<div dir="ltr">It was my understanding that the replica 3 requirement is for glusterfs fuse storage, it's not clear that this would extend to NFS provided by a gluster volume. I'd appreciate clarification.<div><br></div><div>-Alastair</div><div><br></div></div><div class="gmail_extra"><br><div class="gmail_quote">On 5 November 2014 12:57, Daniel Helgenberger <span dir="ltr"><<a href="mailto:daniel.helgenberger@m-box.de" target="_blank">daniel.helgenberger@m-box.de</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><br>
<br>
On 05.11.2014 15:30, wodel youchi wrote:<br>
> Hi, I am new on oirt.<br>
Hello and welcome!<br>
<br>
><br>
> I want to know the best way to protect the storage of the hosted-engine?<br>
IHMO reliable hardware and contingency plans.<br>
<br>
><br>
> In Ovirt3.5, only NFS and iSCSI are supported for the engine VM, so this means<br>
> that the NFS server or the iSCSI volume become the weak link.<br>
First we need to define 'weak link'. IMOH this can be Network and / or<br>
storage hardware like controllers and spindles (SSDs). As the later<br>
tends to be reliable I think your mean indeed the data link layer as<br>
'weak link'?<br>
<br>
Gluster can be the same weak link for instance, as it needs a network<br>
layer. If you use iSCSI together with some storage appliance maybe with<br>
redundant controllers, this setup is quite reliable and engine storage<br>
is protected if you use iSCSI Multipath (and the paths are indeed<br>
separate hardware switches).<br>
<br>
You could call NFS a weak link; but even there are quite reliable setups<br>
available witch support fail over, replicated storage and IPMP.<br>
<br>
><br>
> I've read two articles, one using GlsuterFS+NFS and CTDB for High availability<br>
> of the engine storage,: oVirt 3.4, Glusterized<br>
I have to warn you at this point. This setup seems quite tempting; even<br>
using localhost addresses with gluster's build in NFS. This was tried<br>
before (myself included) but it is far from stable.<br>
<br>
You would at least need repica 3 gluster volumes to avoid split brains.<br>
These seem to happen quite often. I include Martin here, we talked about<br>
this at the ovirt workshop in Düsseldorf; maybe he can provide a better<br>
explanation.<br>
<br>
AFAIK gluster will be supported as engine storage in the future; but<br>
this is not the case right now. Of course, you are welcome to try!<br>
<br>
That said, and because you are new to ovirt, the main thing you need to<br>
protect is not the engine storage, but your production data domains.<br>
<br>
The VMs will run fine and continue to run with the engine down or not<br>
available. In case of a real disaster, you will be able to import these<br>
storage domains along with their VMs to a new engine.<br>
<br>
For me, I tend to have my engine backed up using engine-backup [1] and<br>
put the result to a different storage. From that data you can recreate<br>
the whole engine.<br>
<br>
> <<a href="http://community.redhat.com/blog/2014/05/ovirt-3-4-glusterized/" target="_blank">http://community.redhat.com/blog/2014/05/ovirt-3-4-glusterized/</a>>,<br>
><br>
><br>
> image <<a href="http://community.redhat.com/blog/2014/05/ovirt-3-4-glusterized/" target="_blank">http://community.redhat.com/blog/2014/05/ovirt-3-4-glusterized/</a>><br>
><br>
><br>
><br>
><br>
><br>
> oVirt 3.4, Glusterized<br>
> <<a href="http://community.redhat.com/blog/2014/05/ovirt-3-4-glusterized/" target="_blank">http://community.redhat.com/blog/2014/05/ovirt-3-4-glusterized/</a>><br>
> oVirt's Hosted Engine feature, introduced in the project's 3.4 release, enables<br>
> the open source virtualization system to host its own management server, which<br>
> means...<br>
> Afficher sur community.redha...<br>
> <<a href="http://community.redhat.com/blog/2014/05/ovirt-3-4-glusterized/" target="_blank">http://community.redhat.com/blog/2014/05/ovirt-3-4-glusterized/</a>><br>
><br>
> Aperçu par Yahoo<br>
><br>
><br>
> And another using GluserFS+NFS and KeepAlive: How to workaround through the maze<br>
> and reach the goal of the new amazing oVirt Hosted Engine with 3.4.0 Beta |<br>
> <<a href="http://www.andrewklau.com/ovirt-hosted-engine-with-3-4-0-nightly/" target="_blank">http://www.andrewklau.com/ovirt-hosted-engine-with-3-4-0-nightly/</a>><br>
><br>
><br>
><br>
><br>
> How to workaround through the maze and reach the goal of the new amazing oVirt<br>
> Hosted Engine with 3....<br>
> <<a href="http://www.andrewklau.com/ovirt-hosted-engine-with-3-4-0-nightly/" target="_blank">http://www.andrewklau.com/ovirt-hosted-engine-with-3-4-0-nightly/</a>><br>
> andrewklau - My Small World of Rants, Excitement, Snippets and Tuts How to<br>
> workaround through the maze and reach the goal of the new amazing oVirt Hosted<br>
> Engine with 3.4.0 Beta<br>
> Afficher sur <a href="http://www.andrewklau.com" target="_blank">www.andrewklau.com</a><br>
> <<a href="http://www.andrewklau.com/ovirt-hosted-engine-with-3-4-0-nightly/" target="_blank">http://www.andrewklau.com/ovirt-hosted-engine-with-3-4-0-nightly/</a>><br>
><br>
> Aperçu par Yahoo<br>
><br>
><br>
> Is there a better way to achieve this goal?<br>
><br>
> Thanks<br>
><br>
HTH<br>
<br>
[1] <a href="http://www.ovirt.org/Ovirt-engine-backup" target="_blank">http://www.ovirt.org/Ovirt-engine-backup</a><br>
<br>
--<br>
Daniel Helgenberger<br>
m box bewegtbild GmbH<br>
<br>
P: +49/30/2408781-22<br>
F: +49/30/2408781-10<br>
<br>
ACKERSTR. 19<br>
D-10115 BERLIN<br>
<br>
<br>
<a href="http://www.m-box.de" target="_blank">www.m-box.de</a> <a href="http://www.monkeymen.tv" target="_blank">www.monkeymen.tv</a><br>
<br>
Geschäftsführer: Martin Retschitzegger / Michaela Göllner<br>
Handeslregister: Amtsgericht Charlottenburg / HRB 112767<br>
_______________________________________________<br>
Users mailing list<br>
<a href="mailto:Users@ovirt.org">Users@ovirt.org</a><br>
<a href="http://lists.ovirt.org/mailman/listinfo/users" target="_blank">http://lists.ovirt.org/mailman/listinfo/users</a><br>
</blockquote></div><br></div>