<div dir="ltr">Ok, I've got the plan:<div>1. unmount volume on all clients</div><div>2. shut down glusterd on all nodes</div><div>3. set old op.version (310xx) via /var/lib/glusterd/<a href="http://glusterd.info">glusterd.info</a> on all nodes</div><div>4. downgrade rpm packages everywhere</div><div>5. start glusterd & mount volume</div><div><br></div><div>Does it look fine?</div><div>Also is there any way to roll back without downtime?</div><div><br></div><div>Thanks.</div></div><div class="gmail_extra"><br><div class="gmail_quote">2018-01-13 4:51 GMT+03:00 Pavel Znamensky <span dir="ltr"><<a href="mailto:kompastver@gmail.com" target="_blank">kompastver@gmail.com</a>></span>:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr"><div>Hi,</div><div>I want to roll back a distributed replicated cluster from 3.12 to 3.10 because of performance issues which appeared after upgrade.</div><div>But I didn't find any docs which could help me.</div><div>Is it even possible?</div></div>
</blockquote></div><br></div>