1

I've recently encountered a very annoying effect: when a process on Xen DomU apparently consumes all available RAM/swap (that's my assumption; the root of the problem might lie elsewhere), the DomU locks up, becomes totally unresponsive, and the only way to bring it back online is to login to Dom0 and destroy/recreate the VM. A typical part of the console log of DomU follows:

[88751.207692] INFO: task cron:1318 blocked for more than 120 seconds.
[88751.207717] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[88875.710604] Out of memory: kill process 1315 (apache2) score 51765 or a child
[88875.710655] Killed process 1315 (apache2)

Dom0 is Debian Lenny; DomU is Ubuntu Jaunty. Xen version is 3.1.

  • What may cause such behavior?
  • What tests should I make to help figure it out?
  • What configuration options (such as RAM or VCPU distribution between the Dom0 and the DomU) can possibly solve the problem?

1 Answer 1

3
  • What may cause such behavior? Software that for whatever reason (bugs, misconfiguration, imprecisely predicted load) chews more memory than you had predicted it would.
  • What tests should I make to help figure it out? System profiling, such as sar (sysstat) can help you identify which process is consuming all the RAM, if you don't know that already.
  • What configuration options (such as RAM or VCPU distribution between the Dom0 and the DomU) can possibly solve the problem? Give the domU more RAM, or -- better still -- install a healthy chunk of swap.
1
  • Thanks for the advice. I think I'll be able to handle it from here.
    – dpq
    Oct 18, 2009 at 9:01

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .