+        "message": "Could not parse latitude in the document with the id: `12`. Was expecting a finite number but instead got `null`.",
+        "code": "invalid_document_geo_field",
+        "type": "invalid_request",
+        "link": "https://docs.meilisearch.com/errors#invalid_document_geo_field"
+      },
+      "duration": "[duration]",
+      "enqueuedAt": "[date]",
+      "startedAt": "[date]",
+      "finishedAt": "[date]"
+    }
+    "###);
+
+    // _geo has a lat and lng but set to `null`
+    let documents = json!([
+        {
+            "id": "13",
+            "_geo": { "lng": null, "lat": null }
+        }
+    ]);
+
+    let (response, code) = index.add_documents(documents, None).await;
+    snapshot!(code, @"202 Accepted");
+    let response = index.wait_task(response.uid()).await;
+    snapshot!(json_string!(response, { ".duration" => "[duration]", ".enqueuedAt" => "[date]", ".startedAt" => "[date]", ".finishedAt" => "[date]" }),
+        @r###"
+    {
+      "uid": 16,
+      "indexUid": "test",
+      "status": "failed",
+      "type": "documentAdditionOrUpdate",
+      "canceledBy": null,
+      "details": {
+        "receivedDocuments": 1,
+        "indexedDocuments": 0
+      },
+      "error": {
+        "message": "Could not parse latitude nor longitude in the document with the id: `13`. Was expecting finite numbers but instead got `null` and `null`.",
+        "code": "invalid_document_geo_field",
+        "type": "invalid_request",
+        "link": "https://docs.meilisearch.com/errors#invalid_document_geo_field"
+      },
+      "duration": "[duration]",
+      "enqueuedAt": "[date]",
+      "startedAt": "[date]",
+      "finishedAt": "[date]"
+    }
+    "###);
+}
+
+// Related to #4333
+#[actix_rt::test]
+async fn add_invalid_geo_and_then_settings() {
+    let server = Server::new().await;
+    let index = server.index("test");
+    index.create(Some("id")).await;
+
+    // _geo is not an object
+    let documents = json!([
+        {
+            "id": "11",
+            "_geo": { "lat": null, "lng": null },
+        }
+    ]);
+    let (ret, code) = index.add_documents(documents, None).await;
+    snapshot!(code, @"202 Accepted");
+    let ret = index.wait_task(ret.uid()).await;
+    snapshot!(ret, @r###"
+    {
+      "uid": 1,
+      "indexUid": "test",
+      "status": "succeeded",
+      "type": "documentAdditionOrUpdate",
+      "canceledBy": null,
+      "details": {
+        "receivedDocuments": 1,
+        "indexedDocuments": 1
+      },
+      "error": null,
+      "duration": "[duration]",
+      "enqueuedAt": "[date]",
+      "startedAt": "[date]",
+      "finishedAt": "[date]"
+    }
+    "###);
+
+    let (ret, code) = index.update_settings(json!({"sortableAttributes": ["_geo"]})).await;
+    snapshot!(code, @"202 Accepted");
+    let ret = index.wait_task(ret.uid()).await;
+    snapshot!(ret, @r###"
+    {
+      "uid": 2,
+      "indexUid": "test",
+      "status": "failed",
+      "type": "settingsUpdate",
+      "canceledBy": null,
+      "details": {
+        "sortableAttributes": [
+          "_geo"
+        ]
+      },
+      "error": {
+        "message": "Could not parse latitude in the document with the id: `\"11\"`. Was expecting a finite number but instead got `null`.",
+        "code": "invalid_document_geo_field",
+        "type": "invalid_request",
+        "link": "https://docs.meilisearch.com/errors#invalid_document_geo_field"
+      },
+      "duration": "[duration]",
+      "enqueuedAt": "[date]",
+      "startedAt": "[date]",
+      "finishedAt": "[date]"
+    }
+    "###);
 }
 
 #[actix_rt::test]

meilisearch/tests/dumps/mod.rs

@@ -1845,9 +1845,9 @@ async fn import_dump_v6_containing_experimental_features() {
     meili_snap::snapshot!(code, @"200 OK");
     meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
     {
-      "scoreDetails": false,
       "vectorStore": false,
       "metrics": false,
+      "logsRoute": false,
       "exportPuffinReports": false
     }
     "###);

meilisearch/tests/features/mod.rs

@@ -18,9 +18,9 @@ async fn experimental_features() {
     meili_snap::snapshot!(code, @"200 OK");
     meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
     {
-      "scoreDetails": false,
       "vectorStore": false,
       "metrics": false,
+      "logsRoute": false,
       "exportPuffinReports": false
     }
     "###);
@@ -30,9 +30,9 @@ async fn experimental_features() {
     meili_snap::snapshot!(code, @"200 OK");
     meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
     {
-      "scoreDetails": false,
       "vectorStore": true,
       "metrics": false,
+      "logsRoute": false,
       "exportPuffinReports": false
     }
     "###);
@@ -42,9 +42,9 @@ async fn experimental_features() {
     meili_snap::snapshot!(code, @"200 OK");
     meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
     {
-      "scoreDetails": false,
       "vectorStore": true,
       "metrics": false,
+      "logsRoute": false,
       "exportPuffinReports": false
     }
     "###);
@@ -55,9 +55,9 @@ async fn experimental_features() {
     meili_snap::snapshot!(code, @"200 OK");
     meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
     {
-      "scoreDetails": false,
       "vectorStore": true,
       "metrics": false,
+      "logsRoute": false,
       "exportPuffinReports": false
     }
     "###);
@@ -68,9 +68,9 @@ async fn experimental_features() {
     meili_snap::snapshot!(code, @"200 OK");
     meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
     {
-      "scoreDetails": false,
       "vectorStore": true,
       "metrics": false,
+      "logsRoute": false,
       "exportPuffinReports": false
     }
     "###);
@@ -88,9 +88,9 @@ async fn experimental_feature_metrics() {
     meili_snap::snapshot!(code, @"200 OK");
     meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
     {
-      "scoreDetails": false,
       "vectorStore": false,
       "metrics": true,
+      "logsRoute": false,
       "exportPuffinReports": false
     }
     "###);
@@ -146,7 +146,7 @@ async fn errors() {
     meili_snap::snapshot!(code, @"400 Bad Request");
     meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
     {
-      "message": "Unknown field `NotAFeature`: expected one of `scoreDetails`, `vectorStore`, `metrics`, `exportPuffinReports`",
+      "message": "Unknown field `NotAFeature`: expected one of `vectorStore`, `metrics`, `logsRoute`, `exportPuffinReports`",
       "code": "bad_request",
       "type": "invalid_request",
       "link": "https://docs.meilisearch.com/errors#bad_request"

meilisearch/tests/integration.rs

@@ -5,6 +5,7 @@ mod documents;
 mod dumps;
 mod features;
 mod index;
+mod logs;
 mod search;
 mod settings;
 mod snapshot;

meilisearch/tests/logs/error.rs

@@ -0,0 +1,182 @@
+use meili_snap::*;
+
+use crate::common::Server;
+use crate::json;
+
+#[actix_rt::test]
+async fn logs_stream_bad_target() {
+    let server = Server::new().await;
+
+    // Wrong type
+    let (response, code) = server.service.post("/logs/stream", json!({ "target": true })).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Invalid value type at `.target`: expected a string, but found a boolean: `true`",
+      "code": "bad_request",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#bad_request"
+    }
+    "###);
+
+    // Wrong type
+    let (response, code) = server.service.post("/logs/stream", json!({ "target": [] })).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Invalid value type at `.target`: expected a string, but found an array: `[]`",
+      "code": "bad_request",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#bad_request"
+    }
+    "###);
+
+    // Our help message
+    let (response, code) = server.service.post("/logs/stream", json!({ "target": "" })).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Invalid value at `.target`: Empty string is not a valid target. If you want to get no logs use `OFF`. Usage: `info`, `meilisearch=info`, or you can write multiple filters in one target: `index_scheduler=info,milli=trace`",
+      "code": "bad_request",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#bad_request"
+    }
+    "###);
+
+    // An error from the target parser
+    let (response, code) = server.service.post("/logs/stream", json!({ "target": "==" })).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Invalid value at `.target`: invalid filter directive: too many '=' in filter directive, expected 0 or 1",
+      "code": "bad_request",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#bad_request"
+    }
+    "###);
+}
+
+#[actix_rt::test]
+async fn logs_stream_bad_mode() {
+    let server = Server::new().await;
+
+    // Wrong type
+    let (response, code) = server.service.post("/logs/stream", json!({ "mode": true })).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Invalid value type at `.mode`: expected a string, but found a boolean: `true`",
+      "code": "bad_request",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#bad_request"
+    }
+    "###);
+
+    // Wrong type
+    let (response, code) = server.service.post("/logs/stream", json!({ "mode": [] })).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Invalid value type at `.mode`: expected a string, but found an array: `[]`",
+      "code": "bad_request",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#bad_request"
+    }
+    "###);
+
+    // Wrong value
+    let (response, code) = server.service.post("/logs/stream", json!({ "mode": "tamo" })).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Unknown value `tamo` at `.mode`: expected one of `human`, `profile`",
+      "code": "bad_request",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#bad_request"
+    }
+    "###);
+}
+
+#[actix_rt::test]
+async fn logs_stream_bad_profile_memory() {
+    let server = Server::new().await;
+
+    // Wrong type
+    let (response, code) =
+        server.service.post("/logs/stream", json!({ "profileMemory": "tamo" })).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Invalid value type at `.profileMemory`: expected a boolean, but found a string: `\"tamo\"`",
+      "code": "bad_request",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#bad_request"
+    }
+    "###);
+
+    // Wrong type
+    let (response, code) =
+        server.service.post("/logs/stream", json!({ "profileMemory": ["hello", "kefir"] })).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Invalid value type at `.profileMemory`: expected a boolean, but found an array: `[\"hello\",\"kefir\"]`",
+      "code": "bad_request",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#bad_request"
+    }
+    "###);
+
+    // Used with default parameters
+    let (response, code) =
+        server.service.post("/logs/stream", json!({ "profileMemory": true })).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Invalid value: `profile_memory` can only be used while profiling code and is not compatible with the Human mode.",
+      "code": "invalid_settings_typo_tolerance",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#invalid_settings_typo_tolerance"
+    }
+    "###);
+
+    // Used with an unsupported mode
+    let (response, code) =
+        server.service.post("/logs/stream", json!({ "mode": "fmt", "profileMemory": true })).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "Unknown value `fmt` at `.mode`: expected one of `human`, `profile`",
+      "code": "bad_request",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#bad_request"
+    }
+    "###);
+}
+
+#[actix_rt::test]
+async fn logs_stream_without_enabling_the_route() {
+    let server = Server::new().await;
+
+    let (response, code) = server.service.post("/logs/stream", json!({})).await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "getting logs through the `/logs/stream` route requires enabling the `logs route` experimental feature. See https://github.com/orgs/meilisearch/discussions/721",
+      "code": "feature_not_enabled",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#feature_not_enabled"
+    }
+    "###);
+
+    let (response, code) = server.service.delete("/logs/stream").await;
+    snapshot!(code, @"400 Bad Request");
+    snapshot!(response, @r###"
+    {
+      "message": "getting logs through the `/logs/stream` route requires enabling the `logs route` experimental feature. See https://github.com/orgs/meilisearch/discussions/721",
+      "code": "feature_not_enabled",
+      "type": "invalid_request",
+      "link": "https://docs.meilisearch.com/errors#feature_not_enabled"
+    }
+    "###);
+}

meilisearch/tests/logs/mod.rs

@@ -0,0 +1,92 @@
+mod error;
+
+use std::rc::Rc;
+use std::str::FromStr;
+
+use actix_web::http::header::ContentType;
+use meili_snap::snapshot;
+use meilisearch::{analytics, create_app, Opt};
+use tracing::level_filters::LevelFilter;
+use tracing_subscriber::layer::SubscriberExt;
+use tracing_subscriber::Layer;
+
+use crate::common::{default_settings, Server};
+use crate::json;
+
+#[actix_web::test]
+async fn basic_test_log_stream_route() {
+    let db_path = tempfile::tempdir().unwrap();
+    let server = Server::new_with_options(Opt {
+        experimental_enable_logs_route: true,
+        ..default_settings(db_path.path())
+    })
+    .await
+    .unwrap();
+
+    let (route_layer, route_layer_handle) =
+        tracing_subscriber::reload::Layer::new(None.with_filter(
+            tracing_subscriber::filter::Targets::new().with_target("", LevelFilter::OFF),
+        ));
+
+    let subscriber = tracing_subscriber::registry().with(route_layer).with(
+        tracing_subscriber::fmt::layer()
+            .with_span_events(tracing_subscriber::fmt::format::FmtSpan::ACTIVE)
+            .with_filter(tracing_subscriber::filter::LevelFilter::from_str("INFO").unwrap()),
+    );
+
+    let app = actix_web::test::init_service(create_app(
+        server.service.index_scheduler.clone().into(),
+        server.service.auth.clone().into(),
+        server.service.options.clone(),
+        route_layer_handle,
+        analytics::MockAnalytics::new(&server.service.options),
+        true,
+    ))
+    .await;
+
+    // set the subscriber as the default for the application
+    tracing::subscriber::set_global_default(subscriber).unwrap();
+
+    let app = Rc::new(app);
+
+    // First, we start listening on the `/logs/stream` route
+    let handle_app = app.clone();
+    let handle = tokio::task::spawn_local(async move {
+        let req = actix_web::test::TestRequest::post()
+            .uri("/logs/stream")
+            .insert_header(ContentType::json())
+            .set_payload(
+                serde_json::to_vec(&json!({
+                    "mode": "human",
+                    "target": "info",
+                }))
+                .unwrap(),
+            );
+        let req = req.to_request();
+        let ret = actix_web::test::call_service(&*handle_app, req).await;
+        actix_web::test::read_body(ret).await
+    });
+
+    // We're going to create an index to get at least one info log saying we processed a batch of task
+    let (ret, _code) = server.create_index(json!({ "uid": "tamo" })).await;
+    snapshot!(ret, @r###"
+    {
+      "taskUid": 0,
+      "indexUid": "tamo",
+      "status": "enqueued",
+      "type": "indexCreation",
+      "enqueuedAt": "[date]"
+    }
+    "###);
+    server.wait_task(ret.uid()).await;
+
+    let req = actix_web::test::TestRequest::delete().uri("/logs/stream");
+    let req = req.to_request();
+    let ret = actix_web::test::call_service(&*app, req).await;
+    let code = ret.status();
+    snapshot!(code, @"204 No Content");
+
+    let logs = handle.await.unwrap();
+    let logs = String::from_utf8(logs.to_vec()).unwrap();
+    assert!(logs.contains("INFO"), "{logs}");
+}

meilisearch/tests/search/hybrid.rs

@@ -13,9 +13,9 @@ async fn index_with_documents<'a>(server: &'a Server, documents: &Value) -> Inde
     meili_snap::snapshot!(code, @"200 OK");
     meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
     {
-      "scoreDetails": false,
       "vectorStore": true,
       "metrics": false,
+      "logsRoute": false,
       "exportPuffinReports": false
     }
     "###);
@@ -87,6 +87,52 @@ async fn simple_search() {
     snapshot!(response["hits"], @r###"[{"title":"Captain Marvel","desc":"a Shazam ersatz","id":"3","_vectors":{"default":[2.0,3.0]},"_semanticScore":0.99029034},{"title":"Captain Planet","desc":"He's not part of the Marvel Cinematic Universe","id":"2","_vectors":{"default":[1.0,2.0]},"_semanticScore":0.97434163},{"title":"Shazam!","desc":"a Captain Marvel ersatz","id":"1","_vectors":{"default":[1.0,3.0]},"_semanticScore":0.9472136}]"###);
 }
 
