Elasticsearch version 7.14.2

edit

Elasticsearch version 7.14.2

edit

Also see Breaking changes in 7.14.

Known issues

edit
  • 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.

  • Aggregations: In Elasticsearch 7.14.0–7.15.0, when a cross-cluster search (CCS) request is proxied, the memory for the aggregations on the proxy node will not be freed. The trigger is cross-cluster search using aggregations where minimize roundtrips is not effective (for example, when minimize roundtrips is explicitly disabled, or implicitly disabled when using scroll, async and point-in-time searches).

    This affects Kibana CCS aggregations because Kibana uses async search by default. This issue can also happen in all modes of remote connections configured for cross-cluster search (sniff and proxy). In sniff mode, we only connect to a subset of the remote nodes (by default 3). So if the remote node we want to send a request to is not one of those 3, we must send the request as a proxy request. The workaround is to periodically restart nodes with heap pressure.

    We have fixed this issue in Elasticsearch 7.15.1 and later versions. For more details, see #78404.

  • 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)
  • The deprecated index.mapper.dynamic setting can break your cluster. It can only be set using the Update index settings API. Symptoms include nodes failing to start or shards failing to allocate. Do not use this setting in versions prior to 7.17.22. The bug is fixed in 7.17.22. (issue: #109160)

Enhancements

edit
Mapping
  • Restore public visibility of methods on RootObjectMapper #77274

Bug fixes

edit
EQL
  • Use the document fields only excluding the metadata fields #77203 (issue: #77152)
Machine Learning
  • Fix failure on datafeed preview with date_nanos time field #77109
  • Relocating job should not fail if blocked on revert #77207 (issue: #75377)
  • Retain job task upon fail during opening #77206
  • Retry reset if existing task was terminated abruptly #77328 (issue: #77207)
SQL
  • Fix disjunctions (and IN) with multiple date math expressions #76424 (issue: #76336)
Snapshot/Restore
  • Fix queued snapshot clone not starting after data node drops #77111 (issue: #77101)