2005: auto batching r=MarinPostma a=MarinPostma
This pr implements auto batching. The basic functioning of this is that all updates that can be batched together are batched together while the previous batch is being processed.
For now, the only updates that can be batched together are the document addition updates (both update and replace), for a single index.
The batching is disabled by default for multiple reasons:
- We need more experimentation with the scheduling techniques
- Right now, if one task fails in a batch, the whole batch fails. We need more permissive error handling when processing document indexation.
There are four CLI options, for now, to interact with how the batch is scheduled:
- `enable-autobatching`: enable the autobatching feature.
- `debounce-duration-sec`: When an update is received, wait that number of seconds before batching and performing the updates. Defaults to 0s.
- `max-batch-size`: the maximum number of tasks per batch, defaults to unlimited.
- `max-documents-per-batch`: the maximum number of documents in a batch, defaults to unlimited. The batch will always contain a least 1 task, no matter the number of documents in that task.
# Implementation
The current implementation is made of 3 major components:
## TaskStore
The `TaskStore` contains all the tasks. When a task is pushed, it is directly registered to the task store.
## Scheduler
The scheduler is in charge of making the batches. At its core, there is a `TaskQueue` and a job queue. `Job`s are always processed first. They are *volatile* tasks, that is, they don't have a TaskId and are not persisted to disk. Snapshots and dumps are examples of Jobs.
If no `Job` is available for processing, then the scheduler attempts to make a `Task` batch from the `TaskQueue`. The first step is to gather new tasks from the `TaskStore` to populate the `TaskQueue`. When this is done, we can prepare our batch. The `TaskQueue` is itself a `BinaryHeap` of `Tasklist`. Each `index_uid` is associated with a `TaskList` that contains all the updates associated with that index uid. Each `TaskList` in the `TaskQueue` is ordered by the id of its first task.
When preparing a batch, the `TaskList` at the top of the `TaskQueue` is popped, and the tasks are popped from the list to make the next batch. If there are remaining tasks in the list, the list is inserted back in the `TaskQueue`.
## UpdateLoop
The `UpdateLoop` role is to perform batch sequentially. Each time updates are pushed to the update store, the scheduler is notified, and will in turn notify the update loop that work can be performed. When notified, the update loop waits some time to wait for more incoming update and then asks the scheduler for the next batch to perform and perform it. When it is done, the status of the task is put back into the store, and the next batch is processed.
Co-authored-by: mpostma <postma.marin@protonmail.com>
2120: Bring `stable` into `main` r=curquiza a=curquiza
I forgot to do it, tell me `@Kerollmops` or `@irevoire` if it's useful or not. I would say yes, otherwise I will have conflict when I will try to bring `main` into `stable` for the next release. Maybe I'm wrong
Co-authored-by: Irevoire <tamo@meilisearch.com>
Co-authored-by: mpostma <postma.marin@protonmail.com>
Co-authored-by: Tamo <tamo@meilisearch.com>
Co-authored-by: bors[bot] <26634292+bors[bot]@users.noreply.github.com>
Co-authored-by: Clémentine Urquizar - curqui <clementine@meilisearch.com>
2098: feat(dump): Provide the same cli options as the snapshots r=MarinPostma a=irevoire
Add two cli options for the dump:
- `--ignore-missing-dump`
- `--ignore-dump-if-db-exists`
Fix#2087
Co-authored-by: Tamo <tamo@meilisearch.com>
2086: feat(analytics): send the whole set of cli options instead of only the snapshot r=MarinPostma a=irevoire
Fixes#2088
Co-authored-by: Tamo <tamo@meilisearch.com>
2099: feat(analytics): Set the timestamp of the aggregated event as the first aggregate r=MarinPostma a=irevoire
2108: meta(auth): Enhance tests on authorization r=MarinPostma a=ManyTheFish
Enhance auth tests in order to be able to add new actions without changing tests.
Helping #2080
Co-authored-by: Tamo <tamo@meilisearch.com>
Co-authored-by: ManyTheFish <many@meilisearch.com>
2101: chore(all): update actix-web dependency to 4.0.0-beta.21 r=MarinPostma a=robjtede
# Pull Request
## What does this PR do?
I don't expect any more breaking changes to Actix Web that will affect Meilisearch so bump to latest beta.