+#[actix_rt::test]
+async fn highlighter() {
+    let server = Server::new().await;
+    let index = index_with_documents(&server, &SIMPLE_SEARCH_DOCUMENTS).await;
+
+    let (response, code) = index
+        .search_post(json!({"q": "Captain Marvel", "vector": [1.0, 1.0],
+            "hybrid": {"semanticRatio": 0.2},
+            "attributesToHighlight": [
+                     "desc"
+                   ],
+                   "highlightPreTag": "**BEGIN**",
+                   "highlightPostTag": "**END**"
+        }))
+        .await;
+    snapshot!(code, @"200 OK");
+    snapshot!(response["hits"], @r###"[{"title":"Captain Marvel","desc":"a Shazam ersatz","id":"3","_vectors":{"default":[2.0,3.0]},"_formatted":{"title":"Captain Marvel","desc":"a Shazam ersatz","id":"3","_vectors":{"default":["2.0","3.0"]}}},{"title":"Shazam!","desc":"a Captain Marvel ersatz","id":"1","_vectors":{"default":[1.0,3.0]},"_formatted":{"title":"Shazam!","desc":"a **BEGIN**Captain**END** **BEGIN**Marvel**END** ersatz","id":"1","_vectors":{"default":["1.0","3.0"]}}},{"title":"Captain Planet","desc":"He's not part of the Marvel Cinematic Universe","id":"2","_vectors":{"default":[1.0,2.0]},"_formatted":{"title":"Captain Planet","desc":"He's not part of the **BEGIN**Marvel**END** Cinematic Universe","id":"2","_vectors":{"default":["1.0","2.0"]}}}]"###);
+
+    let (response, code) = index
+        .search_post(json!({"q": "Captain Marvel", "vector": [1.0, 1.0],
+            "hybrid": {"semanticRatio": 0.8},
+            "attributesToHighlight": [
+                     "desc"
+                   ],
+                   "highlightPreTag": "**BEGIN**",
+                   "highlightPostTag": "**END**"
+        }))
+        .await;
+    snapshot!(code, @"200 OK");
+    snapshot!(response["hits"], @r###"[{"title":"Captain Marvel","desc":"a Shazam ersatz","id":"3","_vectors":{"default":[2.0,3.0]},"_formatted":{"title":"Captain Marvel","desc":"a Shazam ersatz","id":"3","_vectors":{"default":["2.0","3.0"]}},"_semanticScore":0.99029034},{"title":"Captain Planet","desc":"He's not part of the Marvel Cinematic Universe","id":"2","_vectors":{"default":[1.0,2.0]},"_formatted":{"title":"Captain Planet","desc":"He's not part of the **BEGIN**Marvel**END** Cinematic Universe","id":"2","_vectors":{"default":["1.0","2.0"]}},"_semanticScore":0.97434163},{"title":"Shazam!","desc":"a Captain Marvel ersatz","id":"1","_vectors":{"default":[1.0,3.0]},"_formatted":{"title":"Shazam!","desc":"a **BEGIN**Captain**END** **BEGIN**Marvel**END** ersatz","id":"1","_vectors":{"default":["1.0","3.0"]}},"_semanticScore":0.9472136}]"###);
+
+    // no highlighting on full semantic
+    let (response, code) = index
+        .search_post(json!({"q": "Captain Marvel", "vector": [1.0, 1.0],
+            "hybrid": {"semanticRatio": 1.0},
+            "attributesToHighlight": [
+                     "desc"
+                   ],
+                   "highlightPreTag": "**BEGIN**",
+                   "highlightPostTag": "**END**"
+        }))
+        .await;
+    snapshot!(code, @"200 OK");
+    snapshot!(response["hits"], @r###"[{"title":"Captain Marvel","desc":"a Shazam ersatz","id":"3","_vectors":{"default":[2.0,3.0]},"_formatted":{"title":"Captain Marvel","desc":"a Shazam ersatz","id":"3","_vectors":{"default":["2.0","3.0"]}},"_semanticScore":0.99029034},{"title":"Captain Planet","desc":"He's not part of the Marvel Cinematic Universe","id":"2","_vectors":{"default":[1.0,2.0]},"_formatted":{"title":"Captain Planet","desc":"He's not part of the Marvel Cinematic Universe","id":"2","_vectors":{"default":["1.0","2.0"]}},"_semanticScore":0.97434163},{"title":"Shazam!","desc":"a Captain Marvel ersatz","id":"1","_vectors":{"default":[1.0,3.0]},"_formatted":{"title":"Shazam!","desc":"a Captain Marvel ersatz","id":"1","_vectors":{"default":["1.0","3.0"]}}}]"###);
+}
+
 #[actix_rt::test]
 async fn invalid_semantic_ratio() {
     let server = Server::new().await;

meilisearch/tests/search/mod.rs

@@ -766,38 +766,14 @@ async fn faceting_max_values_per_facet() {
 }
 
 #[actix_rt::test]
-async fn experimental_feature_score_details() {
+async fn test_score_details() {
     let server = Server::new().await;
     let index = server.index("test");
 
     let documents = DOCUMENTS.clone();
 
-    index.add_documents(json!(documents), None).await;
-    index.wait_task(0).await;
-
-    index
-        .search(
-            json!({
-                "q": "train dragon",
-                "showRankingScoreDetails": true,
-            }),
-            |response, code| {
-                meili_snap::snapshot!(code, @"400 Bad Request");
-                meili_snap::snapshot!(meili_snap::json_string!(response), @r###"
-                {
-                  "message": "Computing score details requires enabling the `score details` experimental feature. See https://github.com/meilisearch/product/discussions/674",
-                  "code": "feature_not_enabled",
-                  "type": "invalid_request",
-                  "link": "https://docs.meilisearch.com/errors#feature_not_enabled"
-                }
-                "###);
-            },
-        )
-        .await;
-
-    let (response, code) = server.set_features(json!({"scoreDetails": true})).await;
-    meili_snap::snapshot!(code, @"200 OK");
-    meili_snap::snapshot!(response["scoreDetails"], @"true");
+    let res = index.add_documents(json!(documents), None).await;
+    index.wait_task(res.0.uid()).await;
 
     index
         .search(

meilisearch/tests/tasks/webhook.rs

@@ -44,10 +44,6 @@ struct WebhookHandle {
 }
 
 async fn create_webhook_server() -> WebhookHandle {
-    let mut log_builder = env_logger::Builder::new();
-    log_builder.parse_filters("info");
-    log_builder.init();
-
     let (sender, receiver) = mpsc::unbounded_channel();
     let sender = Arc::new(sender);
 

meilitool/Cargo.toml

@@ -9,11 +9,11 @@ edition.workspace = true
 license.workspace = true
 
 [dependencies]
-anyhow = "1.0.75"
-clap = { version = "4.2.1", features = ["derive"] }
+anyhow = "1.0.79"
+clap = { version = "4.4.17", features = ["derive"] }
 dump = { path = "../dump" }
 file-store = { path = "../file-store" }
 meilisearch-auth = { path = "../meilisearch-auth" }
 meilisearch-types = { path = "../meilisearch-types" }
-time = { version = "0.3.30", features = ["formatting"] }
-uuid = { version = "1.5.0", features = ["v4"], default-features = false }
+time = { version = "0.3.31", features = ["formatting"] }
+uuid = { version = "1.6.1", features = ["v4"], default-features = false }

milli/Cargo.toml

@@ -14,14 +14,14 @@ license.workspace = true
 [dependencies]
 bimap = { version = "0.6.3", features = ["serde"] }
 bincode = "1.3.3"
-bstr = "1.4.0"
-bytemuck = { version = "1.13.1", features = ["extern_crate_alloc"] }
-byteorder = "1.4.3"
+bstr = "1.9.0"
+bytemuck = { version = "1.14.0", features = ["extern_crate_alloc"] }
+byteorder = "1.5.0"
 charabia = { version = "0.8.5", default-features = false }
 concat-arrays = "0.1.2"
-crossbeam-channel = "0.5.8"
-deserr = "0.6.0"
-either = { version = "1.8.1", features = ["serde"] }
+crossbeam-channel = "0.5.11"
+deserr = "0.6.1"
+either = { version = "1.9.0", features = ["serde"] }
 flatten-serde-json = { path = "../flatten-serde-json" }
 fst = "0.4.7"
 fxhash = "0.2.1"
@@ -35,32 +35,32 @@ heed = { version = "0.20.0-alpha.9", default-features = false, features = [
     "serde-bincode",
     "read-txn-no-tls",
 ] }
-indexmap = { version = "2.0.0", features = ["serde"] }
+indexmap = { version = "2.1.0", features = ["serde"] }
 json-depth-checker = { path = "../json-depth-checker" }
 levenshtein_automata = { version = "0.2.1", features = ["fst_automaton"] }
 memmap2 = "0.7.1"
-obkv = "0.2.0"
-once_cell = "1.17.1"
-ordered-float = "3.6.0"
+obkv = "0.2.1"
+once_cell = "1.19.0"
+ordered-float = "4.2.0"
 rand_pcg = { version = "0.3.1", features = ["serde1"] }
-rayon = "1.7.0"
-roaring = "0.10.1"
+rayon = "1.8.0"
+roaring = "0.10.2"
 rstar = { version = "0.11.0", features = ["serde"] }
-serde = { version = "1.0.160", features = ["derive"] }
-serde_json = { version = "1.0.95", features = ["preserve_order"] }
-slice-group-by = "0.3.0"
+serde = { version = "1.0.195", features = ["derive"] }
+serde_json = { version = "1.0.111", features = ["preserve_order"] }
+slice-group-by = "0.3.1"
 smallstr = { version = "0.3.0", features = ["serde"] }
-smallvec = "1.10.0"
+smallvec = "1.12.0"
 smartstring = "1.0.1"
-tempfile = "3.5.0"
-thiserror = "1.0.40"
-time = { version = "0.3.20", features = [
+tempfile = "3.9.0"
+thiserror = "1.0.56"
+time = { version = "0.3.31", features = [
     "serde-well-known",
     "formatting",
     "parsing",
     "macros",
 ] }
-uuid = { version = "1.3.1", features = ["v4"] }
+uuid = { version = "1.6.1", features = ["v4"] }
 
 filter-parser = { path = "../filter-parser" }
 
@@ -71,9 +71,8 @@ itertools = "0.11.0"
 puffin = "0.16.0"
 
 # logging
-log = "0.4.17"
 logging_timer = "1.1.0"
-csv = "1.2.1"
+csv = "1.3.0"
 candle-core = { git = "https://github.com/huggingface/candle.git", version = "0.3.1" }
 candle-transformers = { git = "https://github.com/huggingface/candle.git", version = "0.3.1" }
 candle-nn = { git = "https://github.com/huggingface/candle.git", version = "0.3.1" }
@@ -81,36 +80,29 @@ tokenizers = { git = "https://github.com/huggingface/tokenizers.git", tag = "v0.
 hf-hub = { git = "https://github.com/dureuill/hf-hub.git", branch = "rust_tls", default_features = false, features = [
     "online",
 ] }
-tokio = { version = "1.34.0", features = ["rt"] }
-futures = "0.3.29"
-reqwest = { version = "0.11.16", features = [
+tokio = { version = "1.35.1", features = ["rt"] }
+futures = "0.3.30"
+reqwest = { version = "0.11.23", features = [
     "rustls-tls",
     "json",
 ], default-features = false }
-tiktoken-rs = "0.5.7"
+tiktoken-rs = "0.5.8"
 liquid = "0.26.4"
-arroy = { git = "https://github.com/meilisearch/arroy.git", version = "0.1.0" }
+arroy = "0.2.0"
 rand = "0.8.5"
+tracing = "0.1.40"
 
 [dev-dependencies]
-mimalloc = { version = "0.1.37", default-features = false }
+mimalloc = { version = "0.1.39", default-features = false }
 big_s = "1.0.2"
-insta = "1.29.0"
+insta = "1.34.0"
 maplit = "1.0.2"
 md5 = "0.7.0"
 meili-snap = { path = "../meili-snap" }
 rand = { version = "0.8.5", features = ["small_rng"] }
 
 [features]
-all-tokenizations = [
-    "charabia/chinese",
-    "charabia/hebrew",
-    "charabia/japanese",
-    "charabia/thai",
-    "charabia/korean",
-    "charabia/greek",
-    "charabia/khmer",
-]
+all-tokenizations = ["charabia/chinese", "charabia/hebrew", "charabia/japanese", "charabia/thai", "charabia/korean", "charabia/greek", "charabia/khmer"]
 
 # Use POSIX semaphores instead of SysV semaphores in LMDB
 # For more information on this feature, see heed's Cargo.toml
@@ -137,3 +129,6 @@ greek = ["charabia/greek"]
 
