[ovirt-users] Re: [No question] NFS disabled, hosts wandering tearful

Hello, This is a simple testimony about what happened yesterday in one of our DC. This DC runs on a dedicated bare-metal engine, oversized compared to the need, thus I've added a NFS service on it to host a small storage domain and the ISO storage domain. Yesterday, after having received the colorful announce about the 4.2.5 version, I decided to upgrade. As our engine was still on a CentOS 7.4, I first upgraded its OS version to 7.5, then reboot. Smooth. Then I followed the very usual oVirt engine upgrade path. Smooth. Eventually, I upgraded the hosts with ovirt-ansible-cluster-upgrade as usual. The result was frightening because the hosts were put in maintenance, upgraded, back to life, seen unavailable, unreachable, connecting, alive, rebooted, then back to another turn and looping... During this, the SPM role was obviously jumping around, and that did not help the debug. In the end, it appeared that something during an upgrade stopped and disabled the NFS service. My hosts partially relied on it, so after having restarted the NFS service, all came back to life. The NFS disabling may come from the CentOS upgrade, except if someone tells me it could come from something on the oVirt side?

I'm sure the RH people will advice me not to run NFS on the engine, but apart this event, I had no trouble doing this in years.