| strace vzctl start VEID - not stopping [message #40541] |
Mon, 30 August 2010 18:44 |
bvirtual
Messages: 1
Registered: August 2010
|
Junior Member |
|
|
"kill -9 pid" fails to stop strace of vzctl. Also, we have 5 VE's that start init, and go no further. It might be related.
First the command as seen from ps, then version info, then extra info. Notice it's using 100% CPU, which is 1 of 8 CPUs, on a dual quad with 32gig RAM, 1 terabyte HD.
# ps aux | grep strace
root 29250 99.5 0.0 4116 776 ? R Aug22 11003:47 strace -f vzctl start 99
strace -- version 4.5.18
vzctl version 3.0.24.1
CentOs release 5.5 (Final)
Linux version 2.6.18-194.8.1.el5.028stab070.2 (root@rhel5-build-x64) (gcc version 4.1.2 20080704 (Red Hat 4.1.2-46)) #1 SMP Tue Jul 6 14:55:39 MSD 2010
I've been able to kill -9 other strace processes, but this one, run by another root user, fails to stop. The way I killed strace was to kill off the child vzctl first, or the parent bash shell, and then kill -9 strace.
I've attached strace (strace -p 29250) and sent a KILL signal, and got nothing. Why is strace using 100% CPU? What loop is in it? I was going to read strace source code, to find out, but that will not help me stop this process.
Any hints? Suggestions? Other signals to send? Ok, I could reboot, but there are dozens of containers running.
I've googled and read a lot, apparently strace might be stopping, but starting again. Most strace vzctl posts go back 2 years, which is about how old our OpenVZ install is.
This came about as 'sync' failed to returned. And was state 'uninterruptible.' 'shutdown' failed to complete, leaving 'init' running, along with the two syncs. None could be stopped from the Host OS with kill -9. Other VE's showed the same problem. So reboot the Host OS for another reason, related to changing mount commit values.
Several containers would start init, but go no further, but could be killed. So strace vzctl. The output file showed vzctl *.sh running, but not all of them, and never showed init starting, but ps did.
So, we now have many containers running and one CPU out of action. Killing this one strace process would give us the CPU back. It's been voiced the uneven heating of the quad chip will eventually crack it, and we will lose 4 CPUs. Is that possible? If so, then rebooting is needed. How soon? It's been 6 days now.
Am I posting in the right place? Might there be an strace forum? I'm just starting here.
|
|
|
|
OpenVZ Forum
Members
Search
Help
Register
Login
Home