 # allow khmer specialized tokenization
 khmer = ["charabia/khmer"]
+
+# allow CUDA support, see <https://github.com/meilisearch/meilisearch/issues/4306>
+cuda = ["candle-core/cuda"]

milli/src/documents/reader.rs

@@ -25,6 +25,7 @@ impl<R: io::Read + io::Seek> DocumentsBatchReader<R> {
     ///
     /// It first retrieves the index, then moves to the first document. Use the `into_cursor`
     /// method to iterator over the documents, from the first to the last.
+    #[tracing::instrument(level = "trace", skip_all, target = "indexing::documents")]
     pub fn from_reader(reader: R) -> Result<Self, Error> {
         let reader = grenad::Reader::new(reader)?;
         let mut cursor = reader.into_cursor()?;

milli/src/error.rs

@@ -172,7 +172,7 @@ only composed of alphanumeric characters (a-z A-Z 0-9), hyphens (-) and undersco
     TooManyDocumentIds { primary_key: String, document: Object },
     #[error("The primary key inference failed as the engine did not find any field ending with `id` in its name. Please specify the primary key manually using the `primaryKey` query parameter.")]
     NoPrimaryKeyCandidateFound,
-    #[error("The primary key inference failed as the engine found {} fields ending with `id` in their names: '{}' and '{}'. Please specify the primary key manually using the `primaryKey` query parameter.", .candidates.len(), .candidates.get(0).unwrap(), .candidates.get(1).unwrap())]
+    #[error("The primary key inference failed as the engine found {} fields ending with `id` in their names: '{}' and '{}'. Please specify the primary key manually using the `primaryKey` query parameter.", .candidates.len(), .candidates.first().unwrap(), .candidates.get(1).unwrap())]
     MultiplePrimaryKeyCandidatesFound { candidates: Vec<String> },
     #[error("There is no more space left on the device. Consider increasing the size of the disk/partition.")]
     NoSpaceLeftOnDevice,
@@ -227,6 +227,22 @@ only composed of alphanumeric characters (a-z A-Z 0-9), hyphens (-) and undersco
         source_: crate::vector::settings::EmbedderSource,
         embedder_name: String,
     },
+    #[error("`.embedders.{embedder_name}.dimensions`: Model `{model}` does not support overriding its native dimensions of {expected_dimensions}. Found {dimensions}")]
+    InvalidOpenAiModelDimensions {
+        embedder_name: String,
+        model: &'static str,
+        dimensions: usize,
+        expected_dimensions: usize,
+    },
+    #[error("`.embedders.{embedder_name}.dimensions`: Model `{model}` does not support overriding its dimensions to a value higher than {max_dimensions}. Found {dimensions}")]
+    InvalidOpenAiModelDimensionsMax {
+        embedder_name: String,
+        model: &'static str,
+        dimensions: usize,
+        max_dimensions: usize,
+    },
+    #[error("`.embedders.{embedder_name}.dimensions`: `dimensions` cannot be zero")]
+    InvalidSettingsDimensions { embedder_name: String },
 }
 
 impl From<crate::vector::Error> for Error {
@@ -251,6 +267,7 @@ impl From<arroy::Error> for Error {
             arroy::Error::DatabaseFull
             | arroy::Error::InvalidItemAppend
             | arroy::Error::UnmatchingDistance { .. }
+            | arroy::Error::MissingNode
             | arroy::Error::MissingMetadata => {
                 Error::InternalError(InternalError::ArroyError(value))
             }

milli/src/search/facet/facet_sort_ascending.rs

@@ -181,7 +181,7 @@ mod tests {
             get_simple_string_index_with_multiple_field_ids(),
             get_random_looking_string_index_with_multiple_field_ids(),
         ];
-        for (_i, index) in indexes.iter().enumerate() {
+        for index in indexes {
             let txn = index.env.read_txn().unwrap();
             let candidates = RoaringBitmap::new();
             let mut results = String::new();
@@ -212,7 +212,7 @@ mod tests {
             get_simple_string_index_with_multiple_field_ids(),
             get_random_looking_string_index_with_multiple_field_ids(),
         ];
-        for (_i, index) in indexes.iter().enumerate() {
+        for index in indexes {
             let txn = index.env.read_txn().unwrap();
             let candidates = RoaringBitmap::new();
             let mut results = String::new();

milli/src/search/facet/facet_sort_descending.rs

@@ -195,7 +195,7 @@ mod tests {
             get_simple_string_index_with_multiple_field_ids(),
             get_random_looking_string_index_with_multiple_field_ids(),
         ];
-        for (_i, index) in indexes.iter().enumerate() {
+        for index in indexes {
             let txn = index.env.read_txn().unwrap();
             let candidates = RoaringBitmap::new();
             let mut results = String::new();
@@ -226,7 +226,7 @@ mod tests {
             get_simple_string_index_with_multiple_field_ids(),
             get_random_looking_string_index_with_multiple_field_ids(),
         ];
-        for (_i, index) in indexes.iter().enumerate() {
+        for index in indexes {
             let txn = index.env.read_txn().unwrap();
             let candidates = RoaringBitmap::new();
             let mut results = String::new();

milli/src/search/facet/mod.rs

@@ -142,7 +142,7 @@ pub(crate) mod tests {
         let mut txn = index.env.write_txn().unwrap();
         let mut rng = rand::rngs::SmallRng::from_seed([0; 32]);
 
-        for (_i, key) in std::iter::from_fn(|| Some(rng.gen_range(0..256))).take(128).enumerate() {
+        for key in std::iter::from_fn(|| Some(rng.gen_range(0..256))).take(128) {
             let mut bitmap = RoaringBitmap::new();
             bitmap.insert(key);
             bitmap.insert(key + 100);
@@ -172,7 +172,7 @@ pub(crate) mod tests {
         let keys =
             std::iter::from_fn(|| Some(rng.gen_range(0..256))).take(128).collect::<Vec<u32>>();
         for fid in 0..2 {
-            for (_i, &key) in keys.iter().enumerate() {
+            for &key in &keys {
                 let mut bitmap = RoaringBitmap::new();
                 bitmap.insert(key);
                 bitmap.insert(key + 100);
@@ -207,7 +207,7 @@ pub(crate) mod tests {
         let keys =
             std::iter::from_fn(|| Some(rng.gen_range(0..256))).take(128).collect::<Vec<u32>>();
         for fid in 0..2 {
-            for (_i, &key) in keys.iter().enumerate() {
+            for &key in &keys {
                 let mut bitmap = RoaringBitmap::new();
                 bitmap.insert(key);
                 bitmap.insert(key + 100);

milli/src/search/hybrid.rs

@@ -102,7 +102,7 @@ impl ScoreWithRatioResult {
         }
 
         SearchResult {
-            matching_words: left.matching_words,
+            matching_words: right.matching_words,
             candidates: left.candidates | right.candidates,
             documents_ids,
             document_scores,

milli/src/search/mod.rs

@@ -6,9 +6,9 @@ use charabia::Normalize;
 use fst::automaton::{Automaton, Str};
 use fst::{IntoStreamer, Streamer};
 use levenshtein_automata::{LevenshteinAutomatonBuilder as LevBuilder, DFA};
-use log::error;
 use once_cell::sync::Lazy;
 use roaring::bitmap::RoaringBitmap;
+use tracing::error;
 
 pub use self::facet::{FacetDistribution, Filter, OrderBy, DEFAULT_VALUES_PER_FACET};
 pub use self::new::matches::{FormatOptions, MatchBounds, MatcherBuilder, MatchingWords};

milli/src/search/new/bucket_sort.rs

@@ -166,6 +166,9 @@ pub fn bucket_sort<'ctx, Q: RankingRuleQueryTrait>(
             continue;
         }
 
+        let span = tracing::trace_span!(target: "search::bucket_sort", "next_bucket", id = ranking_rules[cur_ranking_rule_index].id());
+        let entered = span.enter();
+
         let Some(next_bucket) = ranking_rules[cur_ranking_rule_index].next_bucket(
             ctx,
             logger,
@@ -175,6 +178,7 @@ pub fn bucket_sort<'ctx, Q: RankingRuleQueryTrait>(
             back!();
             continue;
         };
+        drop(entered);
 
         ranking_rule_scores.push(next_bucket.score);
 

milli/src/update/clear_documents.rs

@@ -14,6 +14,12 @@ impl<'t, 'i> ClearDocuments<'t, 'i> {
         ClearDocuments { wtxn, index }
     }
 
+    #[tracing::instrument(
+        level = "trace",
+        skip(self),
+        target = "indexing::documents",
+        name = "clear_documents"
+    )]
     pub fn execute(self) -> Result<u64> {
         puffin::profile_function!();
 

milli/src/update/facet/incremental.rs

@@ -751,7 +751,7 @@ mod tests {
         let mut rng = rand::rngs::SmallRng::from_seed([0; 32]);
         keys.shuffle(&mut rng);
 
-        for (_i, key) in keys.into_iter().enumerate() {
+        for key in keys {
             let mut bitmap = RoaringBitmap::new();
             bitmap.insert(key);
             index.insert(&mut txn, 0, &(key as f64), &bitmap);
@@ -770,7 +770,7 @@ mod tests {
         let mut rng = rand::rngs::SmallRng::from_seed([0; 32]);
         keys.shuffle(&mut rng);
 
-        for (_i, key) in keys.into_iter().enumerate() {
+        for key in keys {
             let mut bitmap = RoaringBitmap::new();
             bitmap.insert(key);
             bitmap.insert(rng.gen_range(256..512));

milli/src/update/facet/mod.rs

@@ -85,8 +85,8 @@ use charabia::normalizer::{Normalize, NormalizerOption};
 use grenad::{CompressionType, SortAlgorithm};
 use heed::types::{Bytes, DecodeIgnore, SerdeJson};
 use heed::BytesEncode;
-use log::debug;
 use time::OffsetDateTime;
+use tracing::debug;
 
 use self::incremental::FacetsUpdateIncremental;
 use super::FacetsUpdateBulk;

milli/src/update/index_documents/enrich.rs

@@ -22,6 +22,7 @@ use crate::{FieldId, Index, Result};
 /// # Panics
 ///
 /// - if reader.is_empty(), this function may panic in some cases
+#[tracing::instrument(level = "trace", skip_all, target = "indexing::documents")]
 pub fn enrich_documents_batch<R: Read + Seek>(
     rtxn: &heed::RoTxn,
     index: &Index,
@@ -77,7 +78,7 @@ pub fn enrich_documents_batch<R: Read + Seek>(
                 },
                 [] => return Ok(Err(UserError::NoPrimaryKeyCandidateFound)),
                 [(field_id, name)] => {
-                    log::info!("Primary key was not specified in index. Inferred to '{name}'");
+                    tracing::info!("Primary key was not specified in index. Inferred to '{name}'");
                     PrimaryKey::Flat { name, field_id: *field_id }
                 }
                 multiple => {
@@ -143,6 +144,8 @@ pub fn enrich_documents_batch<R: Read + Seek>(
 
 /// Retrieve the document id after validating it, returning a `UserError`
 /// if the id is invalid or can't be guessed.
+#[tracing::instrument(level = "trace", skip(uuid_buffer, documents_batch_index, document)
+target = "indexing::documents")]
 fn fetch_or_generate_document_id(
     document: &obkv::KvReader<FieldId>,
     documents_batch_index: &DocumentsBatchIndex,

milli/src/update/index_documents/extract/extract_docid_word_positions.rs

@@ -21,12 +21,12 @@ pub type ScriptLanguageDocidsMap = HashMap<(Script, Language), (RoaringBitmap, R
 ///
 /// Returns the generated internal documents ids and a grenad reader
 /// with the list of extracted words from the given chunk of documents.
-#[logging_timer::time]
+#[tracing::instrument(level = "trace", skip_all, target = "indexing::extract")]
 pub fn extract_docid_word_positions<R: io::Read + io::Seek>(
     obkv_documents: grenad::Reader<R>,
     indexer: GrenadParameters,
     searchable_fields: &Option<HashSet<FieldId>>,
-    stop_words: Option<&fst::Set<Vec<u8>>>,
+    stop_words: Option<&fst::Set<&[u8]>>,
     allowed_separators: Option<&[&str]>,
     dictionary: Option<&[&str]>,
     max_positions_per_attributes: Option<u32>,
@@ -181,11 +181,11 @@ fn searchable_fields_changed(
 
 /// Factorize tokenizer building.
 fn tokenizer_builder<'a>(
-    stop_words: Option<&'a fst::Set<Vec<u8>>>,
+    stop_words: Option<&'a fst::Set<&[u8]>>,
     allowed_separators: Option<&'a [&str]>,
     dictionary: Option<&'a [&str]>,
     script_language: Option<&'a HashMap<Script, Vec<Language>>>,
-) -> TokenizerBuilder<'a, Vec<u8>> {
+) -> TokenizerBuilder<'a, &'a [u8]> {
     let mut tokenizer_builder = TokenizerBuilder::new();
     if let Some(stop_words) = stop_words {
         tokenizer_builder.stop_words(stop_words);
@@ -211,7 +211,7 @@ fn lang_safe_tokens_from_document<'a>(
     obkv: &KvReader<FieldId>,
     searchable_fields: &Option<HashSet<FieldId>>,
     tokenizer: &Tokenizer,
-    stop_words: Option<&fst::Set<Vec<u8>>>,
+    stop_words: Option<&fst::Set<&[u8]>>,
     allowed_separators: Option<&[&str]>,
     dictionary: Option<&[&str]>,
     max_positions_per_attributes: u32,

milli/src/update/index_documents/extract/extract_facet_number_docids.rs

@@ -16,7 +16,7 @@ use crate::Result;
 ///
 /// Returns a grenad reader with the list of extracted facet numbers and
 /// documents ids from the given chunk of docid facet number positions.
-#[logging_timer::time]
+#[tracing::instrument(level = "trace", skip_all, target = "indexing::extract")]
 pub fn extract_facet_number_docids<R: io::Read + io::Seek>(
     fid_docid_facet_number: grenad::Reader<R>,
     indexer: GrenadParameters,

milli/src/update/index_documents/extract/extract_facet_string_docids.rs

@@ -15,7 +15,7 @@ use crate::{FieldId, Result};
 ///
 /// Returns a grenad reader with the list of extracted facet strings and
 /// documents ids from the given chunk of docid facet string positions.
-#[logging_timer::time]
+#[tracing::instrument(level = "trace", skip_all, target = "indexing::extract")]
 pub fn extract_facet_string_docids<R: io::Read + io::Seek>(
     docid_fid_facet_string: grenad::Reader<R>,
     indexer: GrenadParameters,

milli/src/update/index_documents/extract/extract_fid_docid_facet_values.rs

@@ -39,7 +39,7 @@ pub struct ExtractedFacetValues {
 /// Returns the generated grenad reader containing the docid the fid and the orginal value as key
 /// and the normalized value as value extracted from the given chunk of documents.
 /// We need the fid of the geofields to correctly parse them as numbers if they were sent as strings initially.
-#[logging_timer::time]
+#[tracing::instrument(level = "trace", skip_all, target = "indexing::extract")]
 pub fn extract_fid_docid_facet_values<R: io::Read + io::Seek>(
     obkv_documents: grenad::Reader<R>,
     indexer: GrenadParameters,
@@ -431,7 +431,7 @@ fn extract_facet_values(value: &Value, geo_field: bool) -> FilterableValues {
                     if let Ok(float) = original.parse() {
                         output_numbers.push(float);
                     } else {
-                        log::warn!(
+                        tracing::warn!(
                             "Internal error, could not parse a geofield that has been validated. Please open an issue."
                         )
                     }

milli/src/update/index_documents/extract/extract_fid_word_count_docids.rs

@@ -19,7 +19,7 @@ const MAX_COUNTED_WORDS: usize = 30;
 ///
 /// Returns a grenad reader with the list of extracted field id word counts
 /// and documents ids from the given chunk of docid word positions.
-#[logging_timer::time]
+#[tracing::instrument(level = "trace", skip_all, target = "indexing::extract")]
 pub fn extract_fid_word_count_docids<R: io::Read + io::Seek>(
     docid_word_positions: grenad::Reader<R>,
     indexer: GrenadParameters,

milli/src/update/index_documents/extract/extract_geo_points.rs

@@ -13,7 +13,7 @@ use crate::{FieldId, InternalError, Result};
 /// Extracts the geographical coordinates contained in each document under the `_geo` field.
 ///
 /// Returns the generated grenad reader containing the docid as key associated to the (latitude, longitude)
-#[logging_timer::time]
+#[tracing::instrument(level = "trace", skip_all, target = "indexing::extract")]
 pub fn extract_geo_points<R: io::Read + io::Seek>(
     obkv_documents: grenad::Reader<R>,
     indexer: GrenadParameters,
@@ -34,7 +34,9 @@ pub fn extract_geo_points<R: io::Read + io::Seek>(
         // since we only need the primary key when we throw an error
         // we create this getter to lazily get it when needed
         let document_id = || -> Value {
-            let document_id = obkv.get(primary_key_id).unwrap();
+            let reader = KvReaderDelAdd::new(obkv.get(primary_key_id).unwrap());
+            let document_id =
+                reader.get(DelAdd::Deletion).or(reader.get(DelAdd::Addition)).unwrap();
             serde_json::from_slice(document_id).unwrap()
         };
 

milli/src/update/index_documents/extract/extract_vector_points.rs

@@ -67,7 +67,7 @@ impl VectorStateDelta {
 /// Extracts the embedding vector contained in each document under the `_vectors` field.
 ///
 /// Returns the generated grenad reader containing the docid as key associated to the Vec<f32>
-#[logging_timer::time]
+#[tracing::instrument(level = "trace", skip_all, target = "indexing::extract")]
 pub fn extract_vector_points<R: io::Read + io::Seek>(
     obkv_documents: grenad::Reader<R>,
     indexer: GrenadParameters,
@@ -135,7 +135,7 @@ pub fn extract_vector_points<R: io::Read + io::Seek>(
                 let del_vectors = extract_vectors(old, document_id, embedder_name)?;
                 let add_vectors = extract_vectors(new, document_id, embedder_name)?;
 
