<div dir="ltr">I understand. So my setup is maybe wrong. Vijay, could you please explain how this dummy node setup would look like? <div><br></div><div>Do you recommend to setup a glusterd on node3 and replicate to 3 servers? In my understanding this would significantly reduce performance as files have to be replicated 3 times. Would it be possible to mount the volume if 2 of 3 glusterd servers are live?</div></div><div class="gmail_extra"><br><div class="gmail_quote">2015-08-31 19:41 GMT+02:00 Vijay Bellur <span dir="ltr"><<a href="mailto:vbellur@redhat.com" target="_blank">vbellur@redhat.com</a>></span>:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span class="">On Monday 31 August 2015 10:42 PM, Atin Mukherjee wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
> 2. Server2 dies. Server1 has to reboot.<br>
><br>
> In this case the service stays down. It is inpossible to remount the<br>
share without Server1. This is not acceptable for a High Availability<br>
System and I believe also not intended, but a misconfiguration or bug.<br>
This is exactly what I gave as an example in the thread (please read<br>
again). GlusterD is not supposed to start brick process if its other<br>
counter part hasn't come up yet in a 2 node setup. The reason it has<br>
been designed in this way is to block GlusterD on operating on a volume<br>
which could be stale as the node was down and cluster was operational<br>
earlier.<br>
</blockquote>
<br></span>
For two node deployments, a third dummy node is recommended to ensure that quorum is maintained when one of the nodes is down.<br>
<br>
Regards,<br>
Vijay<br>
</blockquote></div><br></div>