23417 Commits

Author SHA1 Message Date
cqy123456
822588302a
enhance: embedding_list support mmap in MemVectorIndex (#44764)
issue: https://github.com/milvus-io/milvus/issues/44702

Signed-off-by: cqy123456 <qianya.cheng@zilliz.com>
2025-10-15 15:22:00 +08:00
wei liu
529c98520c
enhance: Add nullable support for Geometry and Timestamptz types (#44846)
issue: #44800
This commit enhances the upsert and validation logic to properly handle
nullable Geometry (WKT/WKB) and Timestamptz data types:

- Add ToCompressedFormatNullable support for TimestamptzData,
GeometryWktData, and GeometryData to filter out null values during data
compression
- Implement GenNullableFieldData for Timestamptz and Geometry types to
generate nullable field data structures
- Update FillWithNullValue to handle both GeometryData and
GeometryWktData with null value filling logic
- Add UpdateFieldData support for Timestamptz, GeometryData, and
GeometryWktData field updates
- Comprehensive unit tests covering all new data type handling scenarios

Signed-off-by: Wei Liu <wei.liu@zilliz.com>
2025-10-15 14:04:00 +08:00
sparknack
3d3fa44745
fix: milvus-common update (#44798)
issue: #44501

Signed-off-by: Shawn Wang <shawn.wang@zilliz.com>
2025-10-15 11:24:00 +08:00
Spade A
c4f3f0ce4c
feat: impl StructArray -- support more types of vector in STRUCT (#44736)
ref: https://github.com/milvus-io/milvus/issues/42148

---------

Signed-off-by: SpadeA <tangchenjie1210@gmail.com>
Signed-off-by: SpadeA-Tang <tangchenjie1210@gmail.com>
2025-10-15 10:25:59 +08:00
yihao.dai
5ad8a29c0b
enhance: Speed up CDC scheduling (#44564)
Make CDC watch etcd replicate pchannel meta instead of listing them
periodically.

issue: https://github.com/milvus-io/milvus/issues/44123

---------

Signed-off-by: bigsheeper <yihao.dai@zilliz.com>
2025-10-15 10:15:59 +08:00
XuanYang-cn
fc46668812
fix: Disk encryption config missing (#44820)
See also: #44823

Signed-off-by: yangxuan <xuan.yang@zilliz.com>
2025-10-14 17:22:00 +08:00
Spade A
b8df1c0cc5
enhance: improve observability in trace for segcore scalar expression (#44260)
Ref https://github.com/milvus-io/milvus/issues/44259

This PR connects the trace between go and segcore, and add full traces
for scalar expression calling chain:
<img width="2418" height="960" alt="image"
src="https://github.com/user-attachments/assets/8cad69d7-bcb7-4002-a4e3-679a3641e229"
/>
<img width="2452" height="850" alt="image"
src="https://github.com/user-attachments/assets/8b44aed0-0f03-48a7-baa0-b022fee994ce"
/>
<img width="2403" height="707" alt="image"
src="https://github.com/user-attachments/assets/cd6f0601-0d5c-4087-8ed8-2385f1bc740b"
/>

---------

Signed-off-by: SpadeA <tangchenjie1210@gmail.com>
2025-10-14 17:15:59 +08:00
congqixia
92910117a8
fix: binlog count use correct value (#44830)
Related to #44789

---------

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-10-14 16:48:05 +08:00
cqy123456
4b84ba2189
fix:remove the limit of deduplicate case when disable autoindex (#44825)
issue: https://github.com/milvus-io/milvus/issues/44702

Signed-off-by: cqy123456 <qianya.cheng@zilliz.com>
2025-10-14 16:14:00 +08:00
yihao.dai
cebe923d4a
enhance: Make GetReplicateInfo API work at the pchannel level (#44809)
issue: https://github.com/milvus-io/milvus/issues/44123

Signed-off-by: bigsheeper <yihao.dai@zilliz.com>
2025-10-14 15:12:00 +08:00
Chun Han
2ea8d85c2f
feat: restful support run analyzer(#44803) (#44805)
related: #44803

Signed-off-by: MrPresent-Han <chun.han@gmail.com>
Co-authored-by: MrPresent-Han <chun.han@gmail.com>
2025-10-14 14:41:59 +08:00
yanliang567
9e5f9277c0
test: Split insert file and add test for allow insert auto id (#44801)
related issue: #44425
1. split insert.py into a few files: upsert.py, insert.py,
partial_upsert.py ...
2. add test for allow insert auto id

