chore(deps): update opensearchproject/opensearch docker tag to v3.9.0

[renovate]

This PR contains the following updates:

Package	Update	Change
opensearchproject/opensearch	minor	3.8.0 → 3.9.0

---

Release Notes

opensearch-project/OpenSearch (opensearchproject/opensearch)

v3.9.0

Compare Source

Version 3.9.0 Release Notes

Compatible with OpenSearch and OpenSearch Dashboards version 3.9.0

Features

• Add DELETE task API for removing stored completed task results (#​21727)
• Add deployment drain and finish APIs for zero-downtime node deployments (#​21448)
• Add adaptive per-action concurrency limiting module with Vegas, Gradient2, and AIMD algorithms (#​22312)
• Add virtual thread-per-task executor support to ThreadPool (#​22485)
• Add extensible payload decoder framework for pull-based ingestion (#​22364)
• Add plugin SPI for dynamic field-type inference and dynamic-template types (#​22607)
• Add pluggable PrimaryOperationPolicy to EngineConfig for cross-cluster replication (#​22886)
• Add object-store fencing token for remote store primary term validation (#​22774)
• Auto-restore remote store primaries on node loss when no valid copy survives (#​22904)
• Add FieldDomain metadata producer API for enhanced can_match shard pruning (#​22483)
• Add intra-segment search support for histogram, auto_date_histogram, and range aggregations (#​22531)
• Add MPP-style distributed join and aggregation execution for the analytics engine (#​21844)

Enhancements

• Add cluster.blocks.read.auto_release setting to prevent unconditional auto-release of user-set index read blocks (#​22610)
• Add cluster-level remote-store flush settings for uncommitted segments (#​22935)
• Restore size-based periodic flush on remote-store shards (#​22865)
• Apply index.periodic_flush_interval changes to running shards dynamically (#​22991)
• Use ordinals instead of byte values for bucket keys in multi-term aggregations (#​21033)
• Optimize bucket counting and tracking in LongKeyedBucketOrds with O(1) lookups (#​22450)
• Speed up reading history operations from Lucene via sequential stored-fields reader (#​22603)
• Use prefixQuery to speed up wildcardQuery and regexpQuery for wildcard field type (#​22510)
• Preserve constant_score through filtered aliases to maintain Lucene no-scoring fast path (#​22776)
• Promote lowest-version replica on document-replication failover to prevent stuck replicas during rolling upgrades (#​22783)
• Avoid cancellation checks on every advance() call for posting enums to reduce CPU overhead (#​22856)
• Avoid quadratic identifier scan when removing snapshots (#​22968)
• Stream translog parts instead of buffering the whole file for upload (#​23015)
• Change tiered cache default partition count to 16 for improved hit rates on high-core instances (#​23032)
• Bound Arrow native memory pools when node.native_memory.limit is unknown (#​22762)
• Lower default arrow.flight.channel.outbound_buffer_threshold from 64 MB to 512 KB (#​22697)
• Deprecate RestHighLevelClient in favor of opensearch-java client (#​23008)
• Remove Jackson 2.x dependencies from OpenSearch core (#​22704)
• Remove legacy client benchmark and noop API plugin modules (#​21839)
• Preserve routing in Translog.Delete for CDC-based replication (#​22584)
• Exempt server-injected preferences from strict weighted routing check and scope them per index (#​22571)
• Reject out-of-range WLM node threshold updates at validation time (#​22649)
• Return HTTP 400 for too_many_clauses and too_many_nested_clauses errors (#​21725)
• Refactor forPattern if-else chains to switch on FormatNames enum for O(1) dispatch (#​22568)
• Drain active streams before closing Flight channel to prevent spurious leak errors (#​22642)
• Skip stream connections to nodes without a stream address in mixed-version clusters (#​22810)
• Remove duplicate native memory limit setting from NodeDuressSettings (#​22620)

