milvus-2.6.12

v2.6.12

Release date: March 17, 2026

Milvus Version	Python SDK Version	Node.js SDK Version	Java SDK Version	Go SDK Version
2.6.12	2.6.10	2.6.11	2.6.15	2.6.1

We are pleased to announce the release of Milvus v2.6.12! This release introduces replication topology inspection and configurable TLS minimum version for object storage. It also delivers significant memory optimizations in segment loading and compaction, along with numerous bug fixes addressing memory leaks, RBAC alias resolution, collection-level rate limiting, and streaming node stability.

Features

• Added GetReplicateConfiguration API for viewing replication topology with redacted tokens (#47543)
• Added configurable TLS minimum version (minio.ssl.tlsMinVersion) for object storage connections across all supported backends (#48000, #48030)
• Supported configuring different replica numbers on secondary CDC cluster independently from the primary (#47914)
• Allowed pchannel count increase in CDC ReplicateConfiguration to support heterogeneous cluster topologies (#47917)
• Added automatic warmup for large tenant collections to reduce cold-start query latency (#47631)
• Added user-specified warmup support for RESTful API (#47825)

Improvements

• Added caching layer resource management for streaming node with automatic memory/disk accounting during segment operations (#47165)
• Optimized query node segment load speed (#47424)
• Optimized MixCoord CPU usage to reduce coordinator overhead (#47619)
• Added phase-level timing logs and metrics for sort compaction (#47674)
• Refactored cluster-level broadcast mechanism to decouple message package from channel registration lifecycle (#47807)
• Added configurable skip list for replicate message types in CDC (#47910)
• Included text index memory cost in segment loading memory estimation (#47963)
• Added per-cluster TLS configuration support for CDC outbound mTLS connections (#48023)
• Bypassed broadcaster for resource group DDL operations on non-primary clusters in streaming replication (#48034)
• Bumped OpenTelemetry to v1.40.0 to address CWE-426 untrusted search path vulnerability (#48059)

Bug fixes

• Fixed an issue where collection-level rate limits were not delivered to proxies when set via collection properties (#46714, #48018)
• Fixed memory accumulation during segment loading by switching to streaming cell-based loading (#47859)
• Fixed streaming node crash-loop during WAL replay for dropped collections (#47887)
• Fixed memory accumulation in compaction by streaming row groups one at a time (#47904)
• Fixed memory leak in broadcaster caused by uncanceled context (#47912)
• Fixed an issue where CreateCollection could cause schema loss on crash (#47972)
• Fixed crash when inserting non-nullable fields with default values (#48016)
• Fixed deadlock when loading V1 segments with more than 16 system field binlog files (#48037)
• Fixed partial update failure when dynamic field is enabled but upsert data contains only static fields (#48044)
• Fixed RBAC permission checks to resolve collection aliases and added automatic grant cleanup on collection drop/rename (#48052, #48151)
• Fixed BulkInsert failure by removing interactive coordination between import and L0 compaction (#48114)
• Fixed Azure buffered output stream issue (#48041)

milvus-2.6.11

v2.6.11

Release date: February 12, 2026

Milvus Version	Python SDK Version	Node.js SDK Version	Java SDK Version	Go SDK Version
2.6.11	2.6.9	2.6.9	2.6.13	2.6.1

We are pleased to announce the release of Milvus 2.6.11! This update continues to enhance query performance and system stability with improvements to filtering execution, segment loading, and Storage V2 I/O pipelining. It also refines geo indexing, reduces memory usage in default-value chunks, and improves developer and build tooling through dependency and test-suite cleanups. This release further fixes several correctness issues across control-channel handling, index building, nullable-expression semantics, and WAL recovery workflows. We recommend all users on the 2.6 branch upgrade to this version for improved reliability and performance.

