Hi Mark; I have same situation with my Solr 4.2.1 nodes. These are from my startup parameters:
-XX:+UseConcMarkSweepGC -XX:CMSInitiatingOccupancyFraction=75 -XX:NewRatio=3 -XX:MaxTenuringThreshold=8 -XX:+CMSParallelRemarkEnabled -XX:+ParallelRefProcEnabled -XX:+UseLargePages -XX:+AggressiveOpts My nodes are at Amazon here is a part from log messages: INFO: A cluster state change: WatchedEvent state:SyncConnected type:NodeDataChanged path:/clusterstate.json, has occurred - updating... (live nodes size: 10) Do you have an idea about it? 2013/4/18 Mark Miller <markrmil...@gmail.com> > > On Apr 18, 2013, at 8:40 AM, jmozah <jmo...@gmail.com> wrote: > > > > > > > On 16-Apr-2013, at 11:16 PM, Mark Miller <markrmil...@gmail.com> wrote: > > > >> Are you using a the concurrent low pause garbage collector or perhaps > G1? > > > > > > I use the default one which comes in jdk 1.7. > > It varies by platform, but 99% that means you are using the throughput > collector and you should try the CMS collector instead. > > - Mark > > > > >> > >> Are you able to use something like visualvm to pinpoint what the > bottleneck might be? > > > > Unfortunately.. it is prod machine and i could not replicate it locally. > > > >> > >> Otherwise, keep raising the timeout. > > > > > > Thats what i did now.. will see if it comes in the next run.. > > > > ./zahoor > > > >
