Indexing performance degrading over time

Elastic Stack Elasticsearch
55

Well, we have tried to change GC, here are the results:

  • Default GC - worked fine till it has been changed;
  • G1 GC without -XX:+DisableExplicitGC - no, didn't helped. One of nodes still becomes overloaded with the same symptoms.
  • CMS GC (original elasticsearch.in.sh without changes) - works perfect, and even throughput is greater, without long GC pauses.

Yeah, you are right with your default configuration. Well done, guys, keep up the good work!

As for G1, may be submit a bug for it? But I don't know how it can be reproduced in code.