[Linux-cluster] virtual machine failover with gfs

[Joe Royall]

2008/6/25 matt whiteley <whiteley at pdx.edu>:

> I have spent lots of hours trying different setups and reading the
> documentation already so I hope this isn't a faq as I am new to the list.
>
> I read the Red Hat Magazine article on this topic[1], but have come to
> realize that it might not be exactly what I am going for. I want to have a
> group of nodes that run a group of virtual machines with automated failover.
> I set things up how the article described but realized I didn't want the gfs
> mount in the fstab file. I would like the gfs mount described in the
> cluster.conf file so that as nodes are added or removed the mount will
> follow the changes (I know about the 1 journal per node so have created a
> few extra already). When I add a service to mount the gfs resource, it only
> gets mounted on one node as is to be expected thinking in terms of other
> resources.
>
> I started thinking about this and it almost seems like gfs is unnecessary.
> Should I have a file system per virtual machine that wouldn't need to be gfs
> since only one node will ever run a virtual machine at a time? Then
> mount/umount the file system as the virtual machine was migrated in the
> cluster?
>
> It seems like I am missing something about how this should be setup and I
> would really appreciate any tips or ideas. I will include my cluster.conf in
> case it provides any more info.
>
> As a side note, what is with all the errors from system-config-kickstart
> telling me my config file is invalid if it was generated by conga. Both
> versions are updated to the newest available.
>

Why not use lvm backed vms, 1 per vm, share the entire partition with all
the lvms via ISCSI to each dom0 and run clvm on the dom0s.  The lvms do not
need to be mounted in dom0.  You can then use RHCS to failover vms between
dom0s.  Consider putting all the vms on a single node into a single resource
group and only allow 1 group to operate on a single node.  You can then
configure N+1 redundancy.

>
>
>
>
>
> [1]
> http://www.redhatmagazine.com/2007/08/23/automated-failover-and-recovery-of-virtualized-guests-in-advanced-platform/
>
> thanks,
> --
> matt whiteley <whiteley at pdx.edu>
>
>
>
>
> --
> Linux-cluster mailing list
> Linux-cluster at redhat.com
> https://www.redhat.com/mailman/listinfo/linux-cluster
>



-- 
Joe Royall
Red Hat Certified Architect
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/linux-cluster/attachments/20080625/0fa34b20/attachment.htm>