[8.2] MOD-12212: Fix index load from RDB temporary memory overhead#7384
Merged
Itzikvaknin merged 3 commits into8.2from Nov 18, 2025
Merged
[8.2] MOD-12212: Fix index load from RDB temporary memory overhead#7384Itzikvaknin merged 3 commits into8.2from
Itzikvaknin merged 3 commits into8.2from
Conversation
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## 8.2 #7384 +/- ##
==========================================
- Coverage 89.45% 89.39% -0.06%
==========================================
Files 253 253
Lines 40800 40800
Branches 3725 3725
==========================================
- Hits 36496 36475 -21
- Misses 4255 4276 +21
Partials 49 49
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
Itzikvaknin
force-pushed
the
manual-backport-7336-to-8.2
branch
from
18b5237 to
c3e8592
Compare
…data from RDB. (#7336) * Condition IndexSpec_StartGC and Cursors_initSpec calls on sp->isDuplicate == false * change compatible_redis_version: "8.2" * Revert "change compatible_redis_version: "8.2"" This reverts commit 0d1e0bd. * Add testDuplicateIndexRdbLoad to test the duplicate spec case * Minor * Remove index spec validation since the env is not isolated * Improve testDuplicateIndexRdbLoad
Itzikvaknin
force-pushed
the
manual-backport-7336-to-8.2
branch
from
c3e8592 to
af42590
Compare
oshadmi
approved these changes
Nov 18, 2025
alonre24
added a commit
to alonre24/redis
that referenced
this pull request
Jan 26, 2026
**Bug Fixes:** * [redis#7219](RediSearch/RediSearch#7219) Fix a concurrency issue on Reducer in `FT.AGGREGATE` * [redis#7255](RediSearch/RediSearch#7255) Fix `BM25STD` underflow wraparound * [redis#7275](RediSearch/RediSearch#7275) Report used memory as `unsigned long long` to avoid underflows * [redis#7264](RediSearch/RediSearch#7264) Fix `totalDocsLen` updates * [redis#6995](RediSearch/RediSearch#6995) Do not fanout `FT.INFO` to replicas * [redis#7350](RediSearch/RediSearch#7350) Fix `FT.CREATE` failure with LeanVec parameters on non-Intel architectures * [redis#7384](RediSearch/RediSearch#7384) Fix index load from RDB temporary memory overhead * [redis#7459](RediSearch/RediSearch#7459) Fix Fork GC potential double-free on error path * [redis#7458](RediSearch/RediSearch#7458) Fix a GC performence regression * [redis#7470](RediSearch/RediSearch#7470) Avoid draining worker thread pool from FLUSH callback to avoid deadlocks * [redis#7554](RediSearch/RediSearch#7554) Handle the case where `SCORE` is sent alone without extra fields (coordinator) * [redis#7685](RediSearch/RediSearch#7685) Fix cursor logical leak * [redis#7794](RediSearch/RediSearch#7794) Fix `cmp_strings()` to correctly handle binary data with embedded NULLs in TOLIST reducer in FT.AGGREGATE * [redis#7873](RediSearch/RediSearch#7873) Handle warnings in empty `FT.AGGREGATE` replies (cluster) * [redis#7886](RediSearch/RediSearch#7886) Remove non-TEXT fields from the spec keys dictionary * [redis#7904](RediSearch/RediSearch#7904) Refactor keys dictionary handling * [redis#7901](RediSearch/RediSearch#7901) Support multiple warnings in reply * [redis#8083](RediSearch/RediSearch#8083) Fix incorrect FULLTEXT field metric counts * [redis#8153](RediSearch/RediSearch#8153) Fix configuration registration issues **Improvements:** * [redis#7154](RediSearch/RediSearch#7154) `FT.AGGREGATE` can return Background Indexing OOM warnings * [redis#7083](RediSearch/RediSearch#7083) Add the default text scorer as a configuration option * [redis#7341](RediSearch/RediSearch#7341) Rename `FT.PROFILE` counter fields * [redis#7436](RediSearch/RediSearch#7436) Enhance `FT.PROFILE` with vector search execution details * [redis#7435](RediSearch/RediSearch#7435) Ensure full `FT.PROFILE` output on timeout with RETURN policy * [redis#7534](RediSearch/RediSearch#7534) Reduce the number of worker threads asynchronously to avoid deadlocks during queries * [redis#7614](RediSearch/RediSearch#7614) Track timeout warnings and errors in INFO * [redis#7646](RediSearch/RediSearch#7646) Track `maxprefixexpansions` warnings and errors in INFO * [redis#7577](RediSearch/RediSearch#7577) Track query syntax/argument errors (basis for query error tracking) * [redis#7737](RediSearch/RediSearch#7737) Add `Internal cursor reads` metric to cluster `FT.PROFILE` output * [redis#7759](RediSearch/RediSearch#7759) Extend indexing metrics * [redis#7710](RediSearch/RediSearch#7710) Support `WITHCOUNT` keyword in `FT.AGGREGATE` * [redis#7957](RediSearch/RediSearch#7957) Persist query warnings across cursor reads * [redis#8054](RediSearch/RediSearch#8054) Add logging for index-related commands * [redis#8151](RediSearch/RediSearch#8151) Fix shard total profile time reporting in `FT.PROFILE` * [redis#8103](RediSearch/RediSearch#8103) Output current thread IndexSpec information on crash
YaacovHazan