Features

• Added a truncate API to remove collection data more efficiently (#47308)

Improvements

• Used PreparedGeometry to improve geo index refinement performance (#47389)
• Switched the OpenSSL dependency to shared linking (#47664)
• Differentiated load priorities by scenario to improve scheduling behavior (#47594)
• Upgraded Go to 1.24.12 and updated gpgv to address CVEs (#47562)
• Reduced memory usage by enabling multi-cell DefaultValueChunk layout (#47166)
• load-diff based segment loading patches to improve load efficiency (#47545)
• Removed redundant bitset count operations during filter execution to reduce CPU overhead (#47546)
• Added semantic highlighting support for dynamic fields (#47464)
• Reduced unnecessary PinWrapper copies in searchPksWith to improve query performance (#47531)
• Normalized constant-folded boolean expressions to AlwaysTrueExpr/AlwaysFalseExpr during rewriting for simpler plans (#47493)
• Added RESTful search_by_pk support (#47318)
• Optimized “latest delete snapshot” handling to reduce overhead (#47409)
• Added support for user-specified warmup settings (#47343)
• Added LoadWithStrategyAsync to enable true I/O pipelining in Storage V2 (#47427)
• Optimized MixCoord's CPU and memory usage by avoiding redundant calculations in the balance checker (#47190)
• Added sparse filtering support in search (#47447)
• Reduced memory allocations and copies during data loading (#47088)

Bug fixes

• Fixed an issue where collection metadata could contain an invalid database name (#47721)
• Ensured exclusive control-channel messages acquire a global lock in the lock interceptor (#47678)
• Fixed channel exclusive mode state loss and vchannel list handling issues (#47702)
• Fixed index building to use the correct global offset for null_offset_ in BuildIndexFromFieldData (#47708)
• Improved v2.5/v2.6 compatibility handling in SyncTargetVersion (QueryNode) (#47693)
• Handled broadcastToAll messages on the control channel in recovery storage (#47640)
• Added warmupKey to the CheckParams filter to make CreateIndex idempotent (#47607)
• Corrected the default mmap value in code (#47490)
• Populated LevelZeroSegmentIDs in GetDataVChanPositions (#47597)
• Corrected null handling on NullExpr, ExistsExpr, and logical operators (#47519)
• Removed segment_loader pre-reserve logic for warmup fields/indexes to avoid incorrect reservations (#47463)
• Updated log_* macros to use {} placeholders to avoid treating error messages as format strings (#47485)
• Fixed bloom filter memory leak when a worker node crashes (#47451)
• Used actual data timestamps for imported segment positions (#47370)
• Rebuilt WAL messages on each append retry to avoid panics (#47480)
• Filled in the log and memory size fields in TextIndexStats metadata (#47476)
• Reduced the empty timetick filtering interval to improve timetick handling (#47471)

milvus-2.5.27

v2.5.27

Release date: February 27, 2026

Milvus Version	Python SDK Version	Node.js SDK Version	Java SDK Version	Go SDK Version
2.5.27	2.5.18	2.5.13	2.5.14	2.5.14

Milvus 2.5.27 is a critical security release that fixes CVE-2026-26190 (CVSS 9.8), an authentication bypass vulnerability on the metrics port (9091) that could allow unauthenticated access to the REST API and sensitive system operations. This release also includes important bug fixes and dependency updates. We strongly recommend all 2.5.x users upgrade immediately.

Improvements

• Upgraded Go to 1.24.12 and updated dependencies to fix CVEs (#47561)

Bug fixes

• Fixed imported segment positions to use actual data timestamps instead of channel checkpoints (#47372)
• Added authentication to metrics endpoint when authorization is enabled (#47405)

milvus-2.6.10

v2.6.10

Release date: February 5, 2026

Milvus Version	Python SDK Version	Node.js SDK Version	Java SDK Version	Go SDK Version
2.6.10	2.6.8	2.6.9	2.6.13	2.6.1

We are pleased to announce the release of Milvus 2.6.10! This update addresses a critical security vulnerability CVE-2026-26190, strengthens security controls around KMS key revocation and improves search and storage performance through automatic FP32-to-FP16/BF16 conversion, optimized segment loading, and updated auto-index configurations. This release also fixes a number of stability issues across compaction, query pagination, and recovery workflows. We recommend all users on the 2.6 branch upgrade to this version for improved reliability and performance.

