v2.6.25

## v2.6.25

Release date: September 29, 2026

| Milvus Version | Python SDK Version | Node.js SDK Version | Java SDK Version | Go SDK Version |
| -------------- | ------------------ | ------------------- | ---------------- | -------------- |
| 2.6.25         | 2.6.17             | 2.6.17              | 2.6.25           | 2.6.5          |

We are excited to announce the release of Milvus v2.6.25! This release tightens security with stricter RBAC checks, a new `ImportBinlog` privilege, and masking of sensitive configuration values. It also adds snappy and s2 gRPC compression, and fixes stability issues around shard leader movement, scalar index replacement, and Pulsar channel takeover.

### Improvements

- Added per-resource-group load config convergence reporting to the management API, covering replica distribution, serviceability, query visibility, and residual resources ([#53518](https://github.com/milvus-io/milvus/pull/53518))
- Added snappy and s2 gRPC compression, made the zstd codec configurable (`grpc.client.compressionAlgorithm`, `compressionLevel`, `compressionCRC`, `compressionConcurrency`), and hardened zstd decompression against oversized messages ([#53599](https://github.com/milvus-io/milvus/pull/53599))
- Masked sensitive values and omitted undeclared keys in configuration views and diagnostic logs, so credentials and unrelated environment variables are no longer exposed ([#53613](https://github.com/milvus-io/milvus/pull/53613))
- Required the new cluster-level `ImportBinlog` privilege for binlog (`backup=true`) and L0 imports, and rejected regular imports whose file paths point into Milvus's internal storage directories ([#53681](https://github.com/milvus-io/milvus/pull/53681))
- Sped up writing Tantivy index results into bitsets, up to 7x faster per doc ID for contiguous IDs in microbenchmarks ([#53701](https://github.com/milvus-io/milvus/pull/53701))
- Hardened RBAC enforcement by disabling the Dummy RPC's query mode, which bypassed authorization and rate limiting, and by requiring Search privilege on the collection for RunAnalyzer requests that specify one ([#53812](https://github.com/milvus-io/milvus/pull/53812))
- Reduced redundant DescribeCollection RPCs from Proxy to the coordinator during insert, delete, and upsert by resolving collection channels from the Proxy metadata cache ([#53793](https://github.com/milvus-io/milvus/pull/53793))

### Bug fixes

- Fixed a DataNode memory leak where a fully built index or analyze result was never released if its task was canceled during the build ([#53349](https://github.com/milvus-io/milvus/pull/53349))
- Fixed an issue where loading segments with FLAT or BIN_FLAT indexes could fail during incremental index loading, especially under restrictive resource limits ([#53383](https://github.com/milvus-io/milvus/pull/53383))
- Fixed an issue where index builds on segments containing malformed JSON retried indefinitely; such tasks now fail fast with a permanent error instead of looping across workers ([#53429](https://github.com/milvus-io/milvus/pull/53429))
- Fixed an issue where replacing a scalar index on a loaded collection could leave the old index loaded indefinitely and cause repeated reloads; both QueryCoord and QueryNode must be upgraded for the complete fix ([#53537](https://github.com/milvus-io/milvus/pull/53537))
- Fixed an issue where a StreamingNode could keep writing to a Pulsar-backed channel after another node took it over; `pulsar.producerAccessMode` now defaults to `exclusive` (enforcement requires Pulsar 2.8.0+; set it to `shared` to keep the previous behavior) ([#53540](https://github.com/milvus-io/milvus/pull/53540))
- Fixed an issue where sustained GIS query load could stall inserts into growing segments when the geometry cache (`enableGeometryCache`) was enabled ([#53610](https://github.com/milvus-io/milvus/pull/53610))
- Fixed an issue where Milvus components could hang during shutdown or block on logging when the async log output was stuck ([#53626](https://github.com/milvus-io/milvus/pull/53626))
- Fixed a QueryNode crash when a LIKE filter on an NGRAM-indexed field contained a literal with fewer than `min_gram` multi-byte characters, such as a single full-width punctuation mark ([#53668](https://github.com/milvus-io/milvus/pull/53668))
- Fixed an issue where search, query, and statistics requests kept failing with "collection not loaded" for an extended period after QueryCoord moved a shard leader to another QueryNode ([#53810](https://github.com/milvus-io/milvus/pull/53810))
- Fixed a crash that could occur when Arrow data passed through the C Data interface contained zero-length buffers, such as all-null string columns, by upgrading the Arrow Go dependency to v17.0.2 ([#53738](https://github.com/milvus-io/milvus/pull/53738))
- Fixed an issue where a restarting node reported ready on `/healthz` before all of its components had registered, causing requests routed to it during startup to fail with connection refused ([#53736](https://github.com/milvus-io/milvus/pull/53736))
- Fixed an issue where Parquet import stored NULL instead of the default value for null rows in Timestamptz fields that define a default value ([#53809](https://github.com/milvus-io/milvus/pull/53809))