---------

Signed-off-by: yanliang567 <yanliang.qiao@zilliz.com>
2025-10-14 14:28:00 +08:00
Bingyi Sun
6cb1f7d7c6
enhance: optimize the performace of bitmap reverse lookup (#44804)
Signed-off-by: sunby <sunbingyi1992@gmail.com>
2025-10-14 11:57:58 +08:00
zhagnlu
2f178f810f
fix:fix json_contains(path, int) bug (#44814)
#44816

Signed-off-by: luzhang <luzhang@zilliz.com>
Co-authored-by: luzhang <luzhang@zilliz.com>
2025-10-14 00:19:59 +08:00
sparknack
df6a4dc1a0
fix: cachinglayer: avoid eviction during json handling (#44812)
issue: #44797

Signed-off-by: Shawn Wang <shawn.wang@zilliz.com>
2025-10-13 22:07:58 +08:00
Buqian Zheng
3140bd0ca6
enhance: enable default json stats (#44810)
issue: https://github.com/milvus-io/milvus/issues/44132

Signed-off-by: Buqian Zheng <zhengbuqian@gmail.com>
2025-10-13 21:29:59 +08:00
congqixia
33060d9cf2
fix: avoid concurrent Reset/Add operations on DataCoord metrics (#44789)
This commit addresses issue #44788 where the
`datacoord_stored_binlog_size` metric could become inaccurate when
multiple concurrent `GetMetrics` calls arrived at DataCoord.

### Problem

The original implementation called `Reset()` followed by `Add()`
operations on Prometheus metrics within the `GetQuotaInfo()` method.
When multiple goroutines invoked this method concurrently, race
conditions occurred:
- Thread 1: Reset() → Add(value1)
- Thread 2: Reset() → Add(value2)
- Result: Metrics could be reset multiple times and values added in an
interleaved fashion, leading to inaccurate and inflated metric values

### Solution

Changed the approach from `Reset() + Add()` to aggregating metric values
in local maps first, then using `Set()` to update metrics atomically:

1. Collect segment size data into local maps:
   - `storedBinlogSize`: tracks size per collection per segment state
   - `binlogFileSize`: tracks total file count per collection
   - `coll2DbName`: maps collection IDs to database names

2. After aggregation is complete, use `Set()` (instead of `Add()`) to
update metrics in a single operation per label combination

This ensures that concurrent `GetMetrics` calls don't interfere with
each other, as each invocation works with its own local state and only
updates the final metric value atomically.

---------

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-10-13 18:39:59 +08:00
aoiasd
1b17e16fc7
fix: expr filter return wrong result when skipped (#44778)
relate: https://github.com/milvus-io/milvus/issues/44777
Should return res with false if skipped. But now return vaild[0], it
almost be true.

Signed-off-by: aoiasd <zhicheng.yue@zilliz.com>
2025-10-13 18:33:59 +08:00
zhagnlu
3dd5deb70a
fix:disable using shredding for json_path contains digital (#44724)
#44132

Signed-off-by: luzhang <luzhang@zilliz.com>
Co-authored-by: luzhang <luzhang@zilliz.com>
2025-10-13 17:25:59 +08:00
Zhen Ye
53e8f150e8
fix: check if qn is sqn with label and streamingnode list (#44792)
issue: #44014

- On standalone, the query node inside need to load segment and watch
channel, so the querynode is not a embeded querynode in streamingnode
without `LabelStreamingNodeEmbeddedQueryNode`. The channel dist manager
can not confirm a standalone node is a embededStreamingNode.

Bug is introduced by #44099

Signed-off-by: chyezh <chyezh@outlook.com>
2025-10-13 16:33:59 +08:00
XuanYang-cn
a444e2f937
test: Increase PyMilvus version to 2.7.0rc44 for master branch (#44796)
Automated daily bump from pymilvus master branch. Updates
tests/python_client/requirements.txt.

Signed-off-by: XuanYang-cn <xuan.yang@zilliz.com>
2025-10-13 15:54:03 +08:00
sparknack
c8a4d6e2ef
enhance: add cachinglayer management for TextMatchIndex (#44741)
issue: #41435, #44502

Signed-off-by: Shawn Wang <shawn.wang@zilliz.com>
2025-10-13 14:37:58 +08:00
sparknack
6d5b41644b
enhance: remove logical usage checks during segment loading (#44743)
issue: #41435

Signed-off-by: Shawn Wang <shawn.wang@zilliz.com>
2025-10-13 14:21:58 +08:00
congqixia
f5f053f1d2
enhance: Refactor privilege management by extracting privilege cache into separate package (#44762)
Related to #44761

This commit refactors the privilege management system in the proxy
component by:

1. **Separation of Concerns**: Extracts privilege-related functionality
from MetaCache into a dedicated `internal/proxy/privilege` package,
improving code organization and maintainability.