Improvements

• Added support to stop WAL consumption when a KMS key is revoked (#47018)
• Updated the default auto-index configuration for vector fields (#47388)
• Disabled storage-version upgrade compaction by default (#47383)
• Added automatic FP32-to-FP16/BF16 conversion in search (#47241)
• Limited segment load concurrency by submitting loads to the load pool (#47335)
• Added the map_populate flag for mmap to reduce page faults during access (#47317)
• Persisted BM25 stats to disk during segment loading to reduce recomputation (#47232)
• Added loading timeout and cancellation support for better control of long-running loads (#47223)
• Allowed alter_collection_field() to update the field description (#47058)
• Added a target manager to ReplicaObserver initialization (#47093)
• Updated the Knowhere version for vector search improvements (#47109)
• Added BM25 search_by_pk support (#47012)
• Extracted assign policy from the balancer and added StoppingBalancer (#47138)
• Prevented import jobs/tasks from rolling back state unexpectedly (#47102)
• Improved slow logs by recording average cost per NQ (#47086)

Bug fixes

• Fixed incorrect group results during pagination of grouped queries (#47248)
• Added boundary validation for threadpool resize operations (#47367)
• Improved error message handling when the error type is missing (#47369)
• Prevented coredumps and improved diagnostics for PhyReScoresNode (#47341)
• Reverted a compaction change related to “fast finish” when L0 compaction hits zero (L1/L2) (#47336)
• Prevented server crashes on division/modulo by zero in filter expressions (#47306)
• [Go SDK] Aligned timestamptz field type and data format with the server (#47328)
• Added authentication to the metrics endpoint when authorization is enabled (#47278)
• Updated milvus_proxy_req_count metrics for RESTful APIs (#47239)
• Submitted TriggerTypeStorageVersionUpgrade compaction tasks correctly (#47234)
• Allowed empty compaction results (#47153)
• Fixed Azure precheck to use a fixed bucket not owned by Milvus (#47168)
• Ignored L0 compaction during PreallocSegmentIDs checks (#47189)
• Fixed compaction fast-finish behavior when L0 compaction hits zero (L1/L2) (#47187)
• Removed unnecessary batching to reduce OOM risk (#47175)
• Fixed deserialization handling for empty vector arrays (#47127)
• Fixed runtime config updates not triggering watchers (#47161)
• Unified primary-key handling logic between deletePreExecute and packDeleteMessage (#47147)
• Used user-provided target size in compaction-related logic (#47115)

milvus-2.5.26

v2.5.26

Release date: January 27, 2026

Milvus Version	Python SDK Version	Node.js SDK Version	Java SDK Version	Go SDK Version
2.5.26	2.5.18	2.5.13	2.5.14	2.5.14

We are pleased to announce the release of Milvus 2.5.26! This release includes critical security enhancements and important stability fixes. A key highlight is the addition of security controls for the internal /expr endpoint to prevent potential remote expression execution vulnerabilities. We have also upgraded the Pulsar client library to fix race conditions, resolved L0 segment data forwarding issues, and fixed Azure storage precheck configuration problems. We strongly recommend all users on the 2.5 branch upgrade to this version.

Security

• Added security controls for the /expr endpoint to prevent remote expression execution vulnerabilities (#47130)

Improvements

• Upgraded pulsar-client-go to v0.17.0 to fix send buffer race conditions (#47184)
• Added gRPC metadata header for client request time tracking (#46946)

Bug fixes

• Fixed L0 growing segment data forwarding to use the correct bulk delta-load API (#47231)
• Fixed Azure blob storage precheck using incorrect bucket configuration (#47204)
• Fixed geometry data conversion inconsistencies in WKT/WKB options (#46872)

milvus-2.6.9

v2.6.9

Release date: January 16, 2026

Milvus Version	Python SDK Version	Node.js SDK Version	Java SDK Version	Go SDK Version
2.6.9	2.6.6	2.6.9	2.6.12	2.6.1

We are pleased to announce the release of Milvus 2.6.9! This update introduces highlight scores for search results, enhances segment management with support for reopening segments when data or schema changes occur, and improves storage version handling. Key improvements include better logging performance, enhanced security controls for expression endpoints, and optimizations for text analyzers and index building. This release also resolves critical issues including memory estimation accuracy, geometry data conversions, and various stability fixes. We recommend all users on the 2.6 branch upgrade to this version for improved system reliability and performance.

