[Bugs] [Bug 1727682] Upgrade from older versions 3.12.15 and 5.5 to gluster 6.3-1 fails - bricks become offline

[bugzilla at redhat.com]

https://bugzilla.redhat.com/show_bug.cgi?id=1727682



--- Comment #11 from Amar Tumballi <atumball at redhat.com> ---
Amgad, I was able to find out the reason for the failure of binding to 24007.
Culprit is me :-) ref: https://review.gluster.org/#/c/glusterfs/+/21249/

With this change, we actually added listen-port value of 24007 to glusterd.vol
file. But as we had not had a case of anyone changing glusterd.vol before, in
most of the upgrades, the update happened properly, as new glusterd.vol file
was overwriting the old one.

But as in your case, you had edited glusterd.vol file, rpm install didn't
overwrite the glusterd.vol file, and that resulted latest (glusterfs-6.x+)
glusterd process to come up in different ports. So, the actual fix is to add
the line in volfile, which we did as a 'workaround' earlier. So, please make
this change through your scripts (to add 'option listen-port 24007' to your
glusterd.vol files before upgrade).


About systemctl, I figured out because we handled the process outside of the
cgroups, and other required parameters once we started, it was getting into
problem.  A systemctl disable glusterd and enable glusterd is the right way to
get back to normalcy.

About 3, I guess Sanju's patch link should be option, which makes all the
issues observed addressed.

-- 
You are receiving this mail because:
You are on the CC list for the bug.