I agree -> fencing is mandatory.<div id="yMail_cursorElementTracker_1622239704146"><br></div><div id="yMail_cursorElementTracker_1622239742860">You can enable the debug logs by editing corosync.conf or /etc/sysconfig/pacemaker.</div><div id="yMail_cursorElementTracker_1622239790426"><br></div><div id="yMail_cursorElementTracker_1622239790668">In case simple reload doesn't work, you can set the cluster in global maintenance, stop and then start the stack.</div><div id="yMail_cursorElementTracker_1622239879060"><br></div><div id="yMail_cursorElementTracker_1622239879265"><br></div><div id="yMail_cursorElementTracker_1622239879452">Best Regards,</div><div id="yMail_cursorElementTracker_1622239885345">Strahil Nikolov</div><div id="yMail_cursorElementTracker_1622239704410"> <br> <blockquote style="margin: 0 0 20px 0;"> <div style="font-family:Roboto, sans-serif; color:#6D00F6;"> <div>On Fri, May 28, 2021 at 22:13, Digimer</div><div><lists@alteeve.ca> wrote:</div> </div> <div style="padding: 10px 0 0 20px; margin: 10px 0 0 0; border-left: 1px solid #6D00F6;"> On 2021-05-28 3:08 p.m., Eric Robinson wrote:<br clear="none">> <br clear="none">>> -----Original Message-----<br clear="none">>> From: Digimer <<a shape="rect" ymailto="mailto:lists@alteeve.ca" href="mailto:lists@alteeve.ca">lists@alteeve.ca</a>><br clear="none">>> Sent: Friday, May 28, 2021 12:43 PM<br clear="none">>> To: Cluster Labs - All topics related to open-source clustering welcomed<br clear="none">>> <<a shape="rect" ymailto="mailto:users@clusterlabs.org" href="mailto:users@clusterlabs.org">users@clusterlabs.org</a>>; Eric Robinson <<a shape="rect" ymailto="mailto:eric.robinson@psmnv.com" href="mailto:eric.robinson@psmnv.com">eric.robinson@psmnv.com</a>>; Strahil<br clear="none">>> Nikolov <<a shape="rect" ymailto="mailto:hunter86_bg@yahoo.com" href="mailto:hunter86_bg@yahoo.com">hunter86_bg@yahoo.com</a>><br clear="none">>> Subject: Re: [ClusterLabs] Cluster Stopped, No Messages?<br clear="none">>><br clear="none">>> Shared storage is not what triggers the need for fencing. Coordinating actions<br clear="none">>> is what triggers the need. Specifically; If you can run resource on both/all<br clear="none">>> nodes at the same time, you don't need HA. If you can't, you need fencing.<br clear="none">>><br clear="none">>> Digimer<br clear="none">> <br clear="none">> Thanks. That said, there is no fencing, so any thoughts on why the node behaved the way it did?<br clear="none"><br clear="none">Without fencing, when a communication or membership issues arises, it's<br clear="none">hard to predict what will happen.<br clear="none"><br clear="none">I don't see anything in the short log snippet to indicate what happened.<br clear="none">What's on the other node during the event? When did the node disappear<br clear="none">and when was it rejoined, to help find relevant log entries?<br clear="none"><br clear="none">Going forward, if you want predictable and reliable operation, implement<br clear="none">fencing asap. Fencing is required.<div class="yqt3966126429" id="yqtfd99493"><br clear="none"><br clear="none">-- <br clear="none">Digimer<br clear="none">Papers and Projects: <a shape="rect" href="https://alteeve.com/w/" target="_blank">https://alteeve.com/w/</a><br clear="none">"I am, somehow, less interested in the weight and convolutions of<br clear="none">Einstein’s brain than in the near certainty that people of equal talent<br clear="none">have lived and died in cotton fields and sweatshops." - Stephen Jay Gould<br clear="none"></div> </div> </blockquote></div>