1
0
mirror of https://github.com/meilisearch/MeiliSearch synced 2025-03-20 06:30:38 +01:00

319 Commits

Author SHA1 Message Date
meili-bors[bot]
885710a07b
Merge
5341: Embeddings stats r=ManyTheFish a=ManyTheFish

# Pull Request

## Related issue
Fixes 

## What does this PR do?
- Add embedding stats
- force dumpless upgrade to recompute stats
- add tests


Co-authored-by: ManyTheFish <many@meilisearch.com>
2025-02-12 15:46:37 +00:00
ManyTheFish
c55fdad2c3 Fix dumpless upgrade target version 2025-02-12 16:35:05 +01:00
ManyTheFish
8419ed52a1 fix clippy 2025-02-12 14:38:51 +01:00
Louis Dureuil
8e0d8d31f9
Add back timeout from v1.11.3 2025-02-12 11:53:00 +01:00
ManyTheFish
bd27fe7d02 force dumpless upgrade to recompute stats 2025-02-12 11:45:02 +01:00
ManyTheFish
41203f0931 Add embedders stats 2025-02-12 11:37:47 +01:00
Louis Dureuil
b83275c9c5
Change the updated* functions to only_new functions, hopefully better communicating what they do 2025-02-11 15:27:10 +01:00
Louis Dureuil
d7f35ee3ba
Use merged document instead of updated 2025-02-11 15:27:10 +01:00
meili-bors[bot]
0c3e7fe963
Merge
5316: Fix the dumpless upgrade corruption r=dureuill a=irevoire

# Pull Request

## Related issue
Fixes https://github.com/meilisearch/meilisearch/issues/5280

## What does this PR do?
- Add a test that ensure we write the version in the index-scheduler even if we have a bug while writing the VERSION file
- Do what was described in the issue


Co-authored-by: Tamo <tamo@meilisearch.com>
2025-02-10 09:53:57 +00:00
Tamo
45f843ccb9 fmt 2025-02-10 10:46:42 +01:00
Kerollmops
2b0e17ede0
Make sure arroy is using the rayon thread-pool 2025-02-06 15:28:10 +01:00
Louis Dureuil
04ac0af54b
Add WeightedScoreValues to be able to compare remote scores 2025-02-05 15:03:16 +01:00
Louis Dureuil
9996533364
Make search types serialize and deserialize so that reading from a proxy is possible 2025-02-05 15:03:16 +01:00
meili-bors[bot]
796acd1aee
Merge
5288: Improve AI logging r=dureuill a=Kerollmops

This PR fixes  and brings the changes from  to simplify debugging indexation and search performance issues related to AI. The following texts can be found in the logs to debug and understand performance issues:

 - `embed_one: search` represents the time we spent waiting for the embedding generation, i.e., OpenAI, local HuggingFace, Ollama.
 - `filtered_universe: search::universe` the time spent filtering the documents.
 - ~`next_bucket: search::vector_sort` is the time spent finding the nearest neighbors (ANNs) in the vector store (arroy), locally~ was being triggered too many times.
 - `indexing::vectors` is the time arroy spends indexing the new vectors for a batch.
 - `documents::extract vectors` and `documents::merge vectors` to see the time spent generating and writing the embeddings.

Co-authored-by: Kerollmops <clement@meilisearch.com>
2025-02-04 10:20:45 +00:00
Kerollmops
cc8df5e11f
Move back the search-side logging to tracing 2025-02-04 11:16:17 +01:00
meili-bors[bot]
ede74ccc42
Merge
5306: Fix internal error when passing `documentTemplateMaxBytes` to a source that doesn't support it r=ManyTheFish a=dureuill

# Pull Request

## Related issue
Fixes  

## What does this PR do?
- add `DOCUMENT_TEMPLATE_MAX_BYTES` to `allowed_sources_for_field` and `allowed_fields_for_source` to prevent a panic