pushed a commit
to redis/redis
that referenced
this pull request
Jan 26, 2026
**Bug Fixes:** * [#7219](RediSearch/RediSearch#7219) Fix a concurrency issue on Reducer in `FT.AGGREGATE` * [#7255](RediSearch/RediSearch#7255) Fix `BM25STD` underflow wraparound * [#7275](RediSearch/RediSearch#7275) Report used memory as `unsigned long long` to avoid underflows * [#7264](RediSearch/RediSearch#7264) Fix `totalDocsLen` updates * [#6995](RediSearch/RediSearch#6995) Do not fanout `FT.INFO` to replicas * [#7350](RediSearch/RediSearch#7350) Fix `FT.CREATE` failure with LeanVec parameters on non-Intel architectures * [#7694](RediSearch/RediSearch#7694) Use asynchronous jobs for GC in SVS to accelerate execution * [#7384](RediSearch/RediSearch#7384) Fix index load from RDB temporary memory overhead * [#7459](RediSearch/RediSearch#7459) Fix Fork GC potential double-free on error path * [#7458](RediSearch/RediSearch#7458) Fix a GC performence regression * [#7470](RediSearch/RediSearch#7470) Avoid draining worker thread pool from FLUSH callback to avoid deadlocks * [#7554](RediSearch/RediSearch#7554) Handle the case where `SCORE` is sent alone without extra fields (coordinator) * [#7685](RediSearch/RediSearch#7685) Fix cursor logical leak * [#7794](RediSearch/RediSearch#7794) Fix `cmp_strings()` to correctly handle binary data with embedded NULLs in TOLIST reducer in FT.AGGREGATE * [#7873](RediSearch/RediSearch#7873) Handle warnings in empty `FT.AGGREGATE` replies (cluster) * [#7886](RediSearch/RediSearch#7886) Remove non-TEXT fields from the spec keys dictionary * [#7904](RediSearch/RediSearch#7904) Refactor keys dictionary handling * [#7901](RediSearch/RediSearch#7901) Support multiple warnings in reply * [#8083](RediSearch/RediSearch#8083) Fix incorrect FULLTEXT field metric counts * [#8153](RediSearch/RediSearch#8153) Fix configuration registration issues **Improvements:** * [#7154](RediSearch/RediSearch#7154) `FT.AGGREGATE` can return Background Indexing OOM warnings * [#7083](RediSearch/RediSearch#7083) Add the default text scorer as a configuration option * [#7341](RediSearch/RediSearch#7341) Rename `FT.PROFILE` counter fields * [#7436](RediSearch/RediSearch#7436) Enhance `FT.PROFILE` with vector search execution details * [#7435](RediSearch/RediSearch#7435) Ensure full `FT.PROFILE` output on timeout with RETURN policy * [#7534](RediSearch/RediSearch#7534) Reduce the number of worker threads asynchronously to avoid deadlocks during queries * [#7614](RediSearch/RediSearch#7614) Track timeout warnings and errors in INFO * [#7646](RediSearch/RediSearch#7646) Track `maxprefixexpansions` warnings and errors in INFO * [#7577](RediSearch/RediSearch#7577) Track query syntax/argument errors (basis for query error tracking) * [#7737](RediSearch/RediSearch#7737) Add `Internal cursor reads` metric to cluster `FT.PROFILE` output * [#7759](RediSearch/RediSearch#7759) Extend indexing metrics * [#7710](RediSearch/RediSearch#7710) Support `WITHCOUNT` keyword in `FT.AGGREGATE` * [#7957](RediSearch/RediSearch#7957) Persist query warnings across cursor reads * [#8054](RediSearch/RediSearch#8054) Add logging for index-related commands * [#8151](RediSearch/RediSearch#8151) Fix shard total profile time reporting in `FT.PROFILE` * [#8103](RediSearch/RediSearch#8103) Output current thread IndexSpec information on crash
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Manual backport of 7336
Note
Avoid starting GC for duplicate indexes during RDB load; add test ensuring multiple duplicate entries result in a single stored index.
IndexSpec_StartGC(ctx, spec_ref, sp)to only run after confirming not a duplicate inIndexSpec_CreateFromRdb, preventing GC init for duplicates.testDuplicateIndexRdbLoadintests/cpptests/test_cpp_rdb.cppto save the same index 30× and verify only one instance loads and is stored.FieldSpec_RdbLoadcall alignment).Written by Cursor Bugbot for commit 8d3086c. This will update automatically on new commits. Configure here.