[ClusterLabs] Antw: VirtualDomain as resources and OCFS2

[Ken Gaillot]

On Tue, 2018-09-11 at 15:21 +0200, Lentes, Bernd wrote:
> ----- On Sep 11, 2018, at 8:54 AM, Ulrich Windl Ulrich.Windl at rz.uni-r
> egensburg.de wrote:
> 
> > 
> > Hi Bernd,
> > 
> > the disappointing answer ist this: With cLVM you cannot make
> > snapshots of the
> > LVs (easily), and in SLES11 SP4 at least the tool to make snapshots
> > of OCFS2
> > also isn't provided. So the major advantage of having separate LVs
> > for each VM
> > is actually gone, making a desaster recovery strategy somewhat
> > difficult. We do
> > backups of the files "inside" the VM (just like a physical
> > machine), but once
> > your VM images are gone, desaster recovery is a bit tricky,
> > however.
> > 
> > Maybe it is most efficient to have a SAN-provided "physical" disk
> > for each VM.
> > The if the SAN can make snapshots of those disks, you have an
> > elegant backup
> > and recovery solution. We had "kind of that" (using iSCSI-provided
> > disks) about
> > 15 years ago in SLES10, but the solution was unable to allow loive
> > migration (a
> > different and longer story)...
> > 
> > Regards,
> > Ulrich
> 
> Hi Ulrich,
> 
> my first idea to gave dedicated logical volumes for each guest was
> just that it looks more tidy.
> But as mentioned already i realize that my cluster configuration will
> grow and grow 
> with each further guest and becomes more and more overseeable.
> 
> I don't want to create snapshots with lvm, i'm thinking of
> snapshotting my guests with libvirt
> and virsh. I think with that i have a resonable solution. I will
> snapshot the image files so DR
> should be fine.
> Do you see any argument which really contradicts to the use of just
> one logical volume formatted with OCFS2 
> for all guests ?

I don't have enough experience in the area to give a strong answer, but
a couple of comments:

Pacemaker does have resource templates to simplify repetitive
configuration. I believe only primitives can be templated, but that
saves you from having to repeat common parameters, operations, etc.

The only advantage I can think of for separate LVs is that they can be
recovered individually without affecting other VMs. How big an
advantage that is depends on how likely it is to have a failure of one
VM's storage but not all.

> I'm using SLES 12 SP3, SLES 11 SP4 is EOL in a few months.
> 
> Bernd
>  
> 
> Helmholtz Zentrum Muenchen
> Deutsches Forschungszentrum fuer Gesundheit und Umwelt (GmbH)
> Ingolstaedter Landstr. 1
> 85764 Neuherberg
> www.helmholtz-muenchen.de
> Aufsichtsratsvorsitzende: MinDir'in Baerbel Brumme-Bothe
> Geschaeftsfuehrer: Prof. Dr. med. Dr. h.c. Matthias H. Tschoep,
> Heinrich Bassler, Dr. rer. nat. Alfons Enhsen
> Registergericht: Amtsgericht Muenchen HRB 6466
> USt-IdNr: DE 129521671
> 
> 
> _______________________________________________
> Users mailing list: Users at clusterlabs.org
> https://lists.clusterlabs.org/mailman/listinfo/users
> 
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.
> pdf
> Bugs: http://bugs.clusterlabs.org
-- 
Ken Gaillot <kgaillot at redhat.com>