<html>
<head>
<meta http-equiv="content-type" content="text/html; charset=UTF-8">
</head>
<body>
<p>Hello,</p>
<p>I am currently testing LINSTOR/DRBD with Proxmox 7.3 on a three
nodes cluster with:</p>
<p>Nodes A and B:</p>
<ul>
<li>DRBD over ZFS</li>
<li>LINSTOR controller and satellite</li>
<li>DRBD reactor for HA setup (followed the guide at
<a class="moz-txt-link-freetext" href="https://linbit.com/drbd-user-guide/linstor-guide-1_0-en/#s-linstor_ha">https://linbit.com/drbd-user-guide/linstor-guide-1_0-en/#s-linstor_ha</a>)<br>
</li>
</ul>
<p>Node C:</p>
<ul>
<li>DRBD diskless</li>
<li>LINSTOR satellite</li>
</ul>
<p>Now, if I do the following:</p>
<ul>
<li>All nodes UP and a VM with Proxmox HA enabled and stored on
DRBD, running on node B.<br>
</li>
<li>Stop Node A.</li>
<li>Stop Node B.</li>
<li>Start again Node B.</li>
<li>The Proxmox GUI of Node B is no longer accessible until I
restart Node B. And the VM doesn't start.<br>
</li>
</ul>
<p>In the logs of Node B i can see "pvestatd[1990]: could not
connect to any LINSTOR controller at
/usr/share/perl5/PVE/Storage/Custom/LINSTORPlugin.pm line 214."</p>
<p>With drbdmon I can see "QUORUM LOST" on all resources.<br>
</p>
<p>I understand the reason that LINSTOR/DRBD won't let access to
DRBD ressources, since it can't really have quorum on the
resources (should I transform the diskless node into one with just
enough storage for the linstor_db?). But I don't think it is
normal that when there is no LINSTOR/DRBD quorum (but Proxmox
quorum), you can't access the GUI of some nodes (for example, for
running VMs located on another storage).</p>
<p>Thank you for any help.<br>
</p>
<p>Regards,</p>
<p>Nicolas.<br>
</p>
</body>
</html>