v2.10.31

This is a maintenance release for Redis Search 2.10.
Update urgency: LOW: No need to upgrade unless there are new features you want to use.

Bug Fixes

• #9938 Server can crash in fork GC when the last document with an empty TAG value is deleted from an INDEXEMPTY WITHSUFFIXTRIE field. (MOD-15996)
• #10104 FT.SEARCH and FT.AGGREGATE return Unknown argument errors on 2.10 shards during rolling upgrades when a newer coordinator injects internal query arguments. (MOD-16047)
• #9578 Server can crash when FT.CREATE is called with an extremely large number of arguments or fields. (MOD-6411)
• #9810 FT.SYNUPDATE leaves the synonym map partially mutated when an update batch exceeds synonym or group-ID limits. (MOD-15402)
• #9533 FT.INFO num_records grows without bound for indexes with vector fields because vector records were counted on insert but never decremented. (MOD-15487)

v2.8.38

This is a maintenance release for Redis Search 2.8.
Update urgency: LOW: No need to upgrade unless there are new features you want to use.

Bug Fixes

• #10105 FT.SEARCH and FT.AGGREGATE return Unknown argument errors on 2.8 shards during rolling upgrades when a newer coordinator injects internal query arguments. (MOD-16047)
• #9579 Server becomes unresponsive when FT.CREATE is called with an extremely large number of fields. (MOD-6411)
• #9825 FT.SYNUPDATE leaves the synonym map partially mutated when an update batch exceeds synonym or group-ID limits. (MOD-15402)
• #9534 FT.INFO num_records grows without bound for indexes with vector fields due to missing decrement on deletion. (MOD-15487)

v2.6.37

This is a maintenance release for Redis Search 2.6.
Update urgency: LOW: No need to upgrade unless there are new features you want to use.

Bug Fixes
• #10106 FT.SEARCH and FT.AGGREGATE return Unknown argument errors on 2.6 shards during rolling upgrades when a newer coordinator injects internal query arguments. (MOD-16047)
• #9826 FT.SYNUPDATE leaves the synonym map partially mutated when an update batch exceeds synonym or group-ID limits. (MOD-15402)
• #9535 FT.INFO num_records is skewed for indexes with vector fields because vector entries are never decremented on deletion. (MOD-15487)

v2.10.30

This is a maintenance release for Redis Search 2.10.

Update urgency: HIGH: There is a critical bug that may affect a subset of users. Upgrade

Bug Fixes:

• #8948 FT.CURSOR READ enters an infinite loop when the caller lacks the required ACL permissions. (MOD-14479)
• #8794 FT.EXPLAIN crashes or produces corrupt output when a concurrent schema change occurs. (MOD-14461)
• #9176, #9314 Coordinator deadlock under mixed FT.SEARCH and FT.AGGREGATE load. (MOD-14268)
• #9273, #9302 Async shard connect hangs indefinitely when a peer endpoint is unreachable; the connect is now bounded by search-connect-timeout. (MOD-12739)
• #9167 Crash on FT.SEARCH when topology validation fails (for example, some nodes unreachable). (MOD-14475)
• #9082 FT.SPELLCHECK ignores $param placeholders even when PARAMS is supplied. (MOD-10596)
• #9050 FT.PROFILE produces malformed output for documents with missing field values. (MOD-10560)
• #9075 Invalid schema option combinations are silently accepted at FT.CREATE time. (MOD-14655)
• #9436 Malformed index definitions in an RDB file crash the shard or load an invalid index. (MOD-13118)
• #7686 Cursors are not released on certain error paths, leaking cursor slots over time. (MOD-12807)
• #8465 Garbage collector mishandles out-of-memory responses on replicas, leaving the index inconsistent. (MOD-14066)
• #7771 String comparisons truncate at embedded NULL bytes, returning wrong results for TAG and binary fields. (MOD-13010)
• #7851 FT.SUG* commands are not registered on Redis Enterprise builds. (MOD-13152)
• #8154 Per-shard total time in cluster FT.PROFILE output is reported incorrectly. (MOD-13735, MOD-13181)
• #8084 FULLTEXT field count in FT.INFO statistics is inaccurate. (MOD-13432)