Features

• Supported searching by primary keys (#46528)

Improvements

• Added a storage version label metric for better observability (#47014)
• QueryCoord now supports segment reopen when manifest path changes (#46921)
• Added support for reopening segments when data or schema changes occur (#46412)
• Improved slow log performance and efficiency (#47086)
• Added storage version upgrade compaction policy to facilitate version migrations (#47011)
• Eliminated extra memory copy operations for C++ logging to improve performance (#46992)
• Added security controls for the /expr endpoint to prevent unauthorized access (#46978)
• Streaming service now remains enabled until the required streaming node count is reached (#46982)
• Removed redundant etcd put operations when updating segment information (#46794)
• Improved row count validation and reduced misleading warning logs for sort compaction (#46824)
• Cleaned up and organized build index log messages (#46769)
• Limited the number of concurrent vector index builds per worker to prevent resource exhaustion (#46877)
• Optimized jieba and lindera analyzer cloning operations for better performance (#46757)
• Added glog sink to transfer CGO logs into zap logger for unified logging (#46741)
• Enforced storage V2 format usage and deprecated V1 writes (#46889)
• Implemented batch processing for ngram operations to improve efficiency (#46703)
• Added automatic retry mechanism for binlog write operations to improve reliability (#46854)
• Filtered empty timetick messages from the consuming side to reduce unnecessary processing (#46730)
• Improved search by primary key with duplicate checking and automatic anns_field inference (#46745)
• Added dimension parameter support for siliconflow and cohere embedding providers (#47081)

Bug fixes

• Fixed double counting of index memory in segment loading estimation (#47046)
• Fixed compilation issues on macOS 14 (#47048)
• Used revision as streaming service discovery global version for better consistency (#47023)
• Ensured all futures complete on exception to prevent use-after-free crashes (#46960)
• Fixed shard interceptor incorrectly skipping FlushAllMsg operations (#47004)
• Added valid range validation for collection TTL to prevent invalid configurations (#47010)
• Fixed GetCredentialInfo not caching RPC responses (#46945)
• Fixed issue where AlterFunction could not be invoked when multiple functions become invalid (#46986)
• Fixed inverted index null offset file not being compacted (#46950)
• Fixed crash when using is_null_expr on indexed JSON fields (#46894)
• Added check for allow_insert_auto_id flag in RESTful v2 insert API (#46931)
• Added field existence check in column groups before reading from loon manifest (#46924)
• Fixed bug where the highlight parameter was not working correctly (#46876)
• Quota center now ignores delegator when it is in recovering state (#46858)
• Aligned WKT/WKB conversion options to ensure consistent behavior across operations (#46874)
• Fixed voyageai model int8 bug (#46821)
• Fixed missing handling of FlushAllMsg in recovery storage operations (#46803)
• Fixed missing shardclientmgr field in querytask to prevent panic (#46838)
• Used leaderid for leaderaction stale check in scheduler to improve accuracy (#46788)
• Restored tenant/namespace support for Pulsar that was lost in 2.6 (#46759)
• Added load config watcher to prevent load config modifications from being lost (#46786)
• Fixed function edit interface bug (#46782)
• Added collection TTL property validation to prevent compaction from getting stuck (#46736)

milvus-2.5.25

v2.5.25

Release date: January 12, 2026

Milvus Version	Python SDK Version	Node.js SDK Version	Java SDK Version	Go SDK Version
2.5.25	2.5.18	2.5.13	2.5.14	2.5.14

We are pleased to announce the release of Milvus 2.5.25! This update focuses on enhancing the reliability of storage operations and optimizing background task scheduling. Key improvements include a more robust retry mechanism for object storage under rate-limiting conditions and more accurate resource estimation for indexing tasks. This release also resolves several critical issues, including ETCD RPC limits during collection drops and consistency in geometry data conversions. We recommend all users on the 2.5 branch upgrade to this version for improved system stability.