-                if add_vectors.len() > u8::MAX.into() {
+                if add_vectors.len() > usize::from(u8::MAX) {
                     return Err(crate::Error::UserError(crate::UserError::TooManyVectors(
                         document_id().to_string(),
                         add_vectors.len(),
@@ -164,7 +164,7 @@ pub fn extract_vector_points<R: io::Read + io::Seek>(
             (None, Some(new)) => {
                 // was possibly autogenerated, remove all vectors for that document
                 let add_vectors = extract_vectors(new, document_id, embedder_name)?;
-                if add_vectors.len() > u8::MAX.into() {
+                if add_vectors.len() > usize::from(u8::MAX) {
                     return Err(crate::Error::UserError(crate::UserError::TooManyVectors(
                         document_id().to_string(),
                         add_vectors.len(),
@@ -186,12 +186,12 @@ pub fn extract_vector_points<R: io::Read + io::Seek>(
                         prompt.render(obkv, DelAdd::Deletion, field_id_map).unwrap_or_default();
                     let new_prompt = prompt.render(obkv, DelAdd::Addition, field_id_map)?;
                     if old_prompt != new_prompt {
-                        log::trace!(
+                        tracing::trace!(
                             "🚀 Changing prompt from\n{old_prompt}\n===to===\n{new_prompt}"
                         );
                         VectorStateDelta::NowGenerated(new_prompt)
                     } else {
-                        log::trace!("⏭️ Prompt unmodified, skipping");
+                        tracing::trace!("⏭️ Prompt unmodified, skipping");
                         VectorStateDelta::NoChange
                     }
                 } else {
@@ -339,9 +339,7 @@ pub fn extract_embeddings<R: io::Read + io::Seek>(
     indexer: GrenadParameters,
     embedder: Arc<Embedder>,
 ) -> Result<grenad::Reader<BufReader<File>>> {
-    let rt = tokio::runtime::Builder::new_current_thread().enable_io().enable_time().build()?;
-
-    let n_chunks = embedder.chunk_count_hint(); // chunk level parellelism
+    let n_chunks = embedder.chunk_count_hint(); // chunk level parallelism
     let n_vectors_per_chunk = embedder.prompt_count_in_chunk_hint(); // number of vectors in a single chunk
 
     // docid, state with embedding
@@ -375,11 +373,8 @@ pub fn extract_embeddings<R: io::Read + io::Seek>(
         current_chunk_ids.push(docid);
 
         if chunks.len() == chunks.capacity() {
-            let chunked_embeds = rt
-                .block_on(
-                    embedder
-                        .embed_chunks(std::mem::replace(&mut chunks, Vec::with_capacity(n_chunks))),
-                )
+            let chunked_embeds = embedder
+                .embed_chunks(std::mem::replace(&mut chunks, Vec::with_capacity(n_chunks)))
                 .map_err(crate::vector::Error::from)
                 .map_err(crate::Error::from)?;
 
@@ -396,8 +391,8 @@ pub fn extract_embeddings<R: io::Read + io::Seek>(
 
     // send last chunk
     if !chunks.is_empty() {
-        let chunked_embeds = rt
-            .block_on(embedder.embed_chunks(std::mem::take(&mut chunks)))
+        let chunked_embeds = embedder
+            .embed_chunks(std::mem::take(&mut chunks))
             .map_err(crate::vector::Error::from)
             .map_err(crate::Error::from)?;
         for (docid, embeddings) in chunks_ids
@@ -410,13 +405,15 @@ pub fn extract_embeddings<R: io::Read + io::Seek>(
     }
 
     if !current_chunk.is_empty() {
-        let embeds = rt
-            .block_on(embedder.embed(std::mem::take(&mut current_chunk)))
+        let embeds = embedder
+            .embed_chunks(vec![std::mem::take(&mut current_chunk)])
             .map_err(crate::vector::Error::from)
             .map_err(crate::Error::from)?;
 
-        for (docid, embeddings) in current_chunk_ids.iter().zip(embeds.iter()) {
-            state_writer.insert(docid.to_be_bytes(), cast_slice(embeddings.as_inner()))?;
+        if let Some(embeds) = embeds.first() {
+            for (docid, embeddings) in current_chunk_ids.iter().zip(embeds.iter()) {
+                state_writer.insert(docid.to_be_bytes(), cast_slice(embeddings.as_inner()))?;
+            }
         }
     }
 

milli/src/update/index_documents/extract/extract_word_docids.rs

@@ -23,7 +23,7 @@ use crate::{DocumentId, FieldId, Result};
 ///
 /// The first returned reader is the one for normal word_docids, and the second one is for
 /// exact_word_docids
-#[logging_timer::time]
+#[tracing::instrument(level = "trace", skip_all, target = "indexing::extract")]
 pub fn extract_word_docids<R: io::Read + io::Seek>(
     docid_word_positions: grenad::Reader<R>,
     indexer: GrenadParameters,
@@ -135,6 +135,7 @@ pub fn extract_word_docids<R: io::Read + io::Seek>(
     ))
 }
 
+#[tracing::instrument(level = "trace", skip_all, target = "indexing::extract")]
 fn words_into_sorter(
     document_id: DocumentId,
     fid: FieldId,

milli/src/update/index_documents/extract/extract_word_pair_proximity_docids.rs

@@ -19,7 +19,7 @@ use crate::{DocumentId, Result};
 ///
 /// Returns a grenad reader with the list of extracted word pairs proximities and
 /// documents ids from the given chunk of docid word positions.
-#[logging_timer::time]
+#[tracing::instrument(level = "trace", skip_all, target = "indexing::extract")]
 pub fn extract_word_pair_proximity_docids<R: io::Read + io::Seek>(
     docid_word_positions: grenad::Reader<R>,
     indexer: GrenadParameters,
@@ -59,6 +59,10 @@ pub fn extract_word_pair_proximity_docids<R: io::Read + io::Seek>(
         if current_document_id.map_or(false, |id| id != document_id) {
             puffin::profile_scope!("Document into sorter");
 
+            // FIXME: span inside of a hot loop might degrade performance and create big reports
+            let span = tracing::trace_span!(target: "indexing::details", "document_into_sorter");
+            let _entered = span.enter();
+
             document_word_positions_into_sorter(
                 current_document_id.unwrap(),
                 &del_word_pair_proximity,
@@ -77,7 +81,7 @@ pub fn extract_word_pair_proximity_docids<R: io::Read + io::Seek>(
                 if let Some(deletion) = KvReaderDelAdd::new(value).get(DelAdd::Deletion) {
                     for (position, word) in KvReaderU16::new(deletion).iter() {
                         // drain the proximity window until the head word is considered close to the word we are inserting.
-                        while del_word_positions.get(0).map_or(false, |(_w, p)| {
+                        while del_word_positions.front().map_or(false, |(_w, p)| {
                             index_proximity(*p as u32, position as u32) >= MAX_DISTANCE
                         }) {
                             word_positions_into_word_pair_proximity(
@@ -106,7 +110,7 @@ pub fn extract_word_pair_proximity_docids<R: io::Read + io::Seek>(
                 if let Some(addition) = KvReaderDelAdd::new(value).get(DelAdd::Addition) {
                     for (position, word) in KvReaderU16::new(addition).iter() {
                         // drain the proximity window until the head word is considered close to the word we are inserting.
-                        while add_word_positions.get(0).map_or(false, |(_w, p)| {
+                        while add_word_positions.front().map_or(false, |(_w, p)| {
                             index_proximity(*p as u32, position as u32) >= MAX_DISTANCE
                         }) {
                             word_positions_into_word_pair_proximity(
@@ -138,6 +142,10 @@ pub fn extract_word_pair_proximity_docids<R: io::Read + io::Seek>(
 
     if let Some(document_id) = current_document_id {
         puffin::profile_scope!("Final document into sorter");
+        // FIXME: span inside of a hot loop might degrade performance and create big reports
+        let span = tracing::trace_span!(target: "indexing::details", "final_document_into_sorter");
+        let _entered = span.enter();
+
         document_word_positions_into_sorter(
             document_id,
             &del_word_pair_proximity,
@@ -147,6 +155,10 @@ pub fn extract_word_pair_proximity_docids<R: io::Read + io::Seek>(
     }
     {
         puffin::profile_scope!("sorter_into_reader");
+        // FIXME: span inside of a hot loop might degrade performance and create big reports
+        let span = tracing::trace_span!(target: "indexing::details", "sorter_into_reader");
+        let _entered = span.enter();
+
         let mut writer = create_writer(
             indexer.chunk_compression_type,
             indexer.chunk_compression_level,

milli/src/update/index_documents/extract/extract_word_position_docids.rs

@@ -18,7 +18,7 @@ use crate::{bucketed_position, DocumentId, Result};
 ///
 /// Returns a grenad reader with the list of extracted words at positions and
 /// documents ids from the given chunk of docid word positions.
-#[logging_timer::time]
+#[tracing::instrument(level = "trace", skip_all, target = "indexing::extract")]
 pub fn extract_word_position_docids<R: io::Read + io::Seek>(
     docid_word_positions: grenad::Reader<R>,
     indexer: GrenadParameters,
@@ -94,6 +94,7 @@ pub fn extract_word_position_docids<R: io::Read + io::Seek>(
     Ok(word_position_docids_reader)
 }
 
+#[tracing::instrument(level = "trace", skip_all, target = "indexing::extract")]
 fn words_position_into_sorter(
     document_id: DocumentId,
     key_buffer: &mut Vec<u8>,

milli/src/update/index_documents/extract/mod.rs

@@ -15,6 +15,7 @@ use std::io::BufReader;
 
 use crossbeam_channel::Sender;
 use rayon::prelude::*;
+use tracing::debug;
 
 use self::extract_docid_word_positions::extract_docid_word_positions;
 use self::extract_facet_number_docids::extract_facet_number_docids;
@@ -28,7 +29,10 @@ use self::extract_vector_points::{
 use self::extract_word_docids::extract_word_docids;
 use self::extract_word_pair_proximity_docids::extract_word_pair_proximity_docids;
 use self::extract_word_position_docids::extract_word_position_docids;
-use super::helpers::{as_cloneable_grenad, CursorClonableMmap, GrenadParameters};
+use super::helpers::{
+    as_cloneable_grenad, merge_deladd_cbo_roaring_bitmaps, CursorClonableMmap, GrenadParameters,
+    MergeFn, MergeableReader,
+};
 use super::{helpers, TypedChunk};
 use crate::proximity::ProximityPrecision;
 use crate::vector::EmbeddingConfigs;
@@ -37,6 +41,7 @@ use crate::{FieldId, FieldsIdsMap, Result};
 /// Extract data for each databases from obkv documents in parallel.
 /// Send data in grenad file over provided Sender.
 #[allow(clippy::too_many_arguments)]
+#[tracing::instrument(level = "trace", skip_all, target = "indexing::extract")]
 pub(crate) fn data_from_obkv_documents(
     original_obkv_chunks: impl Iterator<Item = Result<grenad::Reader<BufReader<File>>>> + Send,
     flattened_obkv_chunks: impl Iterator<Item = Result<grenad::Reader<BufReader<File>>>> + Send,
@@ -47,7 +52,7 @@ pub(crate) fn data_from_obkv_documents(
     primary_key_id: FieldId,
     geo_fields_ids: Option<(FieldId, FieldId)>,
     field_id_map: FieldsIdsMap,
-    stop_words: Option<fst::Set<Vec<u8>>>,
+    stop_words: Option<fst::Set<&[u8]>>,
     allowed_separators: Option<&[&str]>,
     dictionary: Option<&[&str]>,
     max_positions_per_attributes: Option<u32>,
@@ -57,170 +62,227 @@ pub(crate) fn data_from_obkv_documents(
 ) -> Result<()> {
     puffin::profile_function!();
 
