It seems we did have Splunk collection garbage collection stats via JMX. I see no Full GCs happening at that time. There are minor GCs but they happen fairly frequently and don’t use much CPU time.
Ralph > On Jan 17, 2018, at 10:52 AM, Ralph Goers <ralph.go...@dslextreme.com> wrote: > > We didn’t have gc logging enabled. I can certainly enable it to for the > future but that still leaves me wondering why Ignite didn’t “re-discover” the > node once the garbage collection finished. > > Ralph > >> On Jan 17, 2018, at 10:35 AM, ilya.kasnacheev <ilya.kasnach...@gmail.com> >> wrote: >> >> Hello! >> >> 2018-01-04 22:16:12 INFO [ ] IgniteKernal:128 - FreeList [name=null, >> buckets=256, dataPages=5, reusePages=0] >> 2018-01-04 22:16:52 INFO [ ] TcpDiscoverySpi:128 - Finished serving remote >> node connection [rmtAddr=/192.168.202.110:55327, rmtPort=55327 >> >> As you may notice, there's a 30 seconds of pause in node logs. It's possible >> that a long GC happened during that time and node failed to acknowledge >> discovery in time. >> >> Can you please collect GC log of such occurrence, post relvant fragments? >> >> Regards, >> >> >> >> -- >> Sent from: http://apache-ignite-users.70518.x6.nabble.com/ >> > > >
