VM cannot be stopped

[wolfon]

We use Proxmox VE 1.7

One openvz Container (Ubuntu 10.04) ist not responding, but consuming 100% CPU on one core.
It is neither possible to stop or enter the VM nor to kill the (postrgres) processes causing the high system load.

How can I stop the VM or kill the processes without rebooting the host.
As there are other important servers running in parallel this would be my last option.

Thanks in advance for any suggestions.

 

[ressel]

have you tried kill -9 pid?

 

[wolfon]

ig-proxmox:~# vzlist -o veid,laverage -s laverage
CTID LAVERAGE
175 0.00/0.00/0.00
110 27.99/27.97/27.91
109 0.00/0.00/0.00
108 0.00/0.00/0.00
105 0.08/0.03/0.01
104 0.00/0.00/0.00
102 0.00/0.00/0.00

ig-proxmox:~# vzctl enter 110
enter into CT 110 failed

ig-proxmox:~# vzctl restart 110
Restarting container
Stopping container ...
Unable to stop container: operation timed out

ig-proxmox:~# ps -aux | grep postgres
Warning: bad ps syntax, perhaps a bogus '-'? See http://procps.sf.net/faq.html
ntp 2111 0.0 0.0 79924 5196 ? D May12 0:00 /usr/lib/postgresql/8.2/bin/postgres -D /var/lib/postgresql/8.2/main -c config_file=/etc/postgresql/8.2/main/postgresql.conf
ntp 2113 0.0 0.0 80064 3232 ? Ds May12 0:01 postgres: writer process
ntp 2114 0.0 0.0 50676 1464 ? Ds May12 0:01 postgres: stats collector process
ntp 2149 0.0 0.4 97516 33308 ? Ds May12 0:01 postgres: projop projop [local] idle
ntp 2150 91.4 0.2 93880 23128 ? Rs May12 6247:00 postgres: projop projop [local] idle
ntp 2151 0.0 0.1 85752 10992 ? Ds May12 0:00 postgres: projop projop [local] idle
ntp 2654 0.0 0.2 97120 22624 ? Ds May12 0:01 postgres: projop projop [local] idle
ntp 4224 0.0 0.1 84956 9692 ? Ds May12 0:00 postgres: projop projop [local] idle
ntp 9221 0.0 0.1 85084 10604 ? Ds May12 0:00 postgres: projop projop [local] idle
ntp 16262 0.0 0.1 84832 9616 ? Ds May12 0:00 postgres: projop projop [local] idle
root 31855 0.0 0.0 5168 812 pts/0 S+ 10:55 0:00 grep postgres

ig-proxmox:~# kill -9 2111
ig-proxmox:~# kill -9 2113
ig-proxmox:~# kill -9 2114
ig-proxmox:~# kill -9 2149
ig-proxmox:~# kill -9 2150
ig-proxmox:~# kill -9 2151
ig-proxmox:~# kill -9 2654
ig-proxmox:~# kill -9 4224
ig-proxmox:~# kill -9 9221
ig-proxmox:~# kill -9 16262

no effect, all keep running

 

[dlegg]

I'm confused. You say that you are using KVM and can't log into the guest, but you're trying to kill Postgres processes above and you're using an OpenVZ container.

 

[tom]

on your first post you talk about KVM, now you show a OpenVZ container - so about what do we talk here really?

 

[wolfon]

Oops, its of course an openvz Container.

 

[tom]

if a kill -9 does not help (try from the host)?

in any case upgrade to the latest version and think of using the 2.6.18 branch, stable OpenVZ.
See http://pve.proxmox.com/wiki/Proxmox_VE_Kernel

 

[wolfon]

This is the current version:
ig-proxmox:~# uname -a
Linux ig-proxmox 2.6.32-4-pve #1 SMP Wed Dec 15 14:04:31 CET 2010 x86_64 GNU/Linux

ig-proxmox:~# pveversion -v
pve-manager: 1.7-11 (pve-manager/1.7/5470)
running kernel: 2.6.32-4-pve
proxmox-ve-2.6.32: 1.7-30
pve-kernel-2.6.32-4-pve: 2.6.32-30
qemu-server: 1.1-28
pve-firmware: 1.0-10
libpve-storage-perl: 1.0-16
vncterm: 0.9-2
vzctl: 3.0.24-1pve4
vzdump: 1.2-10
vzprocps: 2.0.11-1dso2
vzquota: 3.0.11-1
pve-qemu-kvm: 0.13.0-3
ksm-control-daemon: 1.0-4

I dont like the idea to upgrade now, but if this is the only option, I need to replan my weekend

The process is really resistant against killing - a real Zombie.
Knew this before only from M$ Systems.

 

[tom]

a second option: we can ssh to your server and take a look. (but you need a support ticket or a server subscription in this case)