-    let (original_pipeline_result, flattened_pipeline_result): (Result<_>, Result<_>) = rayon::join(
-        || {
-            original_obkv_chunks
-                .par_bridge()
-                .map(|original_documents_chunk| {
-                    send_original_documents_data(
-                        original_documents_chunk,
-                        indexer,
-                        lmdb_writer_sx.clone(),
-                        field_id_map.clone(),
-                        embedders.clone(),
-                    )
-                })
-                .collect::<Result<()>>()
-        },
-        || {
-            flattened_obkv_chunks
-                .par_bridge()
-                .map(|flattened_obkv_chunks| {
-                    send_and_extract_flattened_documents_data(
-                        flattened_obkv_chunks,
-                        indexer,
-                        lmdb_writer_sx.clone(),
-                        &searchable_fields,
-                        &faceted_fields,
-                        primary_key_id,
-                        geo_fields_ids,
-                        &stop_words,
-                        &allowed_separators,
-                        &dictionary,
-                        max_positions_per_attributes,
-                    )
-                })
-                .inspect(|result| {
-                    if proximity_precision == ProximityPrecision::ByWord {
-                        if let Ok((docid_word_positions_chunk, _)) = result {
-                            run_extraction_task::<_, _, grenad::Reader<BufReader<File>>>(
-                                docid_word_positions_chunk.clone(),
-                                indexer,
-                                lmdb_writer_sx.clone(),
-                                extract_word_pair_proximity_docids,
-                                TypedChunk::WordPairProximityDocids,
-                                "word-pair-proximity-docids",
-                            );
-                        }
-                    }
-                })
-                .inspect(|result| {
-                    if let Ok((docid_word_positions_chunk, _)) = result {
-                        run_extraction_task::<_, _, grenad::Reader<BufReader<File>>>(
-                            docid_word_positions_chunk.clone(),
-                            indexer,
-                            lmdb_writer_sx.clone(),
-                            extract_fid_word_count_docids,
-                            TypedChunk::FieldIdWordCountDocids,
-                            "field-id-wordcount-docids",
-                        );
-                    }
-                })
-                .inspect(|result| {
-                    if let Ok((docid_word_positions_chunk, _)) = result {
-                        let exact_attributes = exact_attributes.clone();
-                        run_extraction_task::<
-                            _,
-                            _,
-                            (
-                                grenad::Reader<BufReader<File>>,
-                                grenad::Reader<BufReader<File>>,
-                                grenad::Reader<BufReader<File>>,
-                            ),
-                        >(
-                            docid_word_positions_chunk.clone(),
-                            indexer,
-                            lmdb_writer_sx.clone(),
-                            move |doc_word_pos, indexer| {
-                                extract_word_docids(doc_word_pos, indexer, &exact_attributes)
-                            },
-                            |(
-                                word_docids_reader,
-                                exact_word_docids_reader,
-                                word_fid_docids_reader,
-                            )| {
-                                TypedChunk::WordDocids {
-                                    word_docids_reader,
-                                    exact_word_docids_reader,
-                                    word_fid_docids_reader,
-                                }
-                            },
-                            "word-docids",
-                        );
-                    }
-                })
-                .inspect(|result| {
-                    if let Ok((docid_word_positions_chunk, _)) = result {
-                        run_extraction_task::<_, _, grenad::Reader<BufReader<File>>>(
-                            docid_word_positions_chunk.clone(),
-                            indexer,
-                            lmdb_writer_sx.clone(),
-                            extract_word_position_docids,
-                            TypedChunk::WordPositionDocids,
-                            "word-position-docids",
-                        );
-                    }
-                })
-                .inspect(|result| {
-                    if let Ok((_, (_, fid_docid_facet_strings_chunk))) = result {
-                        run_extraction_task::<_, _, grenad::Reader<BufReader<File>>>(
-                            fid_docid_facet_strings_chunk.clone(),
-                            indexer,
-                            lmdb_writer_sx.clone(),
-                            extract_facet_string_docids,
-                            TypedChunk::FieldIdFacetStringDocids,
-                            "field-id-facet-string-docids",
-                        );
-                    }
-                })
-                .inspect(|result| {
-                    if let Ok((_, (fid_docid_facet_numbers_chunk, _))) = result {
-                        run_extraction_task::<_, _, grenad::Reader<BufReader<File>>>(
-                            fid_docid_facet_numbers_chunk.clone(),
-                            indexer,
-                            lmdb_writer_sx.clone(),
-                            extract_facet_number_docids,
-                            TypedChunk::FieldIdFacetNumberDocids,
-                            "field-id-facet-number-docids",
-                        );
-                    }
-                })
-                .map(|r| r.map(|_| ()))
-                .collect::<Result<()>>()
-        },
+    original_obkv_chunks
+        .par_bridge()
+        .map(|original_documents_chunk| {
+            send_original_documents_data(
+                original_documents_chunk,
+                indexer,
+                lmdb_writer_sx.clone(),
+                field_id_map.clone(),
+                embedders.clone(),
+            )
+        })
+        .collect::<Result<()>>()?;
+
+    #[allow(clippy::type_complexity)]
+    let result: Result<(Vec<_>, (Vec<_>, (Vec<_>, (Vec<_>, (Vec<_>, Vec<_>)))))> =
+        flattened_obkv_chunks
+            .par_bridge()
+            .map(|flattened_obkv_chunks| {
+                send_and_extract_flattened_documents_data(
+                    flattened_obkv_chunks,
+                    indexer,
+                    lmdb_writer_sx.clone(),
+                    &searchable_fields,
+                    &faceted_fields,
+                    primary_key_id,
+                    geo_fields_ids,
+                    &stop_words,
+                    &allowed_separators,
+                    &dictionary,
+                    max_positions_per_attributes,
+                )
+            })
+            .collect();
+
+    let (
+        docid_word_positions_chunks,
+        (
+            fid_docid_facet_numbers_chunks,
+            (
+                fid_docid_facet_strings_chunks,
+                (
+                    facet_is_null_docids_chunks,
+                    (facet_is_empty_docids_chunks, facet_exists_docids_chunks),
+                ),
+            ),
+        ),
+    ) = result?;
+
+    // merge facet_exists_docids and send them as a typed chunk
+    {
+        let lmdb_writer_sx = lmdb_writer_sx.clone();
+        rayon::spawn(move || {
+            debug!(database = "facet-id-exists-docids", "merge");
+            match facet_exists_docids_chunks.merge(merge_deladd_cbo_roaring_bitmaps, &indexer) {
+                Ok(reader) => {
+                    let _ = lmdb_writer_sx.send(Ok(TypedChunk::FieldIdFacetExistsDocids(reader)));
+                }
+                Err(e) => {
+                    let _ = lmdb_writer_sx.send(Err(e));
+                }
+            }
+        });
+    }
+
+    // merge facet_is_null_docids and send them as a typed chunk
+    {
+        let lmdb_writer_sx = lmdb_writer_sx.clone();
+        rayon::spawn(move || {
+            debug!(database = "facet-id-is-null-docids", "merge");
+            match facet_is_null_docids_chunks.merge(merge_deladd_cbo_roaring_bitmaps, &indexer) {
+                Ok(reader) => {
+                    let _ = lmdb_writer_sx.send(Ok(TypedChunk::FieldIdFacetIsNullDocids(reader)));
+                }
+                Err(e) => {
+                    let _ = lmdb_writer_sx.send(Err(e));
+                }
+            }
+        });
+    }
+
+    // merge facet_is_empty_docids and send them as a typed chunk
+    {
+        let lmdb_writer_sx = lmdb_writer_sx.clone();
+        rayon::spawn(move || {
+            debug!(database = "facet-id-is-empty-docids", "merge");
+            match facet_is_empty_docids_chunks.merge(merge_deladd_cbo_roaring_bitmaps, &indexer) {
+                Ok(reader) => {
+                    let _ = lmdb_writer_sx.send(Ok(TypedChunk::FieldIdFacetIsEmptyDocids(reader)));
+                }
+                Err(e) => {
+                    let _ = lmdb_writer_sx.send(Err(e));
+                }
+            }
+        });
+    }
+
+    if proximity_precision == ProximityPrecision::ByWord {
+        spawn_extraction_task::<_, _, Vec<grenad::Reader<BufReader<File>>>>(
+            docid_word_positions_chunks.clone(),
+            indexer,
+            lmdb_writer_sx.clone(),
+            extract_word_pair_proximity_docids,
+            merge_deladd_cbo_roaring_bitmaps,
+            TypedChunk::WordPairProximityDocids,
+            "word-pair-proximity-docids",
+        );
+    }
+
+    spawn_extraction_task::<_, _, Vec<grenad::Reader<BufReader<File>>>>(
+        docid_word_positions_chunks.clone(),
+        indexer,
+        lmdb_writer_sx.clone(),
+        extract_fid_word_count_docids,
+        merge_deladd_cbo_roaring_bitmaps,
+        TypedChunk::FieldIdWordCountDocids,
+        "field-id-wordcount-docids",
     );
 
-    original_pipeline_result.and(flattened_pipeline_result)
+    spawn_extraction_task::<
+        _,
+        _,
+        Vec<(
+            grenad::Reader<BufReader<File>>,
+            grenad::Reader<BufReader<File>>,
+            grenad::Reader<BufReader<File>>,
+        )>,
+    >(
+        docid_word_positions_chunks.clone(),
+        indexer,
+        lmdb_writer_sx.clone(),
+        move |doc_word_pos, indexer| extract_word_docids(doc_word_pos, indexer, &exact_attributes),
+        merge_deladd_cbo_roaring_bitmaps,
+        |(word_docids_reader, exact_word_docids_reader, word_fid_docids_reader)| {
+            TypedChunk::WordDocids {
+                word_docids_reader,
+                exact_word_docids_reader,
+                word_fid_docids_reader,
+            }
+        },
+        "word-docids",
+    );
+
+    spawn_extraction_task::<_, _, Vec<grenad::Reader<BufReader<File>>>>(
+        docid_word_positions_chunks.clone(),
+        indexer,
+        lmdb_writer_sx.clone(),
+        extract_word_position_docids,
+        merge_deladd_cbo_roaring_bitmaps,
+        TypedChunk::WordPositionDocids,
+        "word-position-docids",
+    );
+
+    spawn_extraction_task::<_, _, Vec<grenad::Reader<BufReader<File>>>>(
+        fid_docid_facet_strings_chunks,
+        indexer,
+        lmdb_writer_sx.clone(),
+        extract_facet_string_docids,
+        merge_deladd_cbo_roaring_bitmaps,
+        TypedChunk::FieldIdFacetStringDocids,
+        "field-id-facet-string-docids",
+    );
+
+    spawn_extraction_task::<_, _, Vec<grenad::Reader<BufReader<File>>>>(
+        fid_docid_facet_numbers_chunks,
+        indexer,
+        lmdb_writer_sx,
+        extract_facet_number_docids,
+        merge_deladd_cbo_roaring_bitmaps,
+        TypedChunk::FieldIdFacetNumberDocids,
+        "field-id-facet-number-docids",
+    );
+
+    Ok(())
 }
 
 /// Spawn a new task to extract data for a specific DB using extract_fn.
 /// Generated grenad chunks are merged using the merge_fn.
 /// The result of merged chunks is serialized as TypedChunk using the serialize_fn
 /// and sent into lmdb_writer_sx.
-fn run_extraction_task<FE, FS, M>(
-    chunk: grenad::Reader<CursorClonableMmap>,
+fn spawn_extraction_task<FE, FS, M>(
+    chunks: Vec<grenad::Reader<CursorClonableMmap>>,
     indexer: GrenadParameters,
     lmdb_writer_sx: Sender<Result<TypedChunk>>,
     extract_fn: FE,
+    merge_fn: MergeFn,
     serialize_fn: FS,
     name: &'static str,
 ) where
-    FE: Fn(grenad::Reader<CursorClonableMmap>, GrenadParameters) -> Result<M>
+    FE: Fn(grenad::Reader<CursorClonableMmap>, GrenadParameters) -> Result<M::Output>
         + Sync
         + Send
         + 'static,
-    FS: Fn(M) -> TypedChunk + Sync + Send + 'static,
-    M: Send,
+    FS: Fn(M::Output) -> TypedChunk + Sync + Send + 'static,
+    M: MergeableReader + FromParallelIterator<M::Output> + Send + 'static,
+    M::Output: Send,
 {
-    puffin::profile_scope!("extract_chunk", name);
-    match extract_fn(chunk, indexer) {
-        Ok(chunk) => {
-            let _ = lmdb_writer_sx.send(Ok(serialize_fn(chunk)));
-        }
-        Err(e) => {
-            let _ = lmdb_writer_sx.send(Err(e));
-        }
-    }
+    let current_span = tracing::Span::current();
+
+    rayon::spawn(move || {
+        let child_span =
+            tracing::trace_span!(target: "", parent: &current_span, "extract_multiple_chunks");
+        let _entered = child_span.enter();
+        puffin::profile_scope!("extract_multiple_chunksdexing::details, ", name);
+        let chunks: Result<M> =
+            chunks.into_par_iter().map(|chunk| extract_fn(chunk, indexer)).collect();
+        let current_span = tracing::Span::current();
+
+        rayon::spawn(move || match chunks {
+            Ok(chunks) => {
+                let child_span = tracing::trace_span!(target: "", parent: &current_span, "merge_multiple_chunks");
+                let _entered = child_span.enter();
+                debug!(database = name, "merge");
+                puffin::profile_scope!("merge_multiple_chunks", name);
+                let reader = chunks.merge(merge_fn, &indexer);
+                let _ = lmdb_writer_sx.send(reader.map(serialize_fn));
+            }
+            Err(e) => {
+                let _ = lmdb_writer_sx.send(Err(e));
+            }
+        })
+    });
 }
 
 /// Extract chunked data and send it into lmdb_writer_sx sender:
@@ -298,13 +360,22 @@ fn send_and_extract_flattened_documents_data(
     faceted_fields: &HashSet<FieldId>,
     primary_key_id: FieldId,
     geo_fields_ids: Option<(FieldId, FieldId)>,
-    stop_words: &Option<fst::Set<Vec<u8>>>,
+    stop_words: &Option<fst::Set<&[u8]>>,
     allowed_separators: &Option<&[&str]>,
     dictionary: &Option<&[&str]>,
     max_positions_per_attributes: Option<u32>,
 ) -> Result<(
     grenad::Reader<CursorClonableMmap>,
-    (grenad::Reader<CursorClonableMmap>, grenad::Reader<CursorClonableMmap>),
+    (
+        grenad::Reader<CursorClonableMmap>,
+        (
+            grenad::Reader<CursorClonableMmap>,
+            (
+                grenad::Reader<BufReader<File>>,
+                (grenad::Reader<BufReader<File>>, grenad::Reader<BufReader<File>>),
+            ),
+        ),
+    ),
 )> {
     let flattened_documents_chunk =
         flattened_documents_chunk.and_then(|c| unsafe { as_cloneable_grenad(&c) })?;
@@ -375,17 +446,16 @@ fn send_and_extract_flattened_documents_data(
                     fid_docid_facet_strings_chunk.clone(),
                 )));
 
-                let _ = lmdb_writer_sx
-                    .send(Ok(TypedChunk::FieldIdFacetIsNullDocids(fid_facet_is_null_docids_chunk)));
-
-                let _ = lmdb_writer_sx.send(Ok(TypedChunk::FieldIdFacetIsEmptyDocids(
-                    fid_facet_is_empty_docids_chunk,
-                )));
-
-                let _ = lmdb_writer_sx
-                    .send(Ok(TypedChunk::FieldIdFacetExistsDocids(fid_facet_exists_docids_chunk)));
-
-                Ok((fid_docid_facet_numbers_chunk, fid_docid_facet_strings_chunk))
+                Ok((
+                    fid_docid_facet_numbers_chunk,
+                    (
+                        fid_docid_facet_strings_chunk,
+                        (
+                            fid_facet_is_null_docids_chunk,
+                            (fid_facet_is_empty_docids_chunk, fid_facet_exists_docids_chunk),
+                        ),
+                    ),
+                ))
             },
         );
 

milli/src/update/index_documents/helpers/grenad_helpers.rs

@@ -9,6 +9,10 @@ use super::{ClonableMmap, MergeFn};
 use crate::update::index_documents::valid_lmdb_key;
 use crate::Result;
 
+/// This is something reasonable given the fact
+/// that there is one grenad sorter by thread.
+const MAX_GRENAD_SORTER_USAGE: usize = 500 * 1024 * 1024; // 500 MiB
+
 pub type CursorClonableMmap = io::Cursor<ClonableMmap>;
 
 pub fn create_writer<R: io::Write>(
@@ -24,6 +28,9 @@ pub fn create_writer<R: io::Write>(
     builder.build(BufWriter::new(file))
 }
 
+/// A helper function that creates a grenad sorter
+/// with the given parameters. The max memory is
+/// clamped to something reasonable.
 pub fn create_sorter(
     sort_algorithm: grenad::SortAlgorithm,
     merge: MergeFn,
@@ -41,7 +48,7 @@ pub fn create_sorter(
         builder.max_nb_chunks(nb_chunks);
     }
     if let Some(memory) = max_memory {
-        builder.dump_threshold(memory);
+        builder.dump_threshold(memory.min(MAX_GRENAD_SORTER_USAGE));
         builder.allow_realloc(false);
     }
     builder.sort_algorithm(sort_algorithm);
@@ -49,6 +56,7 @@ pub fn create_sorter(
     builder.build()
 }
 
+#[tracing::instrument(level = "trace", skip_all, target = "indexing::grenad")]
 pub fn sorter_into_reader(
     sorter: grenad::Sorter<MergeFn>,
     indexer: GrenadParameters,
@@ -82,6 +90,90 @@ pub unsafe fn as_cloneable_grenad(
     Ok(reader)
 }
 