Improvements

• Prevented message loss in Pulsar consumers after client library upgrades (#46591)
• Made metadata batch processing configurable to prevent transaction errors (#46514)
• Improved reliability of object storage operations under high load with automatic retry on rate limit errors (#46463)
• Added configuration option for variable-length field size estimation (#46301)
• Improved slot allocation accuracy for indexing tasks (#46260)

Bug fixes

• Fixed geometry data conversion inconsistencies (#46872)
• Fixed inverted index compaction issue that could cause segment load failures (#46868)
• Fixed text log loading failure (#46704)
• Fixed ETCD error when dropping large collections (#46580)
• Fixed resource leak issue with proxy clients (#46491)

milvus-2.6.8

v2.6.8

Release date: January 4, 2026

Milvus Version	Python SDK Version	Node.js SDK Version	Java SDK Version	Go SDK Version
2.6.8	2.6.6	2.6.9	2.6.11	2.6.1

We are excited to announce the release of Milvus 2.6.8! This version introduces search result highlighting, significantly enhancing the retrieval experience. Under the hood, we have optimized query processing, resource scheduling, and caching mechanisms to deliver superior performance and stability. Additionally, this release addresses critical bugs related to data security, storage handling, and concurrency. We highly recommend all users upgrade to this version for a more efficient and reliable production environment.

Features

• Supported search with highlighter. For details, refer to Text Highlighter. (#46052)

Improvements

• Moved query optimization logic to the Proxy to improve performance (#46549)
• Optimized LIKE operator performance using STL sort (#46535)
• Enabled concurrent execution of text index tasks for multiple fields (#46306)
• Supported pausing GC at the collection level (#46201)
• Implemented a penalty policy for QueryNodes to handle resource exhaustion (#46086)
• Optimized data caching by mapping multiple row groups to a single cache cell (#46542)
• Reduced CPU usage in QuotaCenter (#46615)
• Improved TIMESTAMPTZ data comparison performance (#46655)
• Supported nullable dynamic fields with an empty JSON object as the default value (#46445)
• Prevented unnecessary segment sealing when only altering collection properties (#46489)
• Supported DML and DQL forwarding in Proxy for RESTful v2 (#46021, #46037)
• Added retry mechanism for object storage reads on rate limit errors (#46464)
• Enhanced logging for Proxy and RootCoord meta tables (#46701)
• Added validation for embedding models and schema field types (#46422)
• Introduced a tolerance duration to delay collection drop operations (#46252)
• Improved index task scheduling by estimating slots based on field size and type (#46276, #45851)
• Added fallback mechanism for write paths when accessing object storage without condition write support (#46022)
• Optimized IDF oracle synchronization logic (#46079)
• Changed RootCoord default port to a non-ephemeral port (#46268)
• Added metrics to monitor Jemalloc cached memory (#45973)
• Improved disk quota metric accuracy when cluster quota changes (#46304)
• Improved trace observability for scalar expressions (#45823)
• Rejected duplicate primary keys in upsert batch requests (#46035)

Bug fixes

• Fixed RBAC ETCD prefix matching to prevent potential data leakage (#46708)
• Fixed incorrect root path handling for local storage mode (#46693)
• Fixed handling of mixed int64/float types in JSON fields (#46682)
• Fixed text log loading failures during cluster upgrade (#46698)
• Prevented deletion of other fields during raw data cleanup (#46689)
• Fixed failure when using highlighting with multiple analyzers (#46664)
• Ensured logs are flushed when the OS exits (#46609)
• Fixed ETCD RPC size limit exceeded error when dropping collections (#46645)
• Fixed replication lag issues when the server is idle (#46612)
• Fixed validation for invalid TIMESTAMPTZ default values (#46556)
• Fixed restoration of compaction tasks to ensure proper cleanup (#46578)
• Unified read-only node handling to avoid stuck balance channel tasks (#46513)
• Prevented field data drops for multi-field column groups (#46425)
• Removed stale proxy clients when re-watching ETCD (#46490)
• Fixed chunk iterator merge order (#46462)
• Prevented creation of Kafka consumer groups by disabling auto-commit (#46509)
• Prohibited hot-reloading of tiered storage parameters (#46438)
• Enabled search iterator for binary vectors (#46334)
• Fixed race condition in storage initialization (#46338)
• Fixed highlight queries not working for non-BM25 searches (#46295)
• Fixed stack overflow during JSON garbage collection (#46318)
• Ensured retries when writing binlogs (#46310)
• Fixed index usage check for JSON fields (#46281)
• Prevented target update blocking when replicas lack nodes during scaling (#46291)
• Restricted char_group tokenizer to support only one-byte delimiters (#46196)
• Skipped JSON path index usage if the query path includes numbers (#46247)
• Fixed path concatenation errors in MinIO when root path is "." (#46221)
• Fixed false-positive health checks by correcting replicate lag metric calculation (#46122)
• Fixed RESTful v2 parsing and schema defaults with TIMESTAMPTZ (#46239)
• Fixed panic when searching empty results with output geometry fields (#46231)
• Added field data alignment validation to prevent panics during partial updates (#46180)
• Fixed database loss issue in RESTful v2 (#46172)
• Fixed incorrect context usage in gRPC client sessions (#46184)
• Fixed incorrect authorization forwarding in RESTful v2 during upgrades (#46140)
• Fixed incorrect struct reduction logic (#46151)
• Fixed error return from highlighter when search results are empty (#46111)
• Corrected logic for loading raw data for fields (#46155)
• Fixed cursor movement issue after skipping chunks in index (#46055)
• Corrected loop logic for TIMESTAMPTZ scalar index output (#46110)
• Fixed setting default values for geometry fields via RESTful API (#46064)
• Implemented fast fail if any component is not ready on startup (#46070)

