Elasticsearch version 7.12.1edit

Also see Breaking changes in 7.12.

Known issuesedit

  • Frozen tier: (Windows only) The frozen data tier relies on multiple caching mechanisms to speed up access to searchable snapshot files. One of these caches uses sparse files to avoid creating large files on disk when it is not strictly required. A bug prevented files from being created with the right options to enable sparse support on Windows, leading Elasticsearch to create potentially large files that can end up consuming all the disk space.

    This issue is fixed in Elasticsearch versions 7.15.2 and later. There is no known workaround for earlier versions. Filesystems that enable sparse files by default are not affected. For more details, see #79371.

  • Snapshot and restore: If an index is deleted while the cluster is concurrently taking more than one snapshot then there is a risk that one of the snapshots may never complete and also that some shard data may be lost from the repository, causing future restore operations to fail. To mitigate this problem, set snapshot.max_concurrent_operations: 1 to prevent concurrent snapshot operations:

    PUT _cluster/settings
      "persistent" : {
        "snapshot.max_concurrent_operations" : 1

    This issue is fixed in Elasticsearch versions 7.13.1 and later. It is not possible to repair a repository once it is affected by this issue, so you must restore the repository from a backup, or clear the repository by executing DELETE _snapshot/<repository name>/*, or move to a fresh repository. For more details, see #73456.

  • Parsing a request when the last element in an array is filtered out (for instance using _source_includes) fails. This is due to a bug in Jackson parser. Fixed in Elasticsearch 8.6.1 (#91456)


  • Prevent snapshot backed indices to be followed using CCR #70580
Features/Data streams
  • Allow closing a write index of a data stream #70908 (issues: #70861, #70903)
  • Improve data stream rollover and simplify cluster metadata validation for data streams #70934 (issue: #70905)
Machine Learning
  • Make ML native processes work with glibc 2.33 on x86_64 #1828
  • Adapt frozen write buffer and thread pool #71172
  • Add CFS index caching support for full_copy searchable snapshots #70646
  • Adjust the length of blob cache docs for Lucene metadata files #69431 (issue: #69283)
  • Always use CacheService for caching metadata blobs #70668 (issues: #70728, #70763)
  • Lazily load soft-deletes for searchable snapshot shards #69203

Bug fixesedit

  • Fix percentiles agg in slow log after transport #70318
  • Fix position increment gap on phrase/prefix analyzers #70096 (issue: #70049)
Cluster Coordination
  • Fully initialize cluster state on ephemeral nodes #71466
  • Map data tiers roles onto DATA legacy role for <7.3 #71628 (issue: #71464)
  • Delete data stream in ILM delete action if last index in data stream #69637
  • Fix handling of non-integer port values in community_id processor #70148 (issue: #70131)
  • Fix typo in validation for destination port of community ID processor #70883
  • Templates match indices with date math expressions #71433 (issue: #69727)
  • URI parts processor handles URLs containing spaces #71559 (issue: #70947)
  • Enable setting master node timeout in Watcher start/stop requests #70425
  • Do not over-allocate when resizing in GeoGridTiler #70159
  • Fix geo_line agg behavior with missing values #69395 (issue: #69346)
  • Fix infinite loop when polygonizing a circle with centre on the pole #70875
  • Fix overflow bug in SortingNumericDocValues #70154
  • Always wrap date parsing exception into IllegalArgumentException #71038
  • Remove loop counter for foreach loops #71602 (issue: #71584)
Machine Learning
  • Make ML memory tracker more robust to flipping on/off master nodes #71067 (issue: #68685)
  • Legacy geo-shape mapper not detecting [points_only] parameter #70765 (issue: #70751)
  • Enforce and document dedicated client version compatibility #70451 (issue: #70400)
  • Fix manifest version tag in Tableau connector #71524
  • Resolve attributes recursively for improved subquery support #69765 (issue: #67237)
  • Verify binary fields found in non-project to have the doc_values #69128 (issue: #68229)
  • Fix exception when merging completion suggestions #70414 (issue: #70328)
  • Fix query cache reporting negative used memory #70273 (issue: #55434)
  • Fix search states of CCS requests in mixed cluster #70948 (issue: #52741)
  • Avoid atomic overwrite tests on FS repositories #70483 (issue: #70303)
  • Drop alloc filters on mount of searchable snapshot #70007 (issue: #69759)
  • Fix source only snapshot permanently broken on broken _snapshot directory #71459
  • Fix isDedicatedVotingOnlyNode #71358 (issue: #71352)
  • Fix human readable xcontent for snapshots in progress and deletion #70256
  • Fix wrongly computed offset in checksum #69441 (issues: #69415, #69437)
  • Only allocate partial shards to nodes with cache #69904
  • Optimized read footer checksum from FileInfo #69415
  • Reduce memory use of parallel Azure blob deletes #71330 (issue: #71267)
  • Stop blocking snapshot deletes due to concurrency limits #71050