+pub trait MergeableReader
+where
+    Self: Sized,
+{
+    type Output;
+
+    fn merge(self, merge_fn: MergeFn, indexer: &GrenadParameters) -> Result<Self::Output>;
+}
+
+impl MergeableReader for Vec<grenad::Reader<BufReader<File>>> {
+    type Output = grenad::Reader<BufReader<File>>;
+
+    fn merge(self, merge_fn: MergeFn, params: &GrenadParameters) -> Result<Self::Output> {
+        let mut merger = MergerBuilder::new(merge_fn);
+        self.into_iter().try_for_each(|r| merger.push(r))?;
+        merger.finish(params)
+    }
+}
+
+impl MergeableReader for Vec<(grenad::Reader<BufReader<File>>, grenad::Reader<BufReader<File>>)> {
+    type Output = (grenad::Reader<BufReader<File>>, grenad::Reader<BufReader<File>>);
+
+    fn merge(self, merge_fn: MergeFn, params: &GrenadParameters) -> Result<Self::Output> {
+        let mut m1 = MergerBuilder::new(merge_fn);
+        let mut m2 = MergerBuilder::new(merge_fn);
+        for (r1, r2) in self.into_iter() {
+            m1.push(r1)?;
+            m2.push(r2)?;
+        }
+        Ok((m1.finish(params)?, m2.finish(params)?))
+    }
+}
+
+impl MergeableReader
+    for Vec<(
+        grenad::Reader<BufReader<File>>,
+        grenad::Reader<BufReader<File>>,
+        grenad::Reader<BufReader<File>>,
+    )>
+{
+    type Output = (
+        grenad::Reader<BufReader<File>>,
+        grenad::Reader<BufReader<File>>,
+        grenad::Reader<BufReader<File>>,
+    );
+
+    fn merge(self, merge_fn: MergeFn, params: &GrenadParameters) -> Result<Self::Output> {
+        let mut m1 = MergerBuilder::new(merge_fn);
+        let mut m2 = MergerBuilder::new(merge_fn);
+        let mut m3 = MergerBuilder::new(merge_fn);
+        for (r1, r2, r3) in self.into_iter() {
+            m1.push(r1)?;
+            m2.push(r2)?;
+            m3.push(r3)?;
+        }
+        Ok((m1.finish(params)?, m2.finish(params)?, m3.finish(params)?))
+    }
+}
+
+struct MergerBuilder<R>(grenad::MergerBuilder<R, MergeFn>);
+
+impl<R: io::Read + io::Seek> MergerBuilder<R> {
+    fn new(merge_fn: MergeFn) -> Self {
+        Self(grenad::MergerBuilder::new(merge_fn))
+    }
+
+    fn push(&mut self, reader: grenad::Reader<R>) -> Result<()> {
+        self.0.push(reader.into_cursor()?);
+        Ok(())
+    }
+
+    fn finish(self, params: &GrenadParameters) -> Result<grenad::Reader<BufReader<File>>> {
+        let merger = self.0.build();
+        let mut writer = create_writer(
+            params.chunk_compression_type,
+            params.chunk_compression_level,
+            tempfile::tempfile()?,
+        );
+        merger.write_into_stream_writer(&mut writer)?;
+
+        writer_into_reader(writer)
+    }
+}
+
 #[derive(Debug, Clone, Copy)]
 pub struct GrenadParameters {
     pub chunk_compression_type: CompressionType,
@@ -103,10 +195,15 @@ impl Default for GrenadParameters {
 
 impl GrenadParameters {
     /// This function use the number of threads in the current threadpool to compute the value.
+    ///
     /// This should be called inside of a rayon thread pool,
-    /// Otherwise, it will take the global number of threads.
+    /// otherwise, it will take the global number of threads.
+    ///
+    /// The max memory cannot exceed a given reasonable value.
     pub fn max_memory_by_thread(&self) -> Option<usize> {
-        self.max_memory.map(|max_memory| max_memory / rayon::current_num_threads())
+        self.max_memory.map(|max_memory| {
+            (max_memory / rayon::current_num_threads()).min(MAX_GRENAD_SORTER_USAGE)
+        })
     }
 }
 
@@ -156,6 +253,7 @@ pub fn grenad_obkv_into_chunks<R: io::Read + io::Seek>(
 
 /// Write provided sorter in database using serialize_value function.
 /// merge_values function is used if an entry already exist in the database.
+#[tracing::instrument(level = "trace", skip_all, target = "indexing::grenad")]
 pub fn write_sorter_into_database<K, V, FS, FM>(
     sorter: Sorter<MergeFn>,
     database: &heed::Database<K, V>,

milli/src/update/index_documents/helpers/mod.rs

@@ -10,7 +10,7 @@ use fst::{IntoStreamer, Streamer};
 pub use grenad_helpers::{
     as_cloneable_grenad, create_sorter, create_writer, grenad_obkv_into_chunks,
     merge_ignore_values, sorter_into_reader, write_sorter_into_database, writer_into_reader,
-    GrenadParameters,
+    GrenadParameters, MergeableReader,
 };
 pub use merge_functions::{
     keep_first, keep_latest_obkv, merge_btreeset_string, merge_cbo_roaring_bitmaps,

milli/src/update/index_documents/mod.rs

@@ -5,20 +5,19 @@ mod transform;
 mod typed_chunk;
 
 use std::collections::{HashMap, HashSet};
-use std::io::{Read, Seek};
+use std::io::{Cursor, Read, Seek};
 use std::iter::FromIterator;
 use std::num::NonZeroU32;
 use std::result::Result as StdResult;
 
 use crossbeam_channel::{Receiver, Sender};
-use grenad::{Merger, MergerBuilder};
 use heed::types::Str;
 use heed::Database;
-use log::debug;
 use rand::SeedableRng;
 use roaring::RoaringBitmap;
 use serde::{Deserialize, Serialize};
 use slice_group_by::GroupBy;
+use tracing::debug_span;
 use typed_chunk::{write_typed_chunk_into_index, TypedChunk};
 
 use self::enrich::enrich_documents_batch;
@@ -135,6 +134,7 @@ where
     /// return an error and not the `IndexDocuments` struct as it is invalid to use it afterward.
     ///
     /// Returns the number of documents added to the builder.
+    #[tracing::instrument(level = "trace", skip_all, target = "indexing::documents")]
     pub fn add_documents<R: Read + Seek>(
         mut self,
         reader: DocumentsBatchReader<R>,
@@ -180,6 +180,7 @@ where
     /// Remove a batch of documents from the current builder.
     ///
     /// Returns the number of documents deleted from the builder.
+    #[tracing::instrument(level = "trace", skip_all, target = "indexing::documents")]
     pub fn remove_documents(
         mut self,
         to_delete: Vec<String>,
@@ -215,6 +216,7 @@ where
     /// - No batching using the standards `remove_documents` and `add_documents` took place
     ///
     /// TODO: make it impossible to call `remove_documents` or `add_documents` on an instance that calls this function.
+    #[tracing::instrument(level = "trace", skip_all, target = "indexing::details")]
     pub fn remove_documents_from_db_no_batch(
         mut self,
         to_delete: &RoaringBitmap,
@@ -238,7 +240,12 @@ where
         Ok((self, deleted_documents))
     }
 
-    #[logging_timer::time("IndexDocuments::{}")]
+    #[tracing::instrument(
+        level = "trace"
+        skip_all,
+        target = "indexing::documents",
+        name = "index_documents"
+    )]
     pub fn execute(mut self) -> Result<DocumentAdditionResult> {
         puffin::profile_function!();
 
@@ -274,7 +281,12 @@ where
     }
 
     /// Returns the total number of documents in the index after the update.
-    #[logging_timer::time("IndexDocuments::{}")]
+    #[tracing::instrument(
+        level = "trace",
+        skip_all,
+        target = "profile::indexing::details",
+        name = "index_documents_raw"
+    )]
     pub fn execute_raw(self, output: TransformOutput) -> Result<u64>
     where
         FP: Fn(UpdateIndexingStep) + Sync,
@@ -314,6 +326,9 @@ where
             }
         };
 
+        let original_documents = grenad::Reader::new(original_documents)?;
+        let flattened_documents = grenad::Reader::new(flattened_documents)?;
+
         // create LMDB writer channel
         let (lmdb_writer_sx, lmdb_writer_rx): (
             Sender<Result<TypedChunk>>,
@@ -352,7 +367,11 @@ where
 
         let stop_words = self.index.stop_words(self.wtxn)?;
         let separators = self.index.allowed_separators(self.wtxn)?;
+        let separators: Option<Vec<_>> =
+            separators.as_ref().map(|x| x.iter().map(String::as_str).collect());
         let dictionary = self.index.dictionary(self.wtxn)?;
+        let dictionary: Option<Vec<_>> =
+            dictionary.as_ref().map(|x| x.iter().map(String::as_str).collect());
         let exact_attributes = self.index.exact_attributes_ids(self.wtxn)?;
         let proximity_precision = self.index.proximity_precision(self.wtxn)?.unwrap_or_default();
 
@@ -362,77 +381,59 @@ where
             max_memory: self.indexer_config.max_memory,
             max_nb_chunks: self.indexer_config.max_nb_chunks, // default value, may be chosen.
         };
-        let documents_chunk_size = match self.indexer_config.documents_chunk_size {
-            Some(chunk_size) => chunk_size,
-            None => {
-                let default_chunk_size = 1024 * 1024 * 4; // 4MiB
-                let min_chunk_size = 1024 * 512; // 512KiB
-
-                // compute the chunk size from the number of available threads and the inputed data size.
-                let total_size = flattened_documents.metadata().map(|m| m.len());
-                let current_num_threads = pool.current_num_threads();
-                total_size
-                    .map_or(default_chunk_size, |size| (size as usize) / current_num_threads)
-                    .max(min_chunk_size)
-            }
-        };
-
-        let original_documents = grenad::Reader::new(original_documents)?;
-        let flattened_documents = grenad::Reader::new(flattened_documents)?;
-
+        let documents_chunk_size =
+            self.indexer_config.documents_chunk_size.unwrap_or(1024 * 1024 * 4); // 4MiB
         let max_positions_per_attributes = self.indexer_config.max_positions_per_attributes;
 
         let cloned_embedder = self.embedders.clone();
 
+        let current_span = tracing::Span::current();
+
         // Run extraction pipeline in parallel.
         pool.install(|| {
-            let stop_words = stop_words.map(|sw| sw.map_data(Vec::from).unwrap());
-            rayon::spawn(move || {
-                puffin::profile_scope!("extract_and_send_grenad_chunks");
-                // split obkv file into several chunks
-                let original_chunk_iter =
-                    grenad_obkv_into_chunks(original_documents, pool_params, documents_chunk_size);
+            let child_span = tracing::trace_span!(target: "indexing::details", parent: &current_span, "extract_and_send_grenad_chunks");
+            let _enter = child_span.enter();
+            puffin::profile_scope!("extract_and_send_grenad_chunks");
+            // split obkv file into several chunks
+            let original_chunk_iter =
+                grenad_obkv_into_chunks(original_documents, pool_params, documents_chunk_size);
 
-                // split obkv file into several chunks
-                let flattened_chunk_iter =
-                    grenad_obkv_into_chunks(flattened_documents, pool_params, documents_chunk_size);
+            // split obkv file into several chunks
+            let flattened_chunk_iter =
+                grenad_obkv_into_chunks(flattened_documents, pool_params, documents_chunk_size);
 
-                let separators: Option<Vec<_>> =
-                    separators.as_ref().map(|x| x.iter().map(String::as_str).collect());
-                let dictionary: Option<Vec<_>> =
-                    dictionary.as_ref().map(|x| x.iter().map(String::as_str).collect());
-                let result = original_chunk_iter.and_then(|original_chunk| {
-                    let flattened_chunk = flattened_chunk_iter?;
-                    // extract all databases from the chunked obkv douments
-                    extract::data_from_obkv_documents(
-                        original_chunk,
-                        flattened_chunk,
-                        pool_params,
-                        lmdb_writer_sx.clone(),
-                        searchable_fields,
-                        faceted_fields,
-                        primary_key_id,
-                        geo_fields_ids,
-                        field_id_map,
-                        stop_words,
-                        separators.as_deref(),
-                        dictionary.as_deref(),
-                        max_positions_per_attributes,
-                        exact_attributes,
-                        proximity_precision,
-                        cloned_embedder,
-                    )
-                });
-
-                if let Err(e) = result {
-                    let _ = lmdb_writer_sx.send(Err(e));
-                }
-
-                // needs to be dropped to avoid channel waiting lock.
-                drop(lmdb_writer_sx);
+            let result = original_chunk_iter.and_then(|original_chunk| {
+                let flattened_chunk = flattened_chunk_iter?;
+                // extract all databases from the chunked obkv douments
+                extract::data_from_obkv_documents(
+                    original_chunk,
+                    flattened_chunk,
+                    pool_params,
+                    lmdb_writer_sx.clone(),
+                    searchable_fields,
+                    faceted_fields,
+                    primary_key_id,
+                    geo_fields_ids,
+                    field_id_map,
+                    stop_words,
+                    separators.as_deref(),
+                    dictionary.as_deref(),
+                    max_positions_per_attributes,
+                    exact_attributes,
+                    proximity_precision,
+                    cloned_embedder,
+                )
             });
+
+            if let Err(e) = result {
+                let _ = lmdb_writer_sx.send(Err(e));
+            }
+
+            // needs to be dropped to avoid channel waiting lock.
+            drop(lmdb_writer_sx);
         });
 
+        let index_is_empty = self.index.number_of_documents(self.wtxn)? == 0;
         let mut final_documents_ids = RoaringBitmap::new();
 
         let mut databases_seen = 0;
@@ -460,21 +461,12 @@ where
                     word_fid_docids_reader,
                 } => {
                     let cloneable_chunk = unsafe { as_cloneable_grenad(&word_docids_reader)? };
-                    let word_docids = word_docids.get_or_insert_with(|| {
-                        MergerBuilder::new(merge_deladd_cbo_roaring_bitmaps as MergeFn)
-                    });
-                    word_docids.push(cloneable_chunk.into_cursor()?);
+                    word_docids = Some(cloneable_chunk);
                     let cloneable_chunk =
                         unsafe { as_cloneable_grenad(&exact_word_docids_reader)? };
-                    let exact_word_docids = exact_word_docids.get_or_insert_with(|| {
-                        MergerBuilder::new(merge_deladd_cbo_roaring_bitmaps as MergeFn)
-                    });
-                    exact_word_docids.push(cloneable_chunk.into_cursor()?);
+                    exact_word_docids = Some(cloneable_chunk);
                     let cloneable_chunk = unsafe { as_cloneable_grenad(&word_fid_docids_reader)? };
-                    let word_fid_docids = word_fid_docids.get_or_insert_with(|| {
-                        MergerBuilder::new(merge_deladd_cbo_roaring_bitmaps as MergeFn)
-                    });
-                    word_fid_docids.push(cloneable_chunk.into_cursor()?);
+                    word_fid_docids = Some(cloneable_chunk);
                     TypedChunk::WordDocids {
                         word_docids_reader,
                         exact_word_docids_reader,
@@ -483,10 +475,7 @@ where
                 }
                 TypedChunk::WordPositionDocids(chunk) => {
                     let cloneable_chunk = unsafe { as_cloneable_grenad(&chunk)? };
-                    let word_position_docids = word_position_docids.get_or_insert_with(|| {
-                        MergerBuilder::new(merge_deladd_cbo_roaring_bitmaps as MergeFn)
-                    });
-                    word_position_docids.push(cloneable_chunk.into_cursor()?);
+                    word_position_docids = Some(cloneable_chunk);
                     TypedChunk::WordPositionDocids(chunk)
                 }
                 TypedChunk::VectorPoints {
@@ -509,7 +498,7 @@ where
             };
 
             let (docids, is_merged_database) =
-                write_typed_chunk_into_index(typed_chunk, self.index, self.wtxn)?;
+                write_typed_chunk_into_index(typed_chunk, self.index, self.wtxn, index_is_empty)?;
             if !docids.is_empty() {
                 final_documents_ids |= docids;
                 let documents_seen_count = final_documents_ids.len();
@@ -517,10 +506,7 @@ where
                     documents_seen: documents_seen_count as usize,
                     total_documents: documents_count,
                 });
-                debug!(
-                    "We have seen {} documents on {} total document so far",
-                    documents_seen_count, documents_count
-                );
+                debug_span!("Seen", documents = documents_seen_count, total = documents_count);
             }
             if is_merged_database {
                 databases_seen += 1;
@@ -550,12 +536,8 @@ where
             pool.install(|| {
                 let writer_index = (embedder_index as u16) << 8;
                 for k in 0..=u8::MAX {
-                    let writer = arroy::Writer::prepare(
-                        wtxn,
-                        vector_arroy,
-                        writer_index | (k as u16),
-                        dimension,
-                    )?;
+                    let writer =
+                        arroy::Writer::new(vector_arroy, writer_index | (k as u16), dimension)?;
                     if writer.is_empty(wtxn)? {
                         break;
                     }
@@ -566,22 +548,27 @@ where
         }
 
         self.execute_prefix_databases(
-            word_docids.map(MergerBuilder::build),
-            exact_word_docids.map(MergerBuilder::build),
-            word_position_docids.map(MergerBuilder::build),
-            word_fid_docids.map(MergerBuilder::build),
+            word_docids,
+            exact_word_docids,
+            word_position_docids,
+            word_fid_docids,
         )?;
 
