Elasticsearch version 6.1.1edit


  • Use AmazonS3.doesObjectExist() method in S3BlobContainer #27723
  • Ensured the watcher thread pool size is reasonably bound. In particular, the watcher thread pool size is now five times the number of processors until 50 threads are reached. If more than 50 cores exist and 50 threads exist, the watch thread pool size grows to match the number of processors.

Bug Fixesedit

Inner Hits
  • Add version support for inner hits in field collapsing (#27822) #27833 (issue: #27822)
Java REST Client
  • Do not use system properties when building the HttpAsyncClient #27829 (issue: #27827)
  • Data collectors now all share the same cluster state that existed at the beginning of data collection. This removes the extremely rare race condition where the cluster state can change between some data collectors, which could cause temporary issues in the Monitoring UI.
  • Fix routing with leading or trailing whitespace #27712 (issue: #27708)
Sequence IDs
  • Recovery from snapshot may leave seq# gaps #27850
  • No longer unidle shard during recovery #27757 (issue: #26591)
  • Fixed the pagerduty action to send context data. For more information, see PagerDuty Action.


  • update ingest-attachment to use Tika 1.17 and newer deps #27824