Initial prototype of Ollama embeddings actually working, error handlign / retries still missing.
Allow model to be any String and require dimensions parameter
Fixed rustfmt formatting issues
There were some formatting issues in the initial PR and this should not make the changes comply with the Rust style guidelines
Because I accidentally didn't follow the style guide for commits in my commit messages I squashed them into one to comply
4453: Don't test on nightly r=dureuill a=dureuill
# Pull Request
## Related issue
Fixes#4441 better 😅
## What does this PR do?
- No longer run tests on nightly
The motivation for this change is that we are now updating Rust at fixed points in time, and so no longer need nightly runs to ensure that a change won't get into stable and break our build at the worst possible moment.
Co-authored-by: Louis Dureuil <louis@meilisearch.com>
4451: Fix nightly build r=dureuill a=dureuill
# Pull Request
## Related issue
Fixes#4441
## What does this PR do?
- Change imports following https://github.com/rust-lang/rust/pull/117772
## Note
This one is going to be annoying a bit until the lint stabilizes:
- We only get the warning on nightly, so we will discover them when it runs in the CI that uses the nightly compiler (not on regular PRs)
- There's the case of `TryInto`/`TryFrom` traits. They have been added to the prelude in Rust edition 2021, so it means that `use`ing them is a warning on nightly for 2021 edition crates (most crates), but not `use`ing them is an error anywhere for 2018 Rust edition crates, such as `milli`
Co-authored-by: Louis Dureuil <louis@meilisearch.com>
4433: Enhance facet incremental r=Kerollmops a=ManyTheFish
# Pull Request
## Related issue
Fixes#4367Fixes#4409
## What does this PR do?
- Add a test reproducing #4409
- Fix#4409 by removing a document from a level only if it is no more present in all the linked sub-level nodes
- Optimize facet Incremental indexing by creating or deleting a complete level once per field id instead of for each facet value
- Optimize facet Incremental indexing by doing the additions and the deletions in the same process instead of doing them separately
Co-authored-by: ManyTheFish <many@meilisearch.com>
4446: Do not omit vectors when importing a dump r=irevoire a=dureuill
# Pull Request
## Related issue
Fixes#4447
## What does this PR do?
- Correctly populate the maps of embedders before starting the indexing operations, while importing a dump
Co-authored-by: Louis Dureuil <louis@meilisearch.com>
4435: Make update file deletion atomic r=Kerollmops a=irevoire
# Pull Request
## Related issue
Fixes https://github.com/meilisearch/meilisearch/issues/4432
Fixes https://github.com/meilisearch/meilisearch/issues/4438 by adding the logs the user asked
## What does this PR do?
- Adds a bunch of logs to help debug this kind of issue in the future
- Delete the update files AFTER committing the update in the `index-scheduler` (thus, if a restart happens, we are able to re-process the batch successfully)
- Multi-thread the deletion of all update files.
Co-authored-by: Tamo <tamo@meilisearch.com>
4443: Add GPU analytics r=dureuill a=dureuill
# Pull Request
## Related issue
Adds analytics indicating whether Meilisearch was compiled with the `milli/cuda` feature.
Cc `@macraig`
Co-authored-by: Louis Dureuil <louis@meilisearch.com>
4442: Send custom task r=ManyTheFish a=irevoire
This PR has already been merged on main but was supposed to be merged on `release-v1.7.0` thus we need to merge it a second time; sorry 😓
### This PR implements the necessary parameters for the High Availability
Introduce a new CLI flag called `--experimental-replication-parameters` that changes a few behaviors in the engine:
- [The auto-deletion of tasks is disabled](https://specs.meilisearch.com/specifications/text/0060-tasks-api.html#_2-technical-details)
- Upon registering a task, you can choose its task ID by sending a new header: `TaskId: 456645`. It must be a valid number, which must be superior to the last task id ever seen.
- Add the ability to « dry-register » a task. That means meilisearch will answer to you with a valid task ID like everything went well, but won’t actually write anything in the database. To do that, you need to use the `DryRun: true` header.
- Specification’s here: https://github.com/meilisearch/specifications/pull/266
Co-authored-by: Tamo <tamo@meilisearch.com>
4042: Implements the new replication parameters r=ManyTheFish a=irevoire
### This PR implements the necessary parameters for the High Availability
- [ ] Update the spec
Introduce a new CLI flag called `--experimental-replication-parameters` that changes a few behaviors in the engine:
- [The auto-deletion of tasks is disabled](https://specs.meilisearch.com/specifications/text/0060-tasks-api.html#_2-technical-details)
- Upon registering a task, you can choose its task ID by sending a new header: `TaskId: 456645`. It must be a valid number, which must be superior to the last task id ever seen.
- Add the ability to « dry-register » a task. That means meilisearch will answer to you with a valid task ID like everything went well, but won’t actually write anything in the database. To do that, you need to use the `DryRun: true` header.
----
Old prototype `prototype-custom-task-id-0`:
- Adds the capability to specify your own task ID via the `TaskId` http header
- Make the task IDs a u64 instead of a u32
Co-authored-by: Tamo <tamo@meilisearch.com>
4418: Output logs to stderr r=dureuill a=irevoire
Output the logs to `stderr` instead of `stdout`. This was introduced in the `v1.7.0-rc.0` and is a bug; logs should always be outputted to stderr.
Fix#4419
Co-authored-by: Tamo <tamo@meilisearch.com>
4410: Implement the experimental log mode cli flag and log level updates at runtime r=dureuill a=irevoire
# Pull Request
This PR fixes two issues at once because they’re highly correlated in the codebase.
## Related issue
Fixes https://github.com/meilisearch/meilisearch/issues/4415
Fixes https://github.com/meilisearch/meilisearch/issues/4413
## What does this PR do?
- It makes the fmt logger configurable to output json or human-readable logs (like we already do today)
- It moves the fmt logger under a `reload` layer so we can update its targets at runtime
- Add the possibility to stream logs in the json mode
- Adds an analytics for the new CLI flag
Co-authored-by: Tamo <tamo@meilisearch.com>
4350: Make several indexing optimizations r=Kerollmops a=ManyTheFish
# Summary
Implement several enhancements to reduce the indexing time.
# Steps
- Compute the indexing chunk size dynamically based on the available threads and the data size
- Remove the merging step before the writing step and merge at the writing time
- Remove append function
- Make Facet search indexing incremental
# Running Indexing process
## `main`
Each type of data is written after a merging phase:
> Highlighted parts are the writings
## `remove-merging-phase-from-indexing`
When the extraction of a chunk is finished, the data is written:
> Highlighted parts are the writings
## Related
This PR removes the appending writes on several indexing parts, which may fix https://github.com/meilisearch/meilisearch/issues/4300. However, all of the appending writes are not removed. There are 2 remaining calls that could trigger this bug:
- When [putting embedders in the settings](b6fc181993/milli/src/update/settings.rs (L996))
- when [bulk indexing the facets](b6fc181993/milli/src/update/facet/bulk.rs (L150))
Co-authored-by: ManyTheFish <many@meilisearch.com>
Co-authored-by: Louis Dureuil <louis@meilisearch.com>
Co-authored-by: Many the fish <many@meilisearch.com>
