Failover Issue

[Lon Hohberger]

On Mon, 2004-11-22 at 07:44 -0800, Pramod wrote:
> Hi, 
>        I have two servers running similar
> configuration,clustered using piranha in FOS type.all
> services required for running  the cluster is enabled
> on both servers.Configured One as Primary Node & Other
> one as a Backup node,a virtual server is created with
> a different IP. when i unplug the cable from the
> active node a failover occurs to the backup node,but
> when i plug it back in to the active node later it
> takes long time approximately 15mins to fall back,and
> that period of time i cannot ping the virtual IP also.
> If anyone has come across such issue please respond.

I think this is related:

http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=134007

Basically, the master won't re-arp the VIP after this scenario.  There
are patches in the bugzilla which fix it:

Mr. Bonnet's patch from a long time ago:
http://bugzilla.redhat.com/bugzilla/attachment.cgi?id=104736&action=view

Mine, which works around a race condition WRT heartbeats after plugging
the master back in:
http://bugzilla.redhat.com/bugzilla/attachment.cgi?id=106303&action=view

-- Lon