[zfs-discuss] ZFS, DRBD, GFS and Clustering

Micky mickylmartin at gmail.com
Wed Oct 16 05:51:28 EDT 2013


On Wed, Oct 16, 2013 at 2:39 PM, Micky <mickylmartin at gmail.com> wrote:
>>GFS needs to be run on
>> shared storage (or DRBD at a push), and thus required RHCS to work.
Right. I understood what you meant. RHCS uses CLVM as cluster aware
filesystem in the examples.


> Yes, but in case of VM/hypervisor clustering, you only need GFS for
> sharing provisioning storage. Rest of the VMs can actually run off
> DRDB, as far as I see from the RHCS documentation since rmanager/cman
> can care of fencing resources and starting VMs after a node failure.
>
>> Fine of you are running in active-passive mode with only one side mounted at
>> any time. If you plan to have them both mounted, you will destroy the file
>> system instantly - ZFS is not clustering aware.
>
> Ok, thanks for the heads up. How about only using a zvol block device
> for DRBD and then use rest of the cluster stack on top it? i.e. using
> a pool only for the RAID purposes?

Let me know what do you think about the above though?

To unsubscribe from this group and stop receiving emails from it, send an email to zfs-discuss+unsubscribe at zfsonlinux.org.



More information about the zfs-discuss mailing list