8.17.2.asciidoc 3.0 KB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152
  1. [[release-notes-8.17.2]]
  2. == {es} version 8.17.2
  3. Also see <<breaking-changes-8.17,Breaking changes in 8.17>>.
  4. [[bug-8.17.2]]
  5. [float]
  6. === Bug fixes
  7. Authentication::
  8. * Fix NPE on disabled API auth key cache {es-pull}120483[#120483]
  9. Data streams::
  10. * Map `scope.name` as a dimension {es-pull}120590[#120590]
  11. ES|QL::
  12. * Use `field_caps` native nested fields filtering {es-pull}117201[#117201] (issue: {es-issue}117054[#117054])
  13. Machine Learning::
  14. * Fix infer on and elasticsearch service endpoint created with a deployment id {es-pull}121428[#121428]
  15. * Fix inference update API calls with `task_type` in body or `deployment_id` defined {es-pull}121231[#121231]
  16. * Skip Usage stats update when ML is disabled {es-pull}121559[#121559] (issue: {es-issue}121532[#121532])
  17. Mapping::
  18. * Fix synthetic source issue with deeply nested ignored source fields {es-pull}121715[#121715]
  19. * Lower `_source.mode` mapping attribute deprecation issue level {es-pull}120059[#120059]
  20. Ranking::
  21. * Fix LTR rescorer throws 'local model reference is null' on multi-shards index when explained is enabled {es-pull}120717[#120717] (issue: {es-issue}120739[#120739])
  22. * LTR sometines throw `NullPointerException:` Cannot read field "approximation" because "top" is null {es-pull}120809[#120809]
  23. * Normalize negative scores for `text_similarity_reranker` retriever {es-pull}120930[#120930] (issue: {es-issue}120201[#120201])
  24. * Update Text Similarity Reranker to Properly Handle Aliases {es-pull}120062[#120062] (issue: {es-issue}119617[#119617])
  25. Search::
  26. * Add back `keep_alive` to `async_search.submit` rest-api-spec {es-pull}120781[#120781]
  27. * Fix NPE caused by race condition in async search when minimise round trips is true {es-pull}117504[#117504]
  28. * Fix potential file leak in ES816BinaryQuantizedVectorsWriter {es-pull}120014[#120014] (issue: {es-issue}119981[#119981])
  29. * Use approximation to advance matched queries {es-pull}120133[#120133] (issue: {es-issue}120130[#120130])
  30. Snapshot/Restore::
  31. * Cheaper snapshot-related `toString()` impls {es-pull}121283[#121283]
  32. * Issue S3 web identity token refresh call with sufficient permissions {es-pull}119748[#119748] (issue: {es-issue}119747[#119747])
  33. [[enhancement-8.17.2]]
  34. [float]
  35. === Enhancements
  36. Ingest Node::
  37. * Improve memory aspects of enrich cache {es-pull}120256[#120256] (issues: {es-issue}96050[#96050], {es-issue}120021[#120021])
  38. * `VALUES` aggregate function can run for a long, long time when collecting many, many groups. Hundreds of thousands of groups can run on one thread for many minutes and millions of groups run for days. It is O(n^2^) with the number of groups. These will not respond to the tasks cancellation API the whole time. Fixed by {es-pull}123073[#123073] and available in 8.16.5, 8.17.3, 8.18.0, and all releases after that.
  39. * Some valid queries using an `ENRICH` command can fail when a match field is used that is absent from some indices or shards, either with a 500 status code due to `NullPointerException` or `ClassCastException` or with a 400 status code and `IllegalArgumentException`. This is fixed in {es-pull}126187[#126187].