|
Guys
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.
For a while now we thought it was an IRQ issue, for you just google "ksoftirqd
cpu", and people suggest to look at irqbalancer, setting affinity. Some have
suggested ensuring that your nic supports MSI-X (which ours do).
Googling "lvs 224.0.0.81" has led me on a weird path of reading a lengthy
thread on the LVS mailing list, that looks like has spanned two months and this
was in 2010, dealing with this ksoftirqd issue.
Long story short, and if I understood correctly,the /ksoftirqd/ issue could
actually be a kernel versioning issue. And it was introduced in 2.6.28 and
looks like it was only fixed in 2.6.37.rc1.
We run Debian Squeeze, which is on 2.6.32+29.
For you own viewing:
Start by reading all 'Kernel 2.6.35 and 100% S.I. CPU Time' in
http://lists.graemef.net/pipermail/lvs-users/2010-September/subject.html#start
Then move on to
http://lists.graemef.net/pipermail/lvs-users/2010-October/subject.html#start
I would like to ask, could it be a a kernel issue?
And is or has anyone else experienced problems with the current Debian kernel.
If you did experience problems, did using the backports kernel address this
issue?
Let me know what you think, if someone could assist, it would be appreciated.
Kind Regards
Brent Clark
_______________________________________________
Please read the documentation before posting - it's available at:
http://www.linuxvirtualserver.org/
LinuxVirtualServer.org mailing list - lvs-users@xxxxxxxxxxxxxxxxxxxxxx
Send requests to lvs-users-request@xxxxxxxxxxxxxxxxxxxxxx
or go to http://lists.graemef.net/mailman/listinfo/lvs-users
|
