Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.
  • Host OS: OpenVZ host
  • Guest OS: Debian 6.0 minimal
  • Java: java6 sun
  • Servlet container: Winstone
  • Jenkins for debian: http://pkg.jenkins-ci.org/debian/
  • Jenkins version: 1.449
  • uname -a: Linux hostname 2.6.18-238.9.1.el5.028stab089.1 #1 SMP Thu Apr 14 14:06:01 MSD 2011 i686 GNU/Linux
  • JAVA_ARGS="-Xmx128m -Xms64m -Xrs -XX:HeapDumpPath=/tmp/java-jenkins-dumps -XX:PermSize=128m"
  • free -m:
              total       used       free     shared    buffers     cached
Mem:          2048        532       1515          0          0          0
-/+ buffers/cache:        532       1515
Swap:            0          0          0

After crash:

  • jenkins latests line output: INFO: Jenkins is fully up and running
  • syslog output: Jan 31 17:50:02 cherry jenkins: jenkins: fatal: client (pid 16189) killed by signal 15, exiting
  • user_beancounters has 0 failcnt

The question: what else to debug? Is it worth trying to change servlet container (to tomcat)?

share|improve this question

1 Answer 1

I just fixed a similar issue on one our servers. For some reason the VM we have running Jenkins was using a small memory configuration and the swap partition wasn't currently used. My guess is that at some point Java attempts to allocate a large amount of memory and is killed by the OOM task.

After making sure that the swap partition was activated via a fstab entry the problem went away.

TLDR: you need more virtual or physical memory.

share|improve this answer
    
Did you notice anything in logs? This results in no output for me: egrep -r -i "killed process" /var/log/ –  iElectric Feb 8 '12 at 16:07
    
yeah there were messages in the the log that jenkins had been killed. However - that is likely linux specific. –  cshimer Feb 13 '12 at 19:05
    
I have no such log entry present, that's why I'm doubting. –  iElectric Feb 14 '12 at 19:45
    
You're probably forking a load of processes so is a good candidate for Linux OOM killer to kick in. You won't see much logging from the JDK as it's being force killed. /var/log/messages should mention OOM. –  Alastair McCormack Nov 30 '12 at 19:13
    
Also, you really should allocate some swap to your box. It may be enough for Linux to page off enough to run everything you need. –  Alastair McCormack Nov 30 '12 at 19:15

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.