2. **New Package Structure**: Creates `internal/proxy/privilege/` with:
   - `cache.go`: Core privilege cache implementation (PrivilegeCache)
   - `result_cache.go`: Privilege enforcement result caching
   - `model.go`: Casbin model and policy enforcement functions
   - `meta_cache_adapter.go`: Casbin adapter for MetaCache integration
   - Corresponding test files and mock implementations

3. **MetaCache Simplification**: Removes privilege and credential
management methods from MetaCache interface and implementation:
   - Removed: GetCredentialInfo, RemoveCredential, UpdateCredential
- Removed: GetPrivilegeInfo, GetUserRole, RefreshPolicyInfo,
InitPolicyInfo
   - Deleted: meta_cache_adapter.go, privilege_cache.go and their tests

4. **Updated References**: Updates all callsites to use the new
privilegeCache global:
- Authentication interceptor now uses privilegeCache for password
verification
- Credential cache operations (InvalidateCredentialCache,
UpdateCredentialCache, UpdateCredential) now use privilegeCache
- Policy refresh operations (RefreshPolicyInfoCache) now use
privilegeCache
- Privilege interceptor uses new privilege.GetEnforcer() and privilege
result cache

5. **Improved API**: Renames cache functions for clarity:
   - GetPrivilegeCache → GetResultCache
   - SetPrivilegeCache → SetResultCache
   - CleanPrivilegeCache → CleanResultCache

This refactoring makes the codebase more modular, separates privilege
management concerns from general metadata caching, and provides a
clearer API for privilege enforcement operations.

---------

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-10-13 11:15:58 +08:00
Zhen Ye
369c6eb206
enhance: support remove cluster from replicate topology (#44642)
issue: #44558, #44123
- Update config(A->C) to A and C, config(B) to B on replicate topology
(A->B,A->C) can remove the B from replicate topology
- Fix some metric error of CDC

Signed-off-by: chyezh <chyezh@outlook.com>
2025-10-13 11:07:58 +08:00
foxspy
e7a91f514c
enhance: overwriting current index type during index build stage (#44753)
issue: #44752

Signed-off-by: xianliang.li <xianliang.li@zilliz.com>
2025-10-11 18:31:58 +08:00
foxspy
0eb42a7870
enhance: support load params for vector index (#44747)
issue: #44746 

Support modifying vector index behavior during loading by change/add
aram knowhere.xxx.load.xxx.

Signed-off-by: xianliang.li <xianliang.li@zilliz.com>
2025-10-11 18:30:08 +08:00
congqixia
7b8ecdaad5
enhance: Add accesslog field for template value length info (#44723)
Related to #36672

Add accesslog field displaying value length for search/query request may
help developers debug related issues

---------

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-10-11 18:23:57 +08:00
XuanYang-cn
a3bdabb328
enhance: Unify compaction executor task state management (#44721)
Remove stopTask.
Replace multiple task tracking maps with single unified taskState map.
Fix slot tracking, improve state transitions, and add comprehensive test

See also: #44714

---------

Signed-off-by: yangxuan <xuan.yang@zilliz.com>
2025-10-11 17:53:57 +08:00
aoiasd
09865a5da5
fix: BM25 with boost return result not ordered. (#44744)
relate: https://github.com/milvus-io/milvus/issues/44758
Wrong code which should be `(result.seg_offsets_[i] >= 0 &&
result.seg_offsets_[j] < 0)`, but was `(result.seg_offsets_[j] >= 0 &&
result.seg_offsets_[j] < 0) ` now.
But because all placeholder which was offset -1, will fill with worst
distance value.
For IP, L2 or COSIN, it will be +inf or -inf. So sort distance was
enough.
But when use BM25, it will be NAN. Will case sort out of ordered.

Signed-off-by: aoiasd <zhicheng.yue@zilliz.com>
2025-10-11 17:17:58 +08:00
congqixia
faaf215913
enhance: Bump go version & builder image tag (#44757)
Bump go version to 1.24.6 fixing CVE-2025-47907

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-10-11 13:49:57 +08:00
cai.zhang
d8beafd6d0
fix: Skip running scalar index when segment was compacted (#44690)
issue: #44689

Signed-off-by: Cai Zhang <cai.zhang@zilliz.com>
2025-10-11 12:05:56 +08:00
cai.zhang
7a93cfe890
fix: Fix bug for nullable geometry (#44732)
issue: #44648

Signed-off-by: Cai Zhang <cai.zhang@zilliz.com>
2025-10-11 11:27:57 +08:00
congqixia
5ece760d73
fix: Pass fs via FileManagerContext when loading index (#44733)
Related to #44615

---------

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-10-11 09:55:57 +08:00
congqixia
78b266a44f
enhance: Bump builder image go version to v1.24.6 (#44739)
Bump go version fixing CVE issues

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-10-11 09:53:57 +08:00
Xiaofan
af9bf80726
Update README.md 2025-10-11 09:46:12 +08:00
wei liu
33d1e7de83
fix: Replace incorrect log import with milvus v2 log package (#44731)
issue: #44730
Fix the issue where logs were not outputting as expected due to
incorrect log package imports across multiple components.