Metrics:

• #7558, #7688, #7723, #7791, #7749, #7738, #7744 New FT.INFO metrics expose live thread-pool utilization and pending-job depth for worker, coordinator, IO, and topology-update pools (search_active_io_threads, search_active_worker_threads, search_active_coord_threads, search_active_topology_update_threads, search_workers_low_priority_pending_jobs, search_workers_high_priority_pending_jobs, search_workers_admin_priority_pending_jobs, search_coord_high_priority_pending_jobs). (MOD-12069, MOD-12695, MOD-12694, MOD-12790, MOD-12791)
• #7793, #7816 New FT.INFO counters track per-shard and coordinator query timeouts and MAXPREFIXEXPANSIONS truncations (search_shard_total_query_errors_timeout, search_shard_total_query_warnings_timeout, search_shard_total_query_warnings_max_prefix_expansions, and equivalent search_coord_total_* counters). (MOD-12419, MOD-12417)
• #7766 New FT.INFO per-field-type indexing-operation counters added for TEXT, TAG, NUMERIC, GEO, GEOSHAPE, and VECTOR fields. (MOD-12070)
• #7746 New Internal cursor reads counter in cluster FT.PROFILE output surfaces internal cursor-hop round-trips per query. (MOD-12414)
• #7593 FT.PROFILE now reports GIL time, giving an accurate breakdown of where query time is spent. (MOD-11987, MOD-12816)

v2.8.37

This is a maintenance release for Redis Search 2.8.

Update urgency: HIGH : There is a critical bug that may affect a subset of users. Upgrade

Bug Fixes:

• #7860: FT.SUGADD and FT.SUGDELreturn inconsistent results in sharded pre-Redis-8 deployments due to random shard routing (MOD-13152)
• #8949: FT.CURSOR enters an infinite loop when the caller's ACL restricts key access (MOD-14479)
• #8795: FT.EXPLAIN crashes or reads stale state under concurrent schema changes (MOD-14461)
• #7697: Stack-use-after-scope in _FT.CURSOR PROFILE produces undefined behavior (MOD-12955)
• #7687: Cursor resources leak when the associated index is dropped while the cursor is open (MOD-12807)
• #8466: Garbage collector mishandles Out-of-Memory conditions on replicas (MOD-14066)
• #8163: FT.PROFILE reports incorrect shard total profile time in cluster mode (MOD-13735, MOD-13181)
• #8085: FT.PROFILE reports incorrect FULLTEXT field metric count (MOD-13432)
• #7721: FT.PROFILE always reports GILTime 0 due to missing initialization and accumulation (MOD-11987, MOD-12816)
• #9074: FT.CREATE accepts invalid schema option combinations that fail silently at query time (MOD-14655)
• #9083: FT.SPELLCHECK ignores $param placeholders from the PARAMS clause (MOD-10596)
• #9435: RDB loader accepts out-of-range or malformed values, causing instability on restore (MOD-13118)
• #7703, #7727: Pending coordinator and worker jobs are not correctly tracked in the queue
• #7629, #7653, #7662: FT.PROFILE gains a debug mode to simulate timeouts, crashes, and mid-execution pauses (MOD-8127, MOD-11155, MOD-12627)
• #7966: Module load no longer fails when find_module is undefined (MOD-13330)

Metrics:

• #7597, #7645, #7665: New active_io_threads, active_worker_threads, and active_coord_threads metrics expose live thread activity (MOD-12069, MOD-12694, MOD-12695)
• #7778: New counters track total documents indexed and fields indexed per field type across all indexes (MOD-12070)
• #7788, #7819, #7820: New FT.INFO counters track query syntax errors, timeout errors/warnings, and MAXPREFIXEXPANSIONS warnings (MOD-12416, MOD-12417, MOD-12419)
• #7751: Cluster-mode FT.PROFILE now reports Internal cursor reads per shard (MOD-12414)
• #8066: Index commands now emit informational log entries for operational visibility (MOD-13431)

v2.6.36

This is a maintenance release for Redis Search 2.6.

