See issues on GitHub

Release Notes

Regression:

  • Remove Index Reader warmer introduced in 0.90.6 as it is not a good default behavior for all use cases. This will be reimplemented as an opt in feature.  #4078 & #4079

Enhancements:

  • Expose maximum heap settings + heap usage in percent in JVM memory stats #4145
  • Postings highlighter to support highlighting a query other than the search query #4121
  • Index Stats: Add support for segment count stats #4101

Bug fixes:

  • Query DSL: Wrong result on bool filter with 'must_not' and 'geo_distance' #4130
  • A potential deadlock in the BulkProcessor when client fails while sending the request to the cluster #4153
  • IndexShardGatewayService should not call post_recovery if shard is in STARTED state #4147
  • NPE for has_child query if number of results exceed certain limit #4135
  • Using JAVA_OPTS might fail on certain in inputs in service.bat #4127
  • _default_ mapping not applied when using separate master/data nodes #4124
  • Postings highlighter doesn't return snippets when using "path":"just_name" #4116
  • External method to set rootTypeParsers in DocumentMapperParser incorrect #4113
  • Unable to create a nested filtered alias on a dataless master #4112
  • Highlight hit object cache gets reset per hit, resulting in a performance penalty #4106
  • The new Postings highlighter may result in wrong highlighting #4103
  • Protection against double closing a XContentBuilder #4100
  • indices.recovery.concurrent_small_file_streams is not dynamically settable #4094
  • A work around for Lucene bug which causes increased memory usage when using the RAM Directory #4093
  • Fixed a racing condition causing rivers to potentially not start #4089 , #3840
  • Fixed an issue preventing the  keep words token filter to load words from a file #4073
  • Potentially failing to resolve aliases in  MoreLikeThisQuery #4045
  • NPE when percolating a document that contains a completion field #4028
  • Add an URI param to set track_scores  in search requests #2986