Changes include:
- Add golangci-lint rule to forbid github.com/pingcap/log usage
- Replace github.com/pingcap/log with
github.com/milvus-io/milvus/pkg/v2/log

Signed-off-by: Wei Liu <wei.liu@zilliz.com>
2025-10-10 20:27:57 +08:00
sparknack
7e750190b6
enhance: add a size getter for tantivy inverted index (#44609)
issue: #41435

---------

Signed-off-by: Shawn Wang <shawn.wang@zilliz.com>
2025-10-10 17:43:57 +08:00
Spade A
208481a070
feat: impl StructArray -- support same names in different STRUCT (#44557)
ref: https://github.com/milvus-io/milvus/issues/42148

---------

Signed-off-by: SpadeA <tangchenjie1210@gmail.com>
2025-10-10 15:53:56 +08:00
nico
e5378a64bc
test: update test cases (#44651)
Signed-off-by: nico <cheng.yuan@zilliz.com>
2025-10-10 11:35:56 +08:00
sre-ci-robot
80c6212c43
[automated] Bump milvus version to v2.6.3 (#44716)
Bump milvus version to v2.6.3
Signed-off-by: sre-ci-robot sre-ci-robot@users.noreply.github.com

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
2025-10-10 10:45:56 +08:00
groot
81f0d498be
enhance: Support JSONL/NDJSON files for bulkinsert (#44602)
issue: https://github.com/milvus-io/milvus/issues/44567

Signed-off-by: yhmo <yihua.mo@zilliz.com>
2025-10-10 10:43:57 +08:00
zhenshan.cao
4279f166c6
enhance: Add refine logs for task scheduler in QueryCoord (#44577)
issue: https://github.com/milvus-io/milvus/issues/43968

Signed-off-by: zhenshan.cao <zhenshan.cao@zilliz.com>
2025-10-10 10:07:55 +08:00
congqixia
e83c7e0c92
fix: Use eventually & fix task id appear in both executing&completed (#44698)
Related to #44620

This PR:
- Use eventually instead of `time.Sleep` in accesslog writer unit test
- Make sure compaction task results have only one state from executor
API

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-10-10 10:05:56 +08:00
sre-ci-robot
221e27fc02
[automated] Bump milvus version to v2.6.3 (#44655)
Bump milvus version to v2.6.3
Signed-off-by: sre-ci-robot sre-ci-robot@users.noreply.github.com

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
2025-10-10 09:06:04 +08:00
Goodnight
dd9e26cad0
doc: fix typos in README (#44674)
This PR fixes several documentation issues in README.md

Correct typo "accelaration" → "acceleration".
Minor grammar and wording cleanups for clarity.
Capitalize language and product names: "python" → "Python", "Go", and
"Youtube" → "YouTube".

No functional code changes, documentation-only.

Signed-off-by: Mohamed Arbi Nsibi <mohammedarbinsibi@gmail.com>
2025-10-09 21:19:58 +08:00
congqixia
efbe6669ec
enhance: Make accesslog.$consistency_level represent actual value used (#44706)
Related to #44703

This PR:
- Add `SetActualConsistencyLevel` to `info.AccessInfo` interface and
related util method processing it
- Make `$consistency_level` returning actual value if set

---------

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-10-09 21:03:58 +08:00
Zhen Ye
97c987f313
fix: remove wrong start timetick to avoid filtering DML whose timetick is less than it. (#44691)
issue: #41611

- introduced by #44532

Signed-off-by: chyezh <chyezh@outlook.com>
2025-10-09 20:39:58 +08:00
congqixia
e55e63288e
enhance: [GoSDK] bump pkg dep to v2.6.3 (#44619)
See also #31293 

Bump pkg dependency version to v2.6.2

---------

Signed-off-by: Congqi Xia <congqi.xia@zilliz.com>
2025-10-09 19:43:57 +08:00
yihao.dai
4ae44f1aa9
fix: Fix CDC OOM due to high buffer size (#44607)
Fix CDC OOM by:
1. free msg buffer manually.
2. limit max msg buffer size.
3. reduce scanner msg hander buffer size.

issue: https://github.com/milvus-io/milvus/issues/44123

Signed-off-by: bigsheeper <yihao.dai@zilliz.com>
2025-10-09 19:31:58 +08:00