RHEVM: Monitor - Hypervisors "down"

Latest response

Have a cluster with two RHEV hypervisors.
Looking in the RHEVM GUI monitor page these are reported as down (A red arrow in the graphs, which results in no percent report on the CPU,Mem or Network utilization).

They have status UP in the cluster and have running VMs on them with no issues.

Why is the monitor not working for these hosts?
Any tips on where to start debugging this would be great.

Responses

Has this anything to do with that I haven't setup the power management on these hosts yet?
 

You are saying, that when you look in the "Hosts" tab, the hosts are shown as "Down", but if you look at the "Clusters" tab, "Hosts" subtab, the hosts are "Up"? 

Dan,

Found the issue. For some reason the hypervisors had been added (by mistake) to two different clusters.
Weird that this is even possible. The fauly ones was in "approval" state. Removing them made the correct hypervisors pop up in the Monitor.

definitely shouldn't be happening, can you maybe recall how exactly you made it into such a situation, so I can reproduce and investigate?

Exactly, I'm afraid not.

More or less:
1. Add host(s) to default cluster. 
2. Try to add the host to a different cluster without approving it in default cluster.
3. Approve the host in the second cluster.

I think that's the steps. Then you should see the host on monitor page, but as "down".
Then remove the host in the default cluster and then the correct host should be visible on the monitor page.

Hi Pär,

Did you manage to get this issue resolved? I can follow up on this if you need further assistance.

Hi David,

I did manage to resolve it. Thanks.

Would you mind sharing how you solved it? It may help others in a similar situation.

I did state my resolution.
But to summarize:
The affected hypervisors was added to the default cluster (but not approving them). Then I added them to the correct cluster and approved them.
At this point they were working fine, but the monitor showed them as "down". Removing them from the wrong cluster solved the monitor issue. 

Take a look at support case number 00772702.  I think these are both the same issue from different points of view.

- Greg Scott

Ah sorry, I misunderstood. Thought you were referring to a separate issue. Thanks!