drax68
drax68
Same problem here, SMILEDecodeError with crippled json in stacktrace. @nvitucci have you found solution for that? @spfjr observing same issue with your fork as well.
By hanging I mean that whole kafka-web-console become unresponsible (no webui at all, continous load) and can be temporary fixed with restart. Now I switched back to 1.1.0 - it...
https://gist.github.com/drax68/d36cf1c35fc84ce645a3
Possibly related https://issues.apache.org/jira/browse/ZOOKEEPER-1506 if generic zookeeper client used.
Let me create a new histogram. Regarding changes - nothing else was changed, except of nginx layer in front of atlas instances. We even moved all automated queries to a...
Adding fresh histogram for 100Gb heap [atlas_hist.txt](https://github.com/Netflix/atlas/files/1478735/atlas_hist.txt)
-histo:live wasn't able to connect to jvm process so I've used -F -histo.
@brharrington seems any non-default value for akka's max-connections causes memory leak after short uptime. Can you confirm that 100 requests/second is not problematic for atlas /publish endpoint?
> Interesting. Did you change that setting when you started using nginx? Property was changed more than year ago, atlas was working fine all this time.
Latest 1.6.0 rc looks good in terms of memory consumption, but we observe various timeouts on publishing/quering. Is this something known to you? Tried it with akka.http.host-connection-pool.pool-implementation=new akka.http.server.linger-timeout=infinite Almost 50%...