client/v2.6.2

Release Notes - Milvus Go SDK v2.6.2

Summary

This release introduces QueryIterator support for efficient large result set iteration and adds Struct Array field type support in the Go SDK.

New Features

QueryIterator Support

Added QueryIterator for efficiently iterating over large query result sets using PK-based pagination (#46633).

• Support for Int64 and VarChar primary key types for automatic pagination
• Configurable options via QueryIteratorOption:
  • WithBatchSize(int) - Set batch size for each iteration (default: 1000)
  • WithIteratorLimit(int64) - Set overall limit of entries to iterate
  • WithFilter(string) - Set filter expression
  • WithOutputFields(...string) - Specify output fields
  • WithPartitions(...string) - Specify partition names
  • WithConsistencyLevel(ConsistencyLevel) - Set consistency level

Usage Example:

  opt := milvusclient.NewQueryIteratorOption("collection_name").
      WithBatchSize(500).
      WithFilter("age > 18").
      WithOutputFields("id", "name", "vector")

  iter, err := client.QueryIterator(ctx, opt)
  for {
      rs, err := iter.Next(ctx)
      if errors.Is(err, io.EOF) {
          break
      }
      // process rs...
  }

Enhancements

Struct Array Field Type Support (#45291)

Added support for Struct Array field type in Go SDK:

• New columnStructArray type implementing the Column interface
• Support for parsing StructArrayField from protobuf responses
• Schema construction support for struct array fields via entity.FieldTypeArray

Test Infrastructure Improvements (#45113)

• Refactored go_client test wrapper to use struct embedding pattern
• Improved test structure and organization

Related PRs

• #45113 - Refactor go_client test wrapper
• #45291 - Support struct array field type
• #46633 - Add QueryIterator support

milvus-2.5.24

Release note is coming soon.## v2.5.24

Release date: December 23, 2025

Milvus version	Python SDK version	Node.js SDK version	Java SDK version
2.5.24	2.5.18	2.5.13	2.5.14

We are excited to announce Milvus 2.5.24. This release prioritizes system security and stability, featuring critical updates to dependencies to address high-severity vulnerabilities. Additionally, it resolves significant correctness issues in Hybrid Search and JSON query processing, ensuring reliable performance under complex workloads. We strongly recommend all users upgrade to this version.

Improvements

• Upgraded Go version to 1.24.11 to address security vulnerabilities (#46114).
• Upgraded golang.org/x/crypto dependency to fix CVE vulnerabilities (#46112).
• Optimized the loading performance of BM25 statistics for sparse vector search (#45780).
• Refined logic to exclude stopping nodes from resource groups, improving cluster load balancing stability (#45982).

Bug fixes

• Improved startup logic to fail fast if critical components are not ready, aiding in troubleshooting (#46115).
• Fixed a data scanning issue where the cursor was not correctly updated after skipping data chunks (#46078).
• Fixed undefined behavior that could occur when dumping database snapshots (#45613).
• Fixed an issue where default values for geometry fields were not correctly set via the RESTful API (#46065).
• Fixed an issue where the total search count was incorrectly returned as zero during hybrid searches (#45843).
• Fixed a logic error where QueryCoord ignored stopping nodes during watcher events (#45964).
• Fixed a potential QueryNode crash caused by memory handling in JSON contains expressions (#45785).