Update urgency: HIGH : There is a critical bug that may affect a subset of users. Upgrade

Bug Fixes:

• #8950 FT.CURSOR enters an infinite loop when the invoking user's ACL restricts access to the cursor's index, causing CPU exhaustion and shard restarts. (MOD-14479)
• #8468 Garbage collector mishandles out-of-memory conditions on replicas, causing instability under memory pressure during indexing. (MOD-14066)
• #6792 Counter result-processor cleanup uses an incorrect type cast, risking memory corruption when freeing query pipeline resources. (MOD-11216)
• #6998 FT.INFO is fanned out to replicas in clustered deployments, producing inconsistent responses and unnecessary inter-node traffic.
• #6871 Coupled grow/shrink logic in Flat and HNSW vector index containers causes excess memory churn for large or fluctuating vector datasets. (MOD-10559)

v2.10.25

This is a maintenance release for RediSearch 2.10

Update urgency: LOW: No need to upgrade unless there are new features you want to use.

Bug Fixes:

• #7216 Fix a concurrency issue on Reducer in FT.AGGREGATE
• #7259 Fix underflow in BM25STD
• #7278 Report used memory as unsigned long long to avoid underflows
• #7340 Fix a rare leak in GC
• #7462 Fix Fork GC potential double-free on error path
• #7525 Avoid draining workers thread pool from FLUSHDB callback to avoid potential deadlocks

Improvements:

• #7157 Show background Indexing OOM warning in FT.AGGREGATE reply in RESP3
• #7086 Add the default text scorer as a new configuration
• #7380 Rename FT.PROFILE counter fields
• #7366 Reduce temporary memory overhead upon index load from RDB
• #7393 Improve timeout check for FT.AGGREGATE in cluster mode
• #7451 Enhance FT.PROFILE with vector search execution details
• #7480 Ensure full profile output on timeout with RETURN policy

Full Changelog: v2.10.24...v2.10.25

v2.8.32

This is a maintenance release for RediSearch 2.8

Update urgency: LOW: No need to upgrade unless there are new features you want to use.

Bug Fixes:

• #7216 Fix a concurrency issue on Reducer in FT.AGGREGATE
• #7259 Fix underflow in BM25STD
• #7278 Report used memory as unsigned long long to avoid underflows
• #7340 Fix a rare leak in GC
• #7462 Fix Fork GC potential double-free on error path
• #7525 Avoid draining workers thread pool from FLUSHDB callback to avoid potential deadlocks

Improvements:

• #7157 Show background Indexing OOM warning in FT.AGGREGATE reply in RESP3
• #7086 Add the default text scorer as a new configuration
• #7380 Rename FT.PROFILE counter fields
• #7366 Reduce temporary memory overhead upon index load from RDB
• #7393 Improve timeout check for FT.AGGREGATE in cluster mode
• #7451 Enhance FT.PROFILE with vector search execution details
• #7480 Ensure full profile output on timeout with RETURN policy

Full Changelog: v2.8.31...v2.8.32

v2.10.24

This is a maintenance release for RediSearch 2.10

Update urgency: LOW: No need to upgrade unless there are new features you want to use.

Bug Fixes

• #6960 FT.INFO returns the wrong number of documents in OSS Cluster with replicas
• #6938 Fix in the HIGHLIGHT feature, where if some fields have empty strings, wrong tokens might be highlighted
• #7047 Avoid crashing in FT.AGGREGATE command in clusters where different shards have different ON_TIMEOUT policy configured (fail vs return)

Full Changelog: v2.10.23...v2.10.24

v2.8.31

This is a maintenance release for RediSearch 2.8

Update urgency: LOW: No need to upgrade unless there are new features you want to use.

Bug Fixes:

• #6960 FT.INFO returns the wrong number of documents in OSS Cluster with replicas
• #6938 Fix in the HIGHLIGHT feature, where if some fields have empty strings, wrong tokens might be highlighted
• #7049 Avoid crashing in FT.AGGREGATE command in clusters where different shards have different ON_TIMEOUT policy configured (fail vs return)