High CPU load

[ichosting]

At the moment we suffer with a node with high cpu load.

Cpu(s): 89.7%us load average: 3.90, 4.87, 5.56

I tested everything (as far as i know) but no luck with pinpointing the problem.

On the node there is just one vm which needs all the resources.
It is a quadcore Dell machine. VM is a opvenvz container.

:~# 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


Any idea where to begin finding the solution? I allready searched on forums and google. But no luck so far. So last resort is to get help with starting new thread.

 

[udo]

Hi,
looks like that your load is io-related not cpu-related?! But you don't support much infos...

Udo

 

[ichosting]

Well the cpu load is not dropping. On node and vm.

There is not much going on on the vm. Not much websites, so i'm not sure why it is io related.

What kind of information do you want me to provide?

 

[udo]

Well the cpu load is not dropping. On node and vm.

There is not much going on on the vm. Not much websites, so i'm not sure why it is io related.

What kind of information do you want me to provide?

Hi,
load shows how many processes are ready to use the cpu - if you have equal or more cores than load there is no realy problem. If you processes need more cpu-power the load rises (but you show only 89%).
If you io-system are not fast enough the load rises without (much) rising of cpu-uses - because the processes must wait for the io and don't use the cpu.
Look at the wait-output and perhaps look with iostat (apt-get install sysstat).

BTW. does the same happens with an actual system?

Udo

 

[ichosting]

Well, the wait output does not show very much. It is mostly 0.

Cpu(s): 71.4%us, 2.3%sy, 0.0%ni, 26.3%id, 0.0%wa, 0.0%hi, 0.0%si, 0.0%st

Can it be that the CPU UNITS setting for the VM (now default) is too low?
Normally OPENVZ would ignore that, right?


Some output from iostat:

Time: 07:15:44
avg-cpu: %user %nice %system %iowait %steal %idle
66.67 0.00 1.33 0.00 0.00 32.00


Time: 07:15:49
avg-cpu: %user %nice %system %iowait %steal %idle
64.53 0.00 0.89 0.00 0.00 34.59


Time: 07:15:54
avg-cpu: %user %nice %system %iowait %steal %idle
65.32 0.00 1.13 0.00 0.00 33.55


Time: 07:15:59
avg-cpu: %user %nice %system %iowait %steal %idle
66.55 0.00 1.77 91973858.03 0.00 0.00


Time: 07:16:04
avg-cpu: %user %nice %system %iowait %steal %idle
69.94 0.00 4.46 101444706.67 0.00 74218603.07


Time: 07:16:09
avg-cpu: %user %nice %system %iowait %steal %idle
67.33 0.00 2.38 0.00 0.00 30.30


Time: 07:16:14
avg-cpu: %user %nice %system %iowait %steal %idle
68.27 0.00 2.35 0.00 0.00 29.38


Time: 07:16:19
avg-cpu: %user %nice %system %iowait %steal %idle
65.02 0.00 2.20 0.00 0.00 32.77


Time: 07:16:24
avg-cpu: %user %nice %system %iowait %steal %idle
66.94 0.00 1.22 0.00 0.00 31.84

We have kind of same setups on "normal" dedicated machines which have not the same problem. Also the same type of machines.