[lvs-users] ksoftirqd/0 runs at a 100% and looks like it could a kernel issue.

Brent Clark brentgclarklist at gmail.com
Thu May 31 12:07:55 BST 2012


On 31/05/2012 12:43, Graeme Fowler wrote:
> On Thu, 2012-05-31 at 12:21 +0200, Brent Clark wrote:
>> Im writing to you because we have a two node cluster setup with just apache2, heartbeart and LVS, and every now and then find that ksoftirqd/0 runs at a 100%. We find that the only way to fix it, is to actually reboot the server.
> <snip>
>> Let me know what you think, if someone could assist, it would be appreciated.
>
> I would hazard a guess that at the time you have this issue you have
> packets being "reflected" between the IPVS frameworks on both hosts.
>

Hi Graeme

Thank you so much for replying.

Sorry I should have stated upfront that, our setup is not active/active.

Heartbeat manages and stops and starts ldirectord.

Just to show you.
bclark at host-web01:~$ cat /etc/ha.d/haresources
hostcommentedout        \
        ldirectord::ldirectord.cf \
        LVSSyncDaemonSwap::master \
        IPaddr2::1.2.3.4/27/eth0/1.2.3.0


So the idea is. Heartbeat manages the Virtual IP (vip), and the active server goes down for what ever reason (be it crash, or reboot), on the passive server, the heartbeat daemons sees that its pair is dead, and then brings up the vip.
Thats just managing the "vip" side of things. Then on top of this, we use LVS to load balance all HTTP, between both servers to share the HTTP traffic. But only one instance of LVS runs at a time on the active server.
Its actually a very busy server. Easily in one of the top twenty sites here in South Africa.
We have actually found that this problem has arised since we upgrade to Debian Squeeze. Debian Lenny, was no problems.
The only slight configuration change was the introduction of adding a third webserver (no heartbreat of LVS runs on this server), to the setup. Else the configurations have not been chanced in years.

Thanks
Brent





More information about the lvs-users mailing list