See issues on GitHub

Release Notes

New features:

  • Plugins can now expose versions #2784
  • Rescoring supports multiple sequential runs #4748

Enhancements:

  • Upgrade to Lucene 4.6.1 #4897
  • Recovery: Translog parsing is not throttled anymore #4890
  • Packaging: Mark lucene-expression as provided in pom.xml to prevent too much unused dependent jars in the distribution #4859
  • Plugins: Serving _site plugins now also read index.html in sub directories #4845
  • Aggregations: Rest API now is consistent across all multi-bucket aggs #4926
  • Aggregations: Huge cleanup in the java api #4922
  • cat API: Add thread pool endpoint #4907
  • cat API: Headers are now consistent and complete #4852
  • cat API: RestTable.renderValue() now support tera and peta values #4871
  • cat API: Added dedicated thread pool statistics API#4907
  • REST Get Field Mapping API: Be consistent in response data being returned in case of a not found field #4738
  • Transport: Remove redundant version checks in serialisation #4731
  • Snapshot/Restore API: Added throttling #4855
  • Snapshot/Restore API: Improve boolean parsing #
  • Plugins: Added version support #2784

Bug fixes:

  • Support for source parameter in some REST APIs
    • REST scroll API supports source parameter #4941
    • REST percolate API supports source parameter #4903
    • REST mpercolate API supports source parameter #4900
    • REST msearch API supports source parameter #4901
    • REST mget API supports source parameter #4892
    • REST mtermvectors API supports source parameter #4902
    • Fixed potential problem when no source and no body were specified #4924
  • Aggregations: Sorting terms agg by sub-aggegation doesn't respect asc/desc when executing on a single shard #4951
  • Aggregations: Sub aggregations on a single shard properly propagated #4843
  • Hot Threads API: Preventing possible ArrayIndexOutOfBoundsException if the number of request busiest threads is greater than actual busy threads #4927
  • Rivers: Ensure startup by specifying the primary preference to query the river configuration from #4864
  • Bulk API: If JSON body data has to be parsed to extract metadata from it (like the _id) and fails, ensure, this does not affect the entire request instead of single bulk operation #4745
  • Allocation: During testing, the BalancedShardAllocator could make non-deterministic rebalance decisions #4867
  • REST Cluster State API: Filtered cluster state still returned empty elements instead of omitting #4885
  • Percolator API: Response always includes the matches key, independent from results #4881
  • Query API: Ensure caching of date range filters that use now with rounding #4947
  • Query API: Filtered query now parses _name correctly #4960
  • Query API: Prevent possible NullPointerException in completion suggest #4970, #4788