About a month ago, I completely rebuilt my oVirt cluster, as I needed to move all my hardware from 1 data center to another with minimal downtime.
All my hardware is in the new data center (yay for HIPAA compliance and 24/7 access, unlike the old place!)
I originally built the cluster as a single-node hyperconverged. I then added nodes to it, and then, finally, I reconfigured gluster to run in a replica 2 / arbiter 1 configuration.
As of now:
- I have 4 compute hosts
- Gluster is running fine and replicating fine between the two full replicas, along with the arbiter node
- I need to revisit my recent email about gluster replication speed, though
However, I'm worried about two things:
- I think that my hosts, and hosted-engine, are all still configured to use the single mount point from the original single-node hyperconverged ... i.e. if I shutdown / reboot that host, then the storage "goes away"
- How do I reconfigure oVirt to use the 2nd replica as a secondary mount point?
- Currently, the Engine is deployed on two of the servers (a and c). But any time one of those servers is online (c), the hosted-engine insists on running on that server c. I cannot migrate the engine off of c.
And if the engine is running on the other host a, then I turn host c back on, the engine shuts down from a, and comes back up on c. - How do I "fix" this so that the engine will run on host a, even when host c is turned on?
- How do I deploy the hosted-engine to host b?
- Is it as simple as logging into host b, and running "hosted-engine --deploy" ?