Co-authored-by: Louis Dureuil <louis@meilisearch.com>
2025-02-04 08:46:13 +00:00
Kerollmops
7a9382b115
Better document the rayon limitation condition 2025-02-03 10:24:53 +01:00
Kerollmops
62dabeba5f
Do not create too many rayon tasks when processing the settings 2025-02-03 10:24:52 +01:00
Kerollmops
48812229a9
Remove a log that would log too much 2025-02-03 10:24:52 +01:00
Louis Dureuil
96544bfa43
add DOCUMENT_TEMPLATE_MAX_BYTES to allowed_sources_for_field and allowed_fields_for_source 2025-02-03 09:59:17 +01:00
Kerollmops
aaefbfae1f
Do not create too many rayon tasks 2025-01-30 16:36:12 +01:00
Kerollmops
97e17f52a1
Add more logs to see calls to the embedders 2025-01-30 16:36:12 +01:00
Kerollmops
424c5bde40
Move the embedding computation and extraction log to debug 2025-01-29 16:40:36 +01:00
Kerollmops
cb1b7513af
Log the memory metrics only once 2025-01-29 15:21:52 +01:00
Clément Renault
a9d0f4a002
Improve english comments
Co-authored-by: Louis Dureuil <louis@meilisearch.com>
2025-01-29 15:16:40 +01:00
Kerollmops
db032079d8
Show indexation allocated memory 2025-01-29 14:21:02 +01:00
Clément Renault
a00796c46a
Improve the naming in the log message 2025-01-29 14:21:02 +01:00
Kerollmops
6112bd8caa
Display the channel congestion 2025-01-29 14:21:02 +01:00
Kerollmops
cec88cfc29
Measure the bbqueue congestion 2025-01-29 14:21:02 +01:00
Kerollmops
19bc885b07
Fix the milli logo 2025-01-27 14:30:59 +01:00
Kerollmops
47f70e3d79
Debug the first vector sort fill buffer 2025-01-27 14:22:29 +01:00
Kerollmops
0f8eb3b506
Improve the logs of the search with AI 2025-01-27 14:22:22 +01:00
Kerollmops
4a5923a55e
log the time arroy took to insert embeddings 2025-01-27 14:22:17 +01:00
Clément Renault
9b579069df
Comment the max grant of the bbqueue
Co-authored-by: Louis Dureuil <louis@meilisearch.com>
2025-01-24 12:18:32 +01:00
Louis Dureuil
f5a4a1c8b2
Give more RAM to bbqueue.
- bbqueue buffers used to have (5% * 2%) / num_threads
- they now have 5% / num_threads
2025-01-24 12:18:32 +01:00
Kerollmops
5ab4cdb1f3
Reduce the maximum grant possible we can store in the BBQueue 2025-01-24 12:18:32 +01:00
Tamo
7197ced673
fix the bad index version on opening 2025-01-23 16:51:24 +01:00
Tamo
787472453d
write the version of the index while upgrading it 2025-01-23 16:51:24 +01:00
Tamo
c27c923439
introduce a trait to upgrade the indexes 2025-01-23 16:51:23 +01:00
Tamo
41eeffd88d
fmt 2025-01-23 16:51:20 +01:00
Tamo
20ac59c946
fix the field distribution when upgrading from the v1_12 2025-01-23 16:51:19 +01:00
Tamo
cfc1e193b6
update the test with the stats 2025-01-23 16:51:19 +01:00
Tamo
0cc25c7e4c
add a large test importing a data.ms from the v1.12.0 2025-01-23 16:51:18 +01:00
Tamo
3ef7a478cd
move the version check to the task queue 2025-01-23 16:48:32 +01:00
Tamo
e70ac35e02
fix bugs after rebase 2025-01-23 16:48:32 +01:00
Tamo
d3654906bf
Add the new tasks with most of the job done 2025-01-23 16:48:32 +01:00
meili-bors[bot]
b15de68831
Merge
5257: Fix ollama r=Kerollmops a=dureuill

Fix oversight in ollama embedder 

WIP Integration tests are on branch `ollama-integration-test` and will be added as a future PR.

Co-authored-by: Louis Dureuil <louis@meilisearch.com>
2025-01-22 15:36:15 +00:00
Kerollmops
d142c5e432
Do not panic when the facet string is not found 2025-01-22 15:50:43 +01:00
Louis Dureuil
d6063079af
Unify facet strings by their normalized value 2025-01-22 15:50:42 +01:00
Louis Dureuil
a6470a0c37
Improve error log 2025-01-22 15:50:41 +01:00