On May 22, 2020 6:48:43 PM GMT+03:00, Gianluca Cecchi <gianluca.cecchi@gmail.com> wrote:
On Fri, May 22, 2020 at 4:17 PM Derek Atkins <derek@ihtfp.com> wrote:
FTR: I don't think I need to check that the datacenter status is up;
I
added that in not really understanding the changes between 4.1 and 4.3. The issue is that the storage domain status isn't initialized to 'down' when the engine first comes up so my script was testing that and seeing all domains up when they really weren't.
Actually at least one time last week I had a situation where after a crash of a single host environment with 4.3.9 and gluster on host itself, all 3 gluster storage domains resulted active but actually only the engine rhev mount point was up and not the other 3 configured Something like this: /rhev/data-center/mnt/glusterSD/ovirtst.mydomai.storage:_engine
While for the other 3 storage domains I only had the gluster brick active but not the filesystem mounted. In web admin gui all the storage domains was marked as active.... and as soon as I powered on the first VM the operation went into error of course and only at that time the 3 storage domains were marked as down... From the web admin gui I was able to then activate them and start VMs. I had no time to investigate more and so opening a bug for that... The problem is that in my opinion the datacenter was marked as up too, so your check would not be of great meaning. In my opinion you could crosscheck also storage domains number with expected mount points of type /rhev/data-center/mnt/...
Gianluca
In order to have a full operational environment you will need (they depend on each other): 1. Engine is up and healthy 2. Master storage domain is up 3. Datacenter is up 4. A host is selected for SPM 5. The storage domains for the VM are up 6. At least one Host in the cluster is up and running (usually that's the node in step 4) Best Regards, Strahil Nikolov