         Ok(number_of_documents)
     }
 
-    #[logging_timer::time("IndexDocuments::{}")]
+    #[tracing::instrument(
+        level = "trace",
+        skip_all,
+        target = "indexing::prefix",
+        name = "index_documents_prefix_databases"
+    )]
     pub fn execute_prefix_databases(
         self,
-        word_docids: Option<Merger<CursorClonableMmap, MergeFn>>,
-        exact_word_docids: Option<Merger<CursorClonableMmap, MergeFn>>,
-        word_position_docids: Option<Merger<CursorClonableMmap, MergeFn>>,
-        word_fid_docids: Option<Merger<CursorClonableMmap, MergeFn>>,
+        word_docids: Option<grenad::Reader<CursorClonableMmap>>,
+        exact_word_docids: Option<grenad::Reader<CursorClonableMmap>>,
+        word_position_docids: Option<grenad::Reader<CursorClonableMmap>>,
+        word_fid_docids: Option<grenad::Reader<CursorClonableMmap>>,
     ) -> Result<()>
     where
         FP: Fn(UpdateIndexingStep) + Sync,
@@ -630,6 +617,8 @@ where
         let del_prefix_fst_words;
 
         {
+            let span = tracing::trace_span!(target: "indexing::details", "compute_prefix_diffs");
+            let _entered = span.enter();
             puffin::profile_scope!("compute_prefix_diffs");
 
             current_prefix_fst = self.index.words_prefixes_fst(self.wtxn)?;
@@ -754,9 +743,15 @@ where
 
 /// Run the word prefix docids update operation.
 #[allow(clippy::too_many_arguments)]
+#[tracing::instrument(
+    level = "trace",
+    skip_all,
+    target = "indexing::prefix",
+    name = "index_documents_word_prefix_docids"
+)]
 fn execute_word_prefix_docids(
     txn: &mut heed::RwTxn,
-    merger: Merger<CursorClonableMmap, MergeFn>,
+    reader: grenad::Reader<Cursor<ClonableMmap>>,
     word_docids_db: Database<Str, CboRoaringBitmapCodec>,
     word_prefix_docids_db: Database<Str, CboRoaringBitmapCodec>,
     indexer_config: &IndexerConfig,
@@ -766,12 +761,13 @@ fn execute_word_prefix_docids(
 ) -> Result<()> {
     puffin::profile_function!();
 
+    let cursor = reader.into_cursor()?;
     let mut builder = WordPrefixDocids::new(txn, word_docids_db, word_prefix_docids_db);
     builder.chunk_compression_type = indexer_config.chunk_compression_type;
     builder.chunk_compression_level = indexer_config.chunk_compression_level;
     builder.max_nb_chunks = indexer_config.max_nb_chunks;
     builder.max_memory = indexer_config.max_memory;
-    builder.execute(merger, new_prefix_fst_words, common_prefix_fst_words, del_prefix_fst_words)?;
+    builder.execute(cursor, new_prefix_fst_words, common_prefix_fst_words, del_prefix_fst_words)?;
     Ok(())
 }
 

milli/src/update/index_documents/transform.rs

@@ -146,7 +146,7 @@ impl<'a, 'i> Transform<'a, 'i> {
         })
     }
 
-    #[logging_timer::time]
+    #[tracing::instrument(level = "trace", skip_all, target = "indexing::documents")]
     pub fn read_documents<R, FP, FA>(
         &mut self,
         reader: EnrichedDocumentsBatchReader<R>,
@@ -359,7 +359,7 @@ impl<'a, 'i> Transform<'a, 'i> {
     /// - If the document to remove was inserted by the `read_documents` method before but was NOT present in the db,
     ///   it's added into the grenad to ensure we don't insert it + removed from the list of new documents ids.
     /// - If the document to remove was not present in either the db or the transform we do nothing.
-    #[logging_timer::time]
+    #[tracing::instrument(level = "trace", skip_all, target = "indexing::documents")]
     pub fn remove_documents<FA>(
         &mut self,
         mut to_remove: Vec<String>,
@@ -450,7 +450,7 @@ impl<'a, 'i> Transform<'a, 'i> {
     /// - No batching using the standards `remove_documents` and `add_documents` took place
     ///
     /// TODO: make it impossible to call `remove_documents` or `add_documents` on an instance that calls this function.
-    #[logging_timer::time]
+    #[tracing::instrument(level = "trace", skip_all, target = "indexing::details")]
     pub fn remove_documents_from_db_no_batch<FA>(
         &mut self,
         to_remove: &RoaringBitmap,
@@ -541,6 +541,7 @@ impl<'a, 'i> Transform<'a, 'i> {
 
     // Flatten a document from the fields ids map contained in self and insert the new
     // created fields. Returns `None` if the document doesn't need to be flattened.
+    #[tracing::instrument(level = "trace", skip(self, obkv), target = "indexing::transform")]
     fn flatten_from_fields_ids_map(&mut self, obkv: KvReader<FieldId>) -> Result<Option<Vec<u8>>> {
         if obkv
             .iter()
@@ -661,7 +662,7 @@ impl<'a, 'i> Transform<'a, 'i> {
     /// Generate the `TransformOutput` based on the given sorter that can be generated from any
     /// format like CSV, JSON or JSON stream. This sorter must contain a key that is the document
     /// id for the user side and the value must be an obkv where keys are valid fields ids.
-    #[logging_timer::time]
+    #[tracing::instrument(level = "trace", skip_all, target = "indexing::transform")]
     pub(crate) fn output_from_sorter<F>(
         self,
         wtxn: &mut heed::RwTxn,

milli/src/update/index_documents/typed_chunk.rs

@@ -7,7 +7,7 @@ use bytemuck::allocation::pod_collect_to_vec;
 use charabia::{Language, Script};
 use grenad::MergerBuilder;
 use heed::types::Bytes;
-use heed::RwTxn;
+use heed::{PutFlags, RwTxn};
 use obkv::{KvReader, KvWriter};
 use roaring::RoaringBitmap;
 
@@ -115,16 +115,20 @@ impl TypedChunk {
 
 /// Write typed chunk in the corresponding LMDB database of the provided index.
 /// Return new documents seen.
+#[tracing::instrument(level = "trace", skip_all, target = "indexing::write_db")]
 pub(crate) fn write_typed_chunk_into_index(
     typed_chunk: TypedChunk,
     index: &Index,
     wtxn: &mut RwTxn,
+    index_is_empty: bool,
 ) -> Result<(RoaringBitmap, bool)> {
     puffin::profile_function!(typed_chunk.to_debug_string());
 
     let mut is_merged_database = false;
     match typed_chunk {
         TypedChunk::Documents(obkv_documents_iter) => {
+            let span = tracing::trace_span!(target: "indexing::write_db", "documents");
+            let _entered = span.enter();
             let mut operations: Vec<DocumentOperation> = Default::default();
 
             let mut docids = index.documents_ids(wtxn)?;
@@ -171,10 +175,14 @@ pub(crate) fn write_typed_chunk_into_index(
             index.put_documents_ids(wtxn, &docids)?;
         }
         TypedChunk::FieldIdWordCountDocids(fid_word_count_docids_iter) => {
-            write_entries_into_database(
+            let span =
+                tracing::trace_span!(target: "indexing::write_db", "field_id_word_count_docids");
+            let _entered = span.enter();
+            append_entries_into_database(
                 fid_word_count_docids_iter,
                 &index.field_id_word_count_docids,
                 wtxn,
+                index_is_empty,
                 deladd_serialize_add_side,
                 merge_deladd_cbo_roaring_bitmaps_into_cbo_roaring_bitmap,
             )?;
@@ -185,29 +193,34 @@ pub(crate) fn write_typed_chunk_into_index(
             exact_word_docids_reader,
             word_fid_docids_reader,
         } => {
+            let span = tracing::trace_span!(target: "indexing::write_db", "word_docids");
+            let _entered = span.enter();
             let word_docids_iter = unsafe { as_cloneable_grenad(&word_docids_reader) }?;
-            write_entries_into_database(
+            append_entries_into_database(
                 word_docids_iter.clone(),
                 &index.word_docids,
                 wtxn,
+                index_is_empty,
                 deladd_serialize_add_side,
                 merge_deladd_cbo_roaring_bitmaps_into_cbo_roaring_bitmap,
             )?;
 
             let exact_word_docids_iter = unsafe { as_cloneable_grenad(&exact_word_docids_reader) }?;
-            write_entries_into_database(
+            append_entries_into_database(
                 exact_word_docids_iter.clone(),
                 &index.exact_word_docids,
                 wtxn,
+                index_is_empty,
                 deladd_serialize_add_side,
                 merge_deladd_cbo_roaring_bitmaps_into_cbo_roaring_bitmap,
             )?;
 
             let word_fid_docids_iter = unsafe { as_cloneable_grenad(&word_fid_docids_reader) }?;
-            write_entries_into_database(
+            append_entries_into_database(
                 word_fid_docids_iter,
                 &index.word_fid_docids,
                 wtxn,
+                index_is_empty,
                 deladd_serialize_add_side,
                 merge_deladd_cbo_roaring_bitmaps_into_cbo_roaring_bitmap,
             )?;
@@ -225,66 +238,93 @@ pub(crate) fn write_typed_chunk_into_index(
             is_merged_database = true;
         }
         TypedChunk::WordPositionDocids(word_position_docids_iter) => {
-            write_entries_into_database(
+            let span = tracing::trace_span!(target: "indexing::write_db", "word_position_docids");
+            let _entered = span.enter();
+            append_entries_into_database(
                 word_position_docids_iter,
                 &index.word_position_docids,
                 wtxn,
+                index_is_empty,
                 deladd_serialize_add_side,
                 merge_deladd_cbo_roaring_bitmaps_into_cbo_roaring_bitmap,
             )?;
             is_merged_database = true;
         }
         TypedChunk::FieldIdFacetNumberDocids(facet_id_number_docids_iter) => {
+            let span =
+                tracing::trace_span!(target: "indexing::write_db","field_id_facet_number_docids");
+            let _entered = span.enter();
             let indexer = FacetsUpdate::new(index, FacetType::Number, facet_id_number_docids_iter);
             indexer.execute(wtxn)?;
             is_merged_database = true;
         }
         TypedChunk::FieldIdFacetStringDocids(facet_id_string_docids_iter) => {
+            let span =
+                tracing::trace_span!(target: "indexing::write_db", "field_id_facet_string_docids");
+            let _entered = span.enter();
             let indexer = FacetsUpdate::new(index, FacetType::String, facet_id_string_docids_iter);
             indexer.execute(wtxn)?;
             is_merged_database = true;
         }
         TypedChunk::FieldIdFacetExistsDocids(facet_id_exists_docids) => {
-            write_entries_into_database(
+            let span =
+                tracing::trace_span!(target: "indexing::write_db", "field_id_facet_exists_docids");
+            let _entered = span.enter();
+            append_entries_into_database(
                 facet_id_exists_docids,
                 &index.facet_id_exists_docids,
                 wtxn,
+                index_is_empty,
                 deladd_serialize_add_side,
                 merge_deladd_cbo_roaring_bitmaps_into_cbo_roaring_bitmap,
             )?;
             is_merged_database = true;
         }
         TypedChunk::FieldIdFacetIsNullDocids(facet_id_is_null_docids) => {
-            write_entries_into_database(
+            let span =
+                tracing::trace_span!(target: "indexing::write_db", "field_id_facet_is_null_docids");
+            let _entered = span.enter();
+            append_entries_into_database(
                 facet_id_is_null_docids,
                 &index.facet_id_is_null_docids,
                 wtxn,
+                index_is_empty,
                 deladd_serialize_add_side,
                 merge_deladd_cbo_roaring_bitmaps_into_cbo_roaring_bitmap,
             )?;
             is_merged_database = true;
         }
         TypedChunk::FieldIdFacetIsEmptyDocids(facet_id_is_empty_docids) => {
-            write_entries_into_database(
+            let span = tracing::trace_span!(target: "profile::indexing::write_db", "field_id_facet_is_empty_docids");
+            let _entered = span.enter();
+            append_entries_into_database(
                 facet_id_is_empty_docids,
                 &index.facet_id_is_empty_docids,
                 wtxn,
+                index_is_empty,
                 deladd_serialize_add_side,
                 merge_deladd_cbo_roaring_bitmaps_into_cbo_roaring_bitmap,
             )?;
             is_merged_database = true;
         }
         TypedChunk::WordPairProximityDocids(word_pair_proximity_docids_iter) => {
-            write_entries_into_database(
+            let span =
+                tracing::trace_span!(target: "indexing::write_db", "word_pair_proximity_docids");
+            let _entered = span.enter();
+            append_entries_into_database(
                 word_pair_proximity_docids_iter,
                 &index.word_pair_proximity_docids,
                 wtxn,
+                index_is_empty,
                 deladd_serialize_add_side,
                 merge_deladd_cbo_roaring_bitmaps_into_cbo_roaring_bitmap,
             )?;
             is_merged_database = true;
         }
         TypedChunk::FieldIdDocidFacetNumbers(fid_docid_facet_number) => {
+            let span =
+                tracing::trace_span!(target: "indexing::write_db", "field_id_docid_facet_numbers");
+            let _entered = span.enter();
             let index_fid_docid_facet_numbers =
                 index.field_id_docid_facet_f64s.remap_types::<Bytes, Bytes>();
             let mut cursor = fid_docid_facet_number.into_cursor()?;
@@ -305,6 +345,9 @@ pub(crate) fn write_typed_chunk_into_index(
             }
         }
         TypedChunk::FieldIdDocidFacetStrings(fid_docid_facet_string) => {
+            let span =
+                tracing::trace_span!(target: "indexing::write_db", "field_id_docid_facet_strings");
+            let _entered = span.enter();
             let index_fid_docid_facet_strings =
                 index.field_id_docid_facet_strings.remap_types::<Bytes, Bytes>();
             let mut cursor = fid_docid_facet_string.into_cursor()?;
@@ -325,6 +368,8 @@ pub(crate) fn write_typed_chunk_into_index(
             }
         }
         TypedChunk::GeoPoints(geo_points) => {
+            let span = tracing::trace_span!(target: "indexing::write_db", "geo_points");
+            let _entered = span.enter();
             let mut rtree = index.geo_rtree(wtxn)?.unwrap_or_default();
             let mut geo_faceted_docids = index.geo_faceted_documents_ids(wtxn)?;
 
