<div dir="auto">Have you considered a HA NFS over a 2-node DRBD8 cluster ? Should work well on most hypervisors (qcow2,raw,vmdk based).<div dir="auto"><br></div><div dir="auto">Yannis</div></div><div class="gmail_extra"><br><div class="gmail_quote">On 26 Aug 2017 12:16, "A.Rubio" <<a href="mailto:aurusa@etsii.upv.es">aurusa@etsii.upv.es</a>> wrote:<br type="attribution"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hi all.<br>
<br>
I writted a book about this every years ago for a course.<br>
<br>
The book is about pacemaker, drbd, gfs2 and kvm, for 2 nodes and for many nodes.<br>
<br>
It's writted in spanish because my english is poor (it's evident :-()<br>
<br>
<a href="http://www.lulu.com/shop/aurelio-rubio-sapi%C3%B1a-and-juan-vicente-capella-hern%C3%A1ndez/pacemaker-clusters-de-alta-disponibilidad-para-servidores-virtualizados/ebook/product-22274638.html" rel="noreferrer" target="_blank">http://www.lulu.com/shop/aurel<wbr>io-rubio-sapi%C3%B1a-and-juan-<wbr>vicente-capella-hern%C3%A1ndez<wbr>/pacemaker-clusters-de-alta-<wbr>disponibilidad-para-servidores<wbr>-virtualizados/ebook/product-<wbr>22274638.html</a><br>
<br>
If yo want a copy and you can't buy it ;-), You write to my. I don't have problem send you the ebook.<br>
<br>
<br>
El 25/08/2017 a las 22:18, Digimer escribió:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
On 2017-08-25 04:08 PM, Gionatan Danti wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Il 25-08-2017 22:01 Digimer ha scritto:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
On 2017-08-25 03:37 PM, Gionatan Danti wrote:<br>
<br>
The overhead of clustered locking is likely such that your VM<br>
performance would not be good, I think.<br>
</blockquote>
Mmm... I need to do some more testing with fio, it seems ;)<br>
<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
With raw clustered LVs backing the servers, you don't need cluster<br>
locking on a per-IO basis, only on LV create/change/delete. Because LVM<br>
is sitting on top of DRBD (in dual-primary), live-migration is no<br>
trouble at all and performance is good, too.<br>
</blockquote>
True.<br>
<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
GFS2, being a cluster FS, will work fine if a node is lost, provided it<br>
is fenced succesfully. It's wouldn't be much of a cluster-FS<br>
otherwise. :)<br>
</blockquote>
So no problem with quorum? A loss of a system in a two-node cluster<br>
seems to wreack havok on other cluster filesystems (Gluster, for<br>
example...)<br>
<br>
Thanks.<br>
</blockquote>
Quorum is optional (an often misunderstood thing).<br>
<br>
<a href="https://www.alteeve.com/w/The_2-Node_Myth" rel="noreferrer" target="_blank">https://www.alteeve.com/w/The_<wbr>2-Node_Myth</a><br>
<br>
We've run without quorum for every system we've built over 5+ years,<br>
across dozens of sites and never once needed it. A proper fence setup,<br>
which is needed regardless, is fine. In our opinion, the complexity of a<br>
third quorum node is not justified for the limited benefit of quorum.<br>
Simplicity is simply too valuable in HA.<br>
<br>
</blockquote>
<br>
______________________________<wbr>_________________<br>
drbd-user mailing list<br>
<a href="mailto:drbd-user@lists.linbit.com" target="_blank">drbd-user@lists.linbit.com</a><br>
<a href="http://lists.linbit.com/mailman/listinfo/drbd-user" rel="noreferrer" target="_blank">http://lists.linbit.com/mailma<wbr>n/listinfo/drbd-user</a><br>
</blockquote></div></div>