I would like to mention that invoking * HConnectionManager.deleteAllConnections(true);* after each batch run has fixed the connection leak issue. Also, this invocation is not interfering with the connections of the other Zookeeper clients from other JVMs. So its the quick fix as of now.
But I am not sure if there is any scenario where this might cause any issues if I close the shared connections using * HConnectionManager.deleteAllConnections(true);* . NY On Sun, May 13, 2012 at 11:15 AM, Narendra yadala <narendra.yad...@gmail.com > wrote: > > Hi Stack > > Thanks for the response. The HBase version we are using is* Version > 0.90.4-cdh3u3* > > I do close the tables and scanners cautiously after doing the crud > operations. Though this does not happen (finally clause does not execute) > when I terminate/kill the batches or stop the servlet containers. Please > advise. > > Regards > NY > > > On Sun, May 13, 2012 at 3:26 AM, Stack <st...@duboce.net> wrote: > >> On Fri, May 11, 2012 at 10:45 PM, Narendra yadala >> <narendra.yad...@gmail.com> wrote: >> > Hi Dave >> > >> > I reuse HBase configuration object as much as possible. I make it >> static so >> > that there is one config per JVM. But the problem is stopping and >> starting >> > the batches or the tomcat container which is where I keep getting the >> > issue. The issue is I run in to this issue with HBase shell also to make >> > things worse. I increased the zookeeper max client connections to 200 >> but >> > that will not be of much use because eventually those will also be used >> up. >> > >> >> What version of hbase? >> >> Look back in this mailing list for threads on our using too many zk >> connections later fixed. See if they are similar to your scenario. >> >> St.Ack >> > >
