[Linux-cluster] How do i stop VM's on a failed node?

[Volker Dormeyer]

Hi,

On Fri, Jul 16, 2010 at 01:21:57PM -0400,
Nathan Lager <lagern at lafayette.edu> wrote:
> I have a 4-node cluster, running KVM, and a host of VM's.
> 
> One of the nodes failed unexpectedly.  I'm having two issues.  First,
> the VM's which were on that node are still reported as up and running on
> that node in clustat, and second, i'm unable to interact with the
> cluster using clusvcadm.  Every command hangs.  I've tried disabling
> these VM's, and the commands hang.
> 
> How can i clear out this dead node, without having to forceably restart
> my entire cluster?

To me, it sounds like fencing is not configured, properly.

What has been logged on the remaining nodes? I would assume, they tried to
fence the failed node - but are not able to.

Log-Snippets and Config would be helpful.

Regards,
Volker