Applying maintenance mode to vCPU object

[nic]

Is there any way to apply maintenance mode to a vCPU? We need to schedule maintenance mode on a handful of servers during our weekly full AV scan because the full system scan tends to max out the CPU on Windows 2003 single CPU servers and older. We can do this with Windows OS objects and target only the CPU object with a script to schedule maintenance mode, however we can't get any more granular with the Veeam objects than the VM object itