I have an issue with ovirt where it thinks a task "Handling non
responsive host hv1" is stuck Executing. Every time, after the server
is fenced and comes back online, when I try to activate it this host is
fenced/rebooted. What happened before the attempt to reboot was
related to doing normal OS updates as follows.
1. Update OS/RPMs on ovirt engine host and reboot it
2. Move all VMs from hv1 to other hypervisors and put into maintenance
mode
3. install updates via ovirt web application by right clicking hosts
and selecting "upgrade"
4. install all available updates via yum and reboot hosts
5. Once hosts has rebooted right click > confirm host has been rebooted
6. attempt to activate hosts
At this point the hosts was fenced forcing a reboot and now I can't
seem to get this host up and working. I noticed that with the yum
updates or ovirt upgrade option it installed a 3.8.x virsion of
glusterfs packages which currently everything else is running 3.7.x
versions so I downgraded the ovirt rpms to the previous
version. Manually rebooted and attempted to activate the host which
produced a fencing of this hosts and still a Handling non responsive
host task in the ovirt web interface. My thought on the ovirt is that
the host, after being upgraded to the newer glusterfs rpms, couldn't
connect properly to the gluster sans.
I'm not sure what else to check or do here since I believe until this
stuck task is cleared this hosts will continue to be fenced everytime I
try to activate it.
I can ping all 3 gluster sans and the ovirt engine server from this
affected hosts.
Running:
ovirt 4.0.6.1-1
glusterfs 3.7.20-1
All hosts/sans are running CentOS 7.3 and the one I upgraded now shows
7.4
--
Edward Clay
Systems Adminstrator
UK2 Group -- US Operations
Phone: 1-800-222-2165
E-Mail: edward.clay(a)uk2group.com