Live migration stops metrics collecting

[Cragdoo]

Maybe not a Veeam MP issue, but thought I'd ask anyway.

We had a linux VM running on our Hyper-V cluster that we had to Live Storage Migrate onto a new CSV. Prior to the move we were able to collect and see the 'Commands per Second' metric , however since moving the VM the monitor appears to be greyed out.

Last trick to try was going to be restarting the Health Service on the hyper-v hosts in the cluster.

Anyone seen anything similar ?

[sergey.g]

Hi Craig,

Greyed out monitor means that Health Service which manages the monitor is not available. It's called Health Service outage in SCOM DW DB terms. You really need to check what is going on with the health service if restart didn't bring everything back online. Check if there are any clues in the Operations Manager logs on the affected Health Service (it will be different if besides storage change, VM also migrated to another cluster node).

Let me know if you have any other questions.

[Cragdoo]

It does appear to be a Veeam MP issue, as the standard MPs are showing the CSV and Hosts as healthy/online.

I'll concentrate on the Health Service, and let you know how I get on.

[sergey.g]

Hi,

Well, if standard means SCVMM MP, then it could be collecting and processing info on SCVMM rather than on the Hyper-V host and in this case Health Service is different. It's debatable which one is better, but in case of SCVMM you have single point of failure for your monitoring.

If standard means Microsoft's Hyper-V MP, then it should be same Health Service and in this case it's really strange that one object is in the HS-outage state while the other one is healthy.

Let us know what you find out.
Thanks.

[Cragdoo]

I am referring to Microsoft's Hyper-V MP.

Apologies is this sounds like a newbie question, but the Health Service is running where ...and it can be safely restarted without impacting the cluster and/or SCOM itself?

[Cragdoo]

Here is what I see from the Microsoft Hyper-V 2012 R2 MP




and now what I see from the Veeam MP

[sergey.g]

Hi Craig,

OK, this looks bad and unfortunate for us - this is really our problem, this state is not "grey" (health service outage), this is a "white" state (Unmonitored), which means that monitor failed to initialize for some reason. It could be performance-realted issue (our MP has a little bit more workflows than Microsoft's, plus I beleive when you are looking onto this Hyper-V Host Computer state, you are seeing Windows Computer object which has minimum set of monitors)

You can restart the Health Service, which is Microsoft Monitoring Agent or System Center Monitoring Agent, it's not affecting Hyper-V.
If the situation remains we definitely need to check the Operations Manager event log, there should be clues on why our workflows are failing. Did you previously see anything like this or only VM migration "sparkled" the issue?

I would appreciate if you can open a support case with our tech support team (you can send me a ticket number), we can arrange a webex to understand what is happening on your Hyper-V hosts. I'm sure we can make our MPs to work correctly.

Thanks.

[Cragdoo]

Thanks Sergey

I'll try restarting the Monitoring Agent, and waiting 30 mins or so , to see if the status changes. If it is still failing I'll log a call with support.

[Cragdoo]

restarting the Microsoft Monitoring Agent has resolved the hosts State.


Will leave this overnight to see if the metrics start reporting.

[Cragdoo]

Having left this overnight, in the hope metrics are being collected, I logged on this morning to find they weren't .

Looking at the status of the hosts in SCOM, I see that the "Hyper-V Host Virtual Machines Container (Veeam MP)" is reporting 3 of the hosts as 'not monitored', could this be the reason that the VM metrics are not being collected?

[Alec King]

Hi Craig,

Well the HOSTs are monitored - this is the first column in the State view. But it's the Virtual Machine CONTAINER on those Hosts which is Not Monitored.

Could be because there are no VMs 'attached' to those Hosts - therefore no state to roll up. If you open a Diagram view (right click Host - Open - Diagram View) - are there VMs in the container?

Could be that our discovery has not updated the VMs and moved them to the new Host.

[Cragdoo]

Yeah, there's definitely VMs running on the hosts




I've just been informed that the SSL cert used by the agents has expired. I need to wait on the cert being renewed

Hopefully this is the issue