It sounds like you probably need to raise the default 15 sec zk client timeout. We have it default to a fairly aggressive setting. For high load envs, you may have to bring it to 30 or 45 seconds.
Also, be sure you are not using a stop the world gc collector - CMS (the concurrent low pause collector) is best. - Mark On May 5, 2013, at 1:50 PM, Michael Della Bitta <michael.della.bi...@appinions.com> wrote: > Hi, > > I'm running a Solr 4.2.1 cloud with an external three-node Zookeeper 2.4.5 > setup. > > I'm seeing a lot of these errors in the zookeeper logs: > > 2013-05-05 15:06:22,863 - WARN [NIOServerCxn.Factory: > 0.0.0.0/0.0.0.0:2181:NIOServerCnxn@349] - caught end of stream exception > > Also some of these: > > NIOServerCxn.Factory:0.0.0.0/0.0.0.0:2181:NIOServerCnxn@354] - Exception > causing close of session 0x3e22c637f5063f due to java.io.IOException: > Connection reset by peer > > or > > NIOServerCxn.Factory:0.0.0.0/0.0.0.0:2181:NIOServerCnxn@354] - Exception > causing close of session 0x3e22c637f506a1 due to java.io.IOException: > Connection timed out > > We've had problems with nodes dropping out of collections during indexing. > I'm assuming these are related? Is there some sort of socket tuning I need > to do on the Solr side to keep these connections going? > > Thanks for any input anybody might be able to provide, > > Michael Della Bitta > > ------------------------------------------------ > Appinions > 18 East 41st Street, 2nd Floor > New York, NY 10017-6271 > > www.appinions.com > > Where Influence Isn’t a Game