[ovirt-users] Re: [Gluster-users] Announcing Gluster release 5.5

I’m not quite done with my test upgrade to ovirt 4.3.x with gluster 5.5, but so far it’s looking good. I have NOT encountered the upgrade bugs listed as resolved in the 5.5 release notes. Strahil, I didn’t encounter the brick death issue and don’t have a bug ID handy for it, but so far I haven’t had any bricks die. I’m moving the last node of my hyperconverged test environment over today, and will followup again tomorrow on it. Separately, I upgraded my production nodes from ovirt 4.3.1 to 4.3.2 (they have a separate gluster server cluster which is still on 3.12.15), which seems to have moved to the gluster 5.3.2 release. While 5.3.0 clients were not having any trouble talking to my 3.12.15 servers, 5.3.2 hit https://bugzilla.redhat.com/show_bug.cgi?id=1651246 <https://bugzilla.redhat.com/show_bug.cgi?id=1651246>;, causing disconnects to one of my servers (but only one, oddly enough), raising the load on my other two servers and causing a lot of continuous healing. This lead to some stability issues with my hosted engine and general sluggishness of the ovirt UI. I also experienced problems migrating from 4.3.1 nodes, but that seems to have been related to the underlying gluster issues, as it seems to have cleared up onceI resolved the gluster problems. Since I was testing gluster 5.5 already, I moved my nodes to gluster 5.5 (instead of rolling them back) as the bug above was resolved in that version. That did the trick, and my cluster is back to normal and behaving properly again. So my gluster 5.5 experience has been positive so far, and it looks like 5.3 is a version for laying down and avoiding. I’ll update again tomorrow, and then flag the centos maintainers about 5.5 stability so it gets out of the -testing repo if all continues to go well. -Darrell