Fixes #N/A?
<!-- Please link the issue you're trying to fix with this PR, if none then please create an issue first. -->
## PR checklist
Please check if your PR fulfills the following requirements:
- [ ] Does this PR fix an existing issue?
- [x] Have you read the contributing guidelines?
- [x] Have you made sure that the title is accurate and descriptive of the changes?
Thank you so much for contributing to MeiliSearch!
Co-authored-by: Rob Ede <robjtede@icloud.com>
2095: feat(error): Update the error message when you have no version file r=MarinPostma a=irevoire
Following this [issue](https://github.com/meilisearch/meilisearch-kubernetes/issues/95) we decided to change the error message from:
```
Version file is missing or the previous MeiliSearch engine version was below 0.24.0. Use a dump to update MeiliSearch.
```
to
```
Version file is missing or the previous MeiliSearch engine version was below 0.25.0. Use a dump to update MeiliSearch.
```
Co-authored-by: Tamo <tamo@meilisearch.com>
2075: Allow payloads with no documents r=irevoire a=MarinPostma
accept addition with 0 documents.
0 bytes payload are still refused, since they are not valid json/jsonlines/csv anyways...
close#1987
Co-authored-by: mpostma <postma.marin@protonmail.com>
2068: chore(http): migrate from structopt to clap3 r=Kerollmops a=MarinPostma
migrate from structopt to clap3
This fix the long lasting issue with flags require a value, such as `--no-analytics` or `--schedule-snapshot`.
All flag arguments now take NO argument, i.e:
`meilisearch --schedule-snapshot true` becomes `meilisearch --schedule-snapshot`
as per https://docs.rs/clap/latest/clap/struct.Arg.html#method.env, the env variable is defines as:
> A false literal is n, no, f, false, off or 0. An absent environment variable will also be considered as false. Anything else will considered as true.
`@gmourier`
`@curquiza`
`@meilisearch/docs-team`
Co-authored-by: mpostma <postma.marin@protonmail.com>
2084: bump milli r=Kerollmops a=irevoire
- Fix https://github.com/meilisearch/MeiliSearch/issues/2082 by updating milli dependency
- Fix Clippy error
- Change the MeiliSearch version in the cargo.toml to anticipate the coming release (v0.25.2)
Co-authored-by: Tamo <tamo@meilisearch.com>
2076: fix(dump): Fix the import of dump from the v24 and before r=ManyTheFish a=irevoire
Same as https://github.com/meilisearch/MeiliSearch/pull/2073 but on main this time
Co-authored-by: Irevoire <tamo@meilisearch.com>
2066: bug(http): fix task duration r=MarinPostma a=MarinPostma
`@gmourier` found that the duration in the task view was not computed correctly, this pr fixes it.
`@curquiza,` I let you decide if we need to make a hotfix out of this or wait for the next release. This is not breaking.
Co-authored-by: mpostma <postma.marin@protonmail.com>
2057: fix(dump): Uncompress the dump IN the data.ms r=irevoire a=irevoire
When loading a dump with docker, we had two problems.
After creating a tempdirectory, uncompressing and re-indexing the dump:
1. We try to `move` the new “data.ms” onto the currently present
one. The problem is that if the `data.ms` is a mount point because
that's what peoples do with docker usually. We can't override
a mount point, and thus we were throwing an error.
2. The tempdir is created in `/tmp`, which is usually quite small AND may not
be on the same partition as the `data.ms`. This means when we tried to move
the dump over the `data.ms`, it was also failing because we can't move data
between two partitions.
------------------
1 was fixed by deleting the *content* of the `data.ms` and moving the *content*
of the tempdir *inside* the `data.ms`. If someone tries to create volumes inside
the `data.ms` that's his problem, not ours.
2 was fixed by creating the tempdir *inside* of the `data.ms`. If a user mounted
its `data.ms` on a large partition, there is no reason he could not load a big
dump because his `/tmp` was too small. This solves the issue; now the dump is
extracted and indexed on the same partition the `data.ms` will lay.
fix#1833
Co-authored-by: Tamo <tamo@meilisearch.com>