Bug Fixes

• Fix ApproximateScoreQuery.equals for query cache correctness (#​22737)
• Fix NPE on collapse with search_after sort using _score or _doc (#​23029)
• Fix NPE when creating a match-all composable index template without an inline template (#​22421)
• Fix NPE in routing table checksum when a shard has no primary (#​22723)
• Fix ClassCastException in mixed rollup and raw index avg aggregation reduce (#​22658)
• Fix parent-bucket scoping in StreamStringTermsAggregator for nested terms aggregations (#​21447)
• Fix cache entries removed before promotion being retained indefinitely (#​22662)
• Fix stale fielddata retention by invalidating exact key on reader close (#​22491)
• Use point-in-time key snapshot for fielddata cache cleanup to prevent missed entries (#​22499)
• Invalidate exact bitset filter cache keys on reader close (#​22498)
• Make ICache keys() iteration safe under concurrent mutation (#​22542)
• Fix stranded snapshot markers on publish failure with bounded retry (#​22518)
• Fix OpenSearch process not exiting when startup fails due to StartupException (#​22259)
• Bound completion suggester max_determinized_states to prevent heap exhaustion (CVE-2026-63136) (#​22924)
• Fix silent hang when REST response body cannot be serialized (#​22840)
• Fix opensearch-heap-prof reset crashing on non-numeric lg_prof_sample argument (#​22970)
• Re-register persistent task throttling keys to prevent cluster state restore failure (#​22656)
• Fix node crash on partially-consumed native stream teardown (#​22754)
• Fix virtual-thread scheduler deadlock from Flight client stream failure logging (#​22743)
• Fix constant ~5s stall in coordinator reduce teardown on limited queries (#​22609)
• Fix 500 on PPL query over alias spanning mapping-less index (#​22665)
• Fix use slice() instead of clone() for multipart uploads in RemoteDirectory to prevent AlreadyClosedException (#​22705)
• Gate internal ignore settings on remote data attributes in snapshot restore (#​22570)
• Allow read access to Kubernetes pod cgroup paths (#​22846)
• Fix HTTP connection leaks in plugin install command (#​22507)
• Fix ambiguous Node type reference in build.gradle (#​22514)
• Revert default SSE type to bucket_default to fix S3-compatible storage uploads (#​23041)

Maintenance

• Bump Apache HttpCore5 to 5.4.3 and Apache HttpClient5 to 5.6.4 to fix TLS latency regression (#​22731)
• Update AWS SDK to 2.54.2 to fix S3 upload failures (#​22816)
• Upgrade Lucene to 10.5.1 (#​22898)
• Update bundled JDK to 25.0.4.1+1 (#​22808)
• Update Gradle to 9.7.1 (#​22719)
• Update log4j to 2.25.5 (CVE-2026-49844) (#​22923)
• Bump Jackson to 2.22.2 (#​22847)
• Update OpenTelemetry to 1.66.0 and OpenTelemetry Semconv to 1.44.0 (#​23033)
• Update Project Reactor to 3.8.7 and Reactor Netty to 1.3.7 (#​22848)
• Update SnakeYaml to 2.7 and SnakeYaml Engine to 3.1.1 (#​23006)
• Bump Apache RAT dependency from 0.15 to 0.18 (#​22770)
• Bump com.squareup.okio:okio from 3.16.4 to 3.17.0 (#​21617)
• Bump com.sun.xml.bind:jaxb-impl from 4.0.7 to 4.0.8 (#​21614)
• Bump org.apache.commons:commons-configuration2 from 2.15.0 to 2.15.1 (#​21947)
• Bump org.carrot2:morfologik-stemming from 2.1.9 to 2.2.0 (#​22156)
• Bump org.jsoup:jsoup from 1.22.2 to 1.23.1 (#​22641)
• Bump org.jspecify:jspecify from 1.0.0 to 1.0.1 (#​22640)
• Bump org.jline:jline from 4.0.14 to 4.4.2 (#​23001)
• Bump com.google.jimfs:jimfs from 1.3.1 to 1.3.2 (#​22866)
• Bump com.github.spotbugs:spotbugs-annotations from 4.10.1 to 4.10.4 (#​22158, #​23000)
• Bump com.diffplug.spotless to 8.10.1 (#​22874)
• Remove obsolete OS packaging tests (#​22879)
• Replace deprecated AccessController in core libraries and plugins (#​22503, #​22988)

---

Configuration

📅 Schedule: (UTC)

• Branch creation
  • At any time (no schedule defined)
• Automerge
  • At any time (no schedule defined)

🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.

♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.

🔕 Ignore: Close this PR and you won't be reminded about this update again.

---

• If you want to rebase/retry this PR, check this box

---

This PR was generated by Mend Renovate. View the repository job log.

[coderabbitai]

Important

Review skipped

Bot user detected.

To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Advanced

Run ID: e948a691-3c92-4c2f-8f22-1d93cf33831c

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

• Autopilot · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

Autopilot is currently an internal CodeRabbit preview.

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}