Use unstable arroy version containing incremental indexing

make meilisearch accept cancelation tasks even when the disk is full

Cargo.lock

@@ -394,8 +394,7 @@ checksum = "96d30a06541fbafbc7f82ed10c06164cfbd2c401138f6addd8404629c4b16711"
 [[package]]
 name = "arroy"
 version = "0.6.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "08e6111f351d004bd13e95ab540721272136fd3218b39d3ec95a2ea1c4e6a0a6"
+source = "git+https://github.com/meilisearch/arroy.git?rev=5b748bac2c69c65a97980901b02067a3a545e357#5b748bac2c69c65a97980901b02067a3a545e357"
 dependencies = [
  "bytemuck",
  "byteorder",

crates/index-scheduler/src/lib.rs

@@ -625,8 +625,8 @@ impl IndexScheduler {
         task_id: Option<TaskId>,
         dry_run: bool,
     ) -> Result<Task> {
-        // if the task doesn't delete anything and 50% of the task queue is full, we must refuse to enqueue the incomming task
-        if !matches!(&kind, KindWithContent::TaskDeletion { tasks, .. } if !tasks.is_empty())
+        // if the task doesn't delete or cancel anything and 40% of the task queue is full, we must refuse to enqueue the incoming task
+        if !matches!(&kind, KindWithContent::TaskDeletion { tasks, .. } | KindWithContent::TaskCancelation { tasks, .. } if !tasks.is_empty())
             && (self.env.non_free_pages_size()? * 100) / self.env.info().map_size as u64 > 40
         {
             return Err(Error::NoSpaceLeftInTaskQueue);

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

@@ -292,8 +292,6 @@ impl Queue {
             return Ok(task);
         }
 
-        // Get rid of the mutability.
-        let task = task;
         self.tasks.register(wtxn, &task)?;
 
         Ok(task)

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

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

crates/milli/Cargo.toml

@@ -87,7 +87,8 @@ rhai = { git = "https://github.com/rhaiscript/rhai", rev = "ef3df63121d27aacd838
     "no_time",
     "sync",
 ] }
-arroy = "0.6.1"
+# arroy = "0.6.1"
+arroy = { git = "https://github.com/meilisearch/arroy.git", rev = "5b748bac2c69c65a97980901b02067a3a545e357" } # incremental update
 rand = "0.8.5"
 tracing = "0.1.41"
 ureq = { version = "2.12.1", features = ["json"] }

crates/milli/src/progress.rs

@@ -266,12 +266,15 @@ impl Step for arroy::MainStep {
                 "writing the descendants and metadata"
             }
             arroy::MainStep::RetrieveTheUpdatedItems => "retrieve the updated items",
-            arroy::MainStep::RetrievingTheTreeAndItemNodes => "retrieving the tree and item nodes",
-            arroy::MainStep::UpdatingTheTrees => "updating the trees",
-            arroy::MainStep::CreateNewTrees => "create new trees",
-            arroy::MainStep::WritingNodesToDatabase => "writing nodes to database",
-            arroy::MainStep::DeleteExtraneousTrees => "delete extraneous trees",
             arroy::MainStep::WriteTheMetadata => "write the metadata",
+            arroy::MainStep::RetrievingTheItemsIds => "retrieving the items ids",
+            arroy::MainStep::RetrievingTheUsedTreeNodes => "retrieving the used tree nodes",
+            arroy::MainStep::DeletingExtraTrees => "deleting extra trees",
+            arroy::MainStep::RemoveItemsFromExistingTrees => "remove items from existing trees",
+            arroy::MainStep::InsertItemsInCurrentTrees => "insert items in current trees",
+            arroy::MainStep::IncrementalIndexLargeDescendants => {
+                "incremental index large descendants"
+            }
         }
         .into()
     }