Yep. Thanks! Aaron
On Sat, May 18, 2013 at 1:34 PM, Gagan Juneja <[email protected]>wrote: > Fine Thanks! > Can we mark this issue as fixed then? > > > Regards, > Gagan > > On Sat, May 18, 2013 at 10:57 PM, Aaron McCurry <[email protected]> > wrote: > > I think that setting the memory very small is fine, it's a tough setting > to > > test in a real test. The biggest issue this solves is the hung JVM that > is > > throwing OOM errors but the process is not exiting. This can cause an > > entire cluster to become unresponsive. To be honest since this is a > change > > to the settings on the JVM, I'm fine with it the way it's implemented and > > tested now. Because I know the JVM will fire the command on OOM error. > > > > Aaron > > > > > > On Sat, May 18, 2013 at 1:19 PM, Gagan Juneja <[email protected] > >wrote: > > > >> Aaron, > >> I have committed these changes. But one question how can we test this > >> scenario. Do you have any specific case or we should simulate this > >> issue by just setting memory value to be very small. > >> > >> Regards, > >> Gagan > >> > >> On Sat, May 18, 2013 at 9:54 PM, Aaron McCurry <[email protected]> > wrote: > >> > Yep I think that's its as simple as that. I didn't know %p would give > >> you > >> > the PID of the jvm. Cool, Thanks! > >> > > >> > Aaron > >> > > >> > > >> > On Sat, May 18, 2013 at 2:31 AM, Gagan Juneja < > [email protected] > >> >wrote: > >> > > >> >> I think -XX:OnOutOfMemoryError="kill -9 %p" will automatically > detects > >> >> the pid. Do we need to pass it explicitly or %p would be sufficient. > >> >> > >> >> > >> >> Regards, > >> >> Gagan > >> >> > >> >> On Wed, May 15, 2013 at 11:37 PM, Aaron McCurry (JIRA) < > [email protected] > >> > > >> >> wrote: > >> >> > > >> >> > [ > >> >> > >> > https://issues.apache.org/jira/browse/BLUR-85?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13658614#comment-13658614 > >> ] > >> >> > > >> >> > Aaron McCurry commented on BLUR-85: > >> >> > ----------------------------------- > >> >> > > >> >> > Yep, that sounds right. We should also add to the > >> >> BLUR_SHARD_JVM_OPTIONS. We should issue a kill -9 on the process and > >> you > >> >> can get the pid from the ./pids directory. Take a look at the > >> >> startup/shutdown scripts to get the exact file format. I think it's > >> >> "shard-0.pid" for the first shard process on the computer and > >> >> "controller-0.pid" for the first controller and so on for the 2nd and > >> third > >> >> processes (if any). Thanks! > >> >> > > >> >> > Aaron > >> >> > > >> >> >> Kill JVM on OutOfMemory Error > >> >> >> ----------------------------- > >> >> >> > >> >> >> Key: BLUR-85 > >> >> >> URL: > https://issues.apache.org/jira/browse/BLUR-85 > >> >> >> Project: Apache Blur > >> >> >> Issue Type: Improvement > >> >> >> Affects Versions: 0.1.5 > >> >> >> Reporter: Aaron McCurry > >> >> >> Fix For: 0.1.5 > >> >> >> > >> >> >> > >> >> >> Most of the time if a Blur process runs out of memory it will > likely > >> >> hang and never recover. So on an OOM error we should use the > >> >> -XX:OnOutOfMemoryError="<cmd args>; <cmd args>" JVM option to kill > the > >> >> process. > >> >> > > >> >> > -- > >> >> > This message is automatically generated by JIRA. > >> >> > If you think it was sent incorrectly, please contact your JIRA > >> >> administrators > >> >> > For more information on JIRA, see: > >> >> http://www.atlassian.com/software/jira > >> >> > >> >