@@ -355,6 +400,8 @@ pub(crate) fn write_typed_chunk_into_index(
             expected_dimension,
             embedder_name,
         } => {
+            let span = tracing::trace_span!(target: "indexing::write_db", "vector_points");
+            let _entered = span.enter();
             let embedder_index = index.embedder_category_id.get(wtxn, &embedder_name)?.ok_or(
                 InternalError::DatabaseMissingEntry { db_name: "embedder_category_id", key: None },
             )?;
@@ -362,8 +409,7 @@ pub(crate) fn write_typed_chunk_into_index(
             // FIXME: allow customizing distance
             let writers: std::result::Result<Vec<_>, _> = (0..=u8::MAX)
                 .map(|k| {
-                    arroy::Writer::prepare(
-                        wtxn,
+                    arroy::Writer::new(
                         index.vector_arroy,
                         writer_index | (k as u16),
                         expected_dimension,
@@ -397,7 +443,7 @@ pub(crate) fn write_typed_chunk_into_index(
                             // code error if we somehow got the wrong dimension
                             .unwrap();
 
-                    if embeddings.embedding_count() > u8::MAX.into() {
+                    if embeddings.embedding_count() > usize::from(u8::MAX) {
                         let external_docid = if let Ok(Some(Ok(index))) = index
                             .external_id_of(wtxn, std::iter::once(docid))
                             .map(|it| it.into_iter().next())
@@ -471,9 +517,11 @@ pub(crate) fn write_typed_chunk_into_index(
                 }
             }
 
-            log::debug!("Finished vector chunk for {}", embedder_name);
+            tracing::debug!("Finished vector chunk for {}", embedder_name);
         }
         TypedChunk::ScriptLanguageDocids(sl_map) => {
+            let span = tracing::trace_span!(target: "indexing::write_db", "script_language_docids");
+            let _entered = span.enter();
             for (key, (deletion, addition)) in sl_map {
                 let mut db_key_exists = false;
                 let final_value = match index.script_language_docids.get(wtxn, &key)? {
@@ -527,10 +575,12 @@ fn merge_word_docids_reader_into_fst(
 
 /// Write provided entries in database using serialize_value function.
 /// merge_values function is used if an entry already exist in the database.
+#[tracing::instrument(level = "trace", skip_all, target = "indexing::write_db")]
 fn write_entries_into_database<R, K, V, FS, FM>(
     data: grenad::Reader<R>,
     database: &heed::Database<K, V>,
     wtxn: &mut RwTxn,
+    index_is_empty: bool,
     serialize_value: FS,
     merge_values: FM,
 ) -> Result<()>
@@ -548,9 +598,13 @@ where
     while let Some((key, value)) = cursor.move_on_next()? {
         if valid_lmdb_key(key) {
             buffer.clear();
-            let value = match database.get(wtxn, key)? {
-                Some(prev_value) => merge_values(value, prev_value, &mut buffer)?,
-                None => Some(serialize_value(value, &mut buffer)?),
+            let value = if index_is_empty {
+                Some(serialize_value(value, &mut buffer)?)
+            } else {
+                match database.get(wtxn, key)? {
+                    Some(prev_value) => merge_values(value, prev_value, &mut buffer)?,
+                    None => Some(serialize_value(value, &mut buffer)?),
+                }
             };
             match value {
                 Some(value) => database.put(wtxn, key, value)?,
@@ -563,3 +617,59 @@ where
 
     Ok(())
 }
+
+/// Write provided entries in database using serialize_value function.
+/// merge_values function is used if an entry already exist in the database.
+/// All provided entries must be ordered.
+/// If the index is not empty, write_entries_into_database is called instead.
+#[tracing::instrument(level = "trace", skip_all, target = "indexing::write_db")]
+fn append_entries_into_database<R, K, V, FS, FM>(
+    data: grenad::Reader<R>,
+    database: &heed::Database<K, V>,
+    wtxn: &mut RwTxn,
+    index_is_empty: bool,
+    serialize_value: FS,
+    merge_values: FM,
+) -> Result<()>
+where
+    R: io::Read + io::Seek,
+    FS: for<'a> Fn(&'a [u8], &'a mut Vec<u8>) -> Result<&'a [u8]>,
+    FM: for<'a> Fn(&[u8], &[u8], &'a mut Vec<u8>) -> Result<Option<&'a [u8]>>,
+    K: for<'a> heed::BytesDecode<'a>,
+{
+    puffin::profile_function!(format!("number of entries: {}", data.len()));
+
+    if !index_is_empty {
+        return write_entries_into_database(
+            data,
+            database,
+            wtxn,
+            false,
+            serialize_value,
+            merge_values,
+        );
+    }
+
+    let mut buffer = Vec::new();
+    let mut database = database.iter_mut(wtxn)?.remap_types::<Bytes, Bytes>();
+
+    let mut cursor = data.into_cursor()?;
+    while let Some((key, value)) = cursor.move_on_next()? {
+        if valid_lmdb_key(key) {
+            debug_assert!(
+                K::bytes_decode(key).is_ok(),
+                "Couldn't decode key with the database decoder, key length: {} - key bytes: {:x?}",
+                key.len(),
+                &key
+            );
+            buffer.clear();
+            let value = serialize_value(value, &mut buffer)?;
+            unsafe {
+                // safety: We do not keep a reference to anything that lives inside the database
+                database.put_current_with_options::<Bytes>(PutFlags::APPEND, key, value)?
+            };
+        }
+    }
+
+    Ok(())
+}

milli/src/update/settings.rs

@@ -372,6 +372,11 @@ impl<'a, 't, 'i> Settings<'a, 't, 'i> {
         self.embedder_settings = Setting::Reset;
     }
 
+    #[tracing::instrument(
+        level = "trace"
+        skip(self, progress_callback, should_abort, old_fields_ids_map),
+        target = "indexing::documents"
+    )]
     fn reindex<FP, FA>(
         &mut self,
         progress_callback: &FP,
@@ -974,6 +979,9 @@ impl<'a, 't, 'i> Settings<'a, 't, 'i> {
                             crate::vector::settings::EmbeddingSettings::apply_default_source(
                                 &mut setting,
                             );
+                            crate::vector::settings::EmbeddingSettings::apply_default_openai_model(
+                                &mut setting,
+                            );
                             let setting = validate_embedding_settings(setting, &name)?;
                             changed = true;
                             new_configs.insert(name, setting);
@@ -1119,6 +1127,14 @@ pub fn validate_embedding_settings(
     let Setting::Set(settings) = settings else { return Ok(settings) };
     let EmbeddingSettings { source, model, revision, api_key, dimensions, document_template } =
         settings;
+
+    if let Some(0) = dimensions.set() {
+        return Err(crate::error::UserError::InvalidSettingsDimensions {
+            embedder_name: name.to_owned(),
+        }
+        .into());
+    }
+
     let Some(inferred_source) = source.set() else {
         return Ok(Setting::Set(EmbeddingSettings {
             source,
@@ -1132,14 +1148,34 @@ pub fn validate_embedding_settings(
     match inferred_source {
         EmbedderSource::OpenAi => {
             check_unset(&revision, "revision", inferred_source, name)?;
-            check_unset(&dimensions, "dimensions", inferred_source, name)?;
             if let Setting::Set(model) = &model {
-                crate::vector::openai::EmbeddingModel::from_name(model.as_str()).ok_or(
-                    crate::error::UserError::InvalidOpenAiModel {
+                let model = crate::vector::openai::EmbeddingModel::from_name(model.as_str())
+                    .ok_or(crate::error::UserError::InvalidOpenAiModel {
                         embedder_name: name.to_owned(),
                         model: model.clone(),
-                    },
-                )?;
+                    })?;
+                if let Setting::Set(dimensions) = dimensions {
+                    if !model.supports_overriding_dimensions()
+                        && dimensions != model.default_dimensions()
+                    {
+                        return Err(crate::error::UserError::InvalidOpenAiModelDimensions {
+                            embedder_name: name.to_owned(),
+                            model: model.name(),
+                            dimensions,
+                            expected_dimensions: model.default_dimensions(),
+                        }
+                        .into());
+                    }
+                    if dimensions > model.default_dimensions() {
+                        return Err(crate::error::UserError::InvalidOpenAiModelDimensionsMax {
+                            embedder_name: name.to_owned(),
+                            model: model.name(),
+                            dimensions,
+                            max_dimensions: model.default_dimensions(),
+                        }
+                        .into());
+                    }
+                }
             }
         }
         EmbedderSource::HuggingFace => {

milli/src/update/word_prefix_docids.rs

@@ -39,10 +39,15 @@ impl<'t, 'i> WordPrefixDocids<'t, 'i> {
         }
     }
 
-    #[logging_timer::time("WordPrefixDocids::{}")]
+    #[tracing::instrument(
+        level = "trace",
+        skip_all,
+        target = "indexing::prefix",
+        name = "word_prefix_docids"
+    )]
     pub fn execute(
         self,
-        new_word_docids: grenad::Merger<CursorClonableMmap, MergeFn>,
+        mut new_word_docids_iter: grenad::ReaderCursor<CursorClonableMmap>,
         new_prefix_fst_words: &[String],
         common_prefix_fst_words: &[&[String]],
         del_prefix_fst_words: &HashSet<Vec<u8>>,
@@ -63,8 +68,7 @@ impl<'t, 'i> WordPrefixDocids<'t, 'i> {
         if !common_prefix_fst_words.is_empty() {
             let mut current_prefixes: Option<&&[String]> = None;
             let mut prefixes_cache = HashMap::new();
-            let mut new_word_docids_iter = new_word_docids.into_stream_merger_iter()?;
-            while let Some((word, data)) = new_word_docids_iter.next()? {
+            while let Some((word, data)) = new_word_docids_iter.move_on_next()? {
                 current_prefixes = match current_prefixes.take() {
                     Some(prefixes) if word.starts_with(prefixes[0].as_bytes()) => Some(prefixes),
                     _otherwise => {

milli/src/update/words_prefix_integer_docids.rs

@@ -4,7 +4,7 @@ use std::str;
 use grenad::CompressionType;
 use heed::types::Bytes;
 use heed::{BytesDecode, BytesEncode, Database};
-use log::debug;
+use tracing::debug;
 
 use crate::error::SerializationError;
 use crate::heed_codec::StrBEU16Codec;
@@ -44,10 +44,15 @@ impl<'t, 'i> WordPrefixIntegerDocids<'t, 'i> {
         }
     }
 
-    #[logging_timer::time("WordPrefixIntegerDocids::{}")]
+    #[tracing::instrument(
+        level = "trace",
+        skip_all,
+        target = "indexing::prefix",
+        name = "words_prefix_integer_docids"
+    )]
     pub fn execute(
         self,
-        new_word_integer_docids: grenad::Merger<CursorClonableMmap, MergeFn>,
+        new_word_integer_docids: grenad::Reader<CursorClonableMmap>,
         new_prefix_fst_words: &[String],
         common_prefix_fst_words: &[&[String]],
         del_prefix_fst_words: &HashSet<Vec<u8>>,
@@ -64,14 +69,14 @@ impl<'t, 'i> WordPrefixIntegerDocids<'t, 'i> {
             self.max_memory,
         );
 
+        let mut new_word_integer_docids_iter = new_word_integer_docids.into_cursor()?;
+
         if !common_prefix_fst_words.is_empty() {
             // We fetch all the new common prefixes between the previous and new prefix fst.
             let mut buffer = Vec::new();
             let mut current_prefixes: Option<&&[String]> = None;
             let mut prefixes_cache = HashMap::new();
-            let mut new_word_integer_docids_iter =
-                new_word_integer_docids.into_stream_merger_iter()?;
-            while let Some((key, data)) = new_word_integer_docids_iter.next()? {
+            while let Some((key, data)) = new_word_integer_docids_iter.move_on_next()? {
                 let (word, pos) =
                     StrBEU16Codec::bytes_decode(key).map_err(heed::Error::Decoding)?;
 

milli/src/update/words_prefixes_fst.rs

@@ -38,7 +38,12 @@ impl<'t, 'i> WordsPrefixesFst<'t, 'i> {
         self
     }
 
-    #[logging_timer::time("WordsPrefixesFst::{}")]
+    #[tracing::instrument(
+        level = "trace",
+        skip_all,
+        target = "indexing::prefix",
+        name = "words_prefix_fst"
+    )]
     pub fn execute(self) -> Result<()> {
         puffin::profile_function!();
 

milli/src/vector/error.rs

@@ -67,6 +67,10 @@ pub enum EmbedErrorKind {
     OpenAiUnhandledStatusCode(u16),
     #[error("attempt to embed the following text in a configuration where embeddings must be user provided: {0:?}")]
     ManualEmbed(String),
+    #[error("could not initialize asynchronous runtime: {0}")]
+    OpenAiRuntimeInit(std::io::Error),
+    #[error("initializing web client for sending embedding requests failed: {0}")]
+    InitWebClient(reqwest::Error),
 }
 
 impl EmbedError {
@@ -117,6 +121,14 @@ impl EmbedError {
     pub(crate) fn embed_on_manual_embedder(texts: String) -> EmbedError {
         Self { kind: EmbedErrorKind::ManualEmbed(texts), fault: FaultSource::User }
     }
+
+    pub(crate) fn openai_runtime_init(inner: std::io::Error) -> EmbedError {
+        Self { kind: EmbedErrorKind::OpenAiRuntimeInit(inner), fault: FaultSource::Runtime }
+    }
+
+    pub fn openai_initialize_web_client(inner: reqwest::Error) -> Self {
+        Self { kind: EmbedErrorKind::InitWebClient(inner), fault: FaultSource::Runtime }
+    }
 }
 
 #[derive(Debug, thiserror::Error)]
@@ -183,10 +195,6 @@ impl NewEmbedderError {
         }
     }
 
-    pub fn openai_initialize_web_client(inner: reqwest::Error) -> Self {
-        Self { kind: NewEmbedderErrorKind::InitWebClient(inner), fault: FaultSource::Runtime }
-    }
-
     pub fn openai_invalid_api_key_format(inner: reqwest::header::InvalidHeaderValue) -> Self {
         Self { kind: NewEmbedderErrorKind::InvalidApiKeyFormat(inner), fault: FaultSource::User }
     }
@@ -237,8 +245,6 @@ pub enum NewEmbedderErrorKind {
     #[error("loading model failed: {0}")]
     LoadModel(candle_core::Error),
     // openai
-    #[error("initializing web client for sending embedding requests failed: {0}")]
-    InitWebClient(reqwest::Error),
     #[error("The API key passed to Authorization error was in an invalid format: {0}")]
     InvalidApiKeyFormat(reqwest::header::InvalidHeaderValue),
 }