<html>
<head>
<meta content="text/html; charset=UTF-8" http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<div class="moz-cite-prefix">Hello,<br>
<br>
We still have three clusters running the old v1.9 with OCFS2
(although it was never recommended by Proxmox):<br>
<ul>
<li>One small two-nodes with DRBD Primary/Primary and OCFS2on
top. Never had any problems.</li>
<li>One mid four-nodes with iSCSI(openfiler) and OCFS on top.
Also on the same cluster we have another storage via iSCSI
with LVM on top. Both are stable, (i think LVM is a bit more)
but OCFS has been more flexible (qcow2, file system access,
etc).</li>
<li>One mid three-nodes with FC and OCFS. Only once time it was
necessary to do fsck (after 2 years), without loose any data.<br>
</li>
</ul>
If neccesary, I would use DRBD with LVM on top (as doc says).<br>
It was a pity that there is no more OCFS2 support since v2.0, but
as Eneko says, the good news is that we will have new storage
options on 3.1.<br>
<br>
Thank you Proxmox team!!<br>
<br>
On 08/28/2013 10:14 AM, Rob Fantini wrote:<br>
</div>
<blockquote cite="mid:521E13BF.4060609@fantinibakery.com"
type="cite">Hello Eneko
<br>
thank you for the info.
<br>
<br>
We're also going to use drbd active/active . We used drbd in
1.9 Primary/Secondary . And for awhile in pve version2 used
active/active.
<br>
<br>
We'll use active-active and run our high available kvm's only
on one node. I think that active-active drbd can be dangerous .
<br>
<br>
DRBD seems to be the most dependable way to have high available
storage for our setup. Next year we'll test Ceph , Sheepdog and
Glusterfs .
<br>
<br>
On Wed 28 Aug 2013 07:15:23 AM EDT, Eneko Lacunza wrote:
<br>
<blockquote type="cite">Hi Rob,
<br>
<br>
On 28/08/13 03:41, Rob Fantini wrote:
<br>
<blockquote type="cite">Please, can you give some information
about the 'shared-storage HA
<br>
setup' you mentioned?
<br>
<blockquote type="cite">We finally discarded it (were trying
active-active configuration)
<br>
in-house, we felt it gave us more headaches than the
peace-of-mind
<br>
potentially gained by a shared-storage HA setup. We didn't
have
<br>
experience with DRBD so maybe it was us not setting up
correctly.
<br>
</blockquote>
</blockquote>
What I meant is that we discarded to use DRBD in active-active
<br>
configuration, which is a way to have "shared" storage without
<br>
actually sharing physically an storage.
<br>
<br>
Cheers
<br>
Eneko
<br>
--
<br>
Zuzendari Teknikoa / Director Técnico
<br>
Binovo IT Human Project, S.L.
<br>
Telf. 943575997
<br>
943493611
<br>
Astigarraga bidea 2, planta 6 dcha., ofi. 3-2; 20180 Oiartzun
(Gipuzkoa)
<br>
<a class="moz-txt-link-abbreviated" href="http://www.binovo.es">www.binovo.es</a>
<br>
<br>
<br>
_______________________________________________
<br>
pve-user mailing list
<br>
<a class="moz-txt-link-abbreviated" href="mailto:pve-user@pve.proxmox.com">pve-user@pve.proxmox.com</a>
<br>
<a class="moz-txt-link-freetext" href="http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user">http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user</a>
<br>
</blockquote>
_______________________________________________
<br>
pve-user mailing list
<br>
<a class="moz-txt-link-abbreviated" href="mailto:pve-user@pve.proxmox.com">pve-user@pve.proxmox.com</a>
<br>
<a class="moz-txt-link-freetext" href="http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user">http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user</a>
<br>
</blockquote>
<br>
<br>
<div class="moz-signature">-- <br>
<img src="cid:part1.00040003.01040906@cie.com.mx" border="0"></div>
</body>
</html>