We recently upgraded our cluster from 5.4.1 to 5.6.2 and the GC time has started to skyrocket. The GC time has over the course of 4 weeks risen 200%. Restarting the individual nodes lowers the GC time to pre-update times, but immediately starts increasing again. There is no more searches and indexing going on than before the update. I am pretty new to ES, so pointer on where to look for the cause is much appreciated.
The cluster is 6 nodes running on AWS (c4.2xl) 7 GB heap for ES, 18 shards, a replication factor of 2, size of the index is 238.1 gb and total size is 714.4gb.
Any hin ton finding out why we started seeing increased GC times is much appreciated.