Subject: Re: [vserver] Re: vserver heartbeat
From: Martin Fick <mogulguy@yahoo.com>
Date: Fri, 26 Sep 2008 07:12:04 -0700 (PDT)

--- On Fri, 9/26/08, randall <randall@songshu.org> wrote:

> still having problems with my heartbeat though, and i must
> admit i find it hard to grasp the logic.
> Using a R1 style config i can get the failover of the disks
> to happen and filesystems mounted but the Vserver 
> start script does not seem to 
> happen, so i would have to vserver start manually, already
> very happy i got this far.

Unfortunately, I never tried R1 style scripts, but 
I believe that they will not be any good for what
you want because I believe that there is not way
to add an instance parameter to them.  In other 
words I think that it is impossible to fail over
more than one drbd resource independently with them.

From your website it seems like you would like both
nodes to act as primaries to different resources
during normal operation, I think that you need R2
ocf agents for this.

 
> The CRM style config seems to be preffered but 
> i'm just getting a split-brain trying to 
> understand the .xml thingie. 

Yes, they can be rather complicated, the tools
are lacking, I ended up developing my own, you may
want to check out:
 http://www.theficks.name/Hacks/Heartbeat-ResourceCIBS

> So the question is do people have more Vserver specific
> examples besides http://linux-vserver.org/Fail-over, so 
> i can do a little comparing to get the logic of the whole thing?

I am not sure that I have better examples, but I will
try to help you work through things if you choose to
switch to R2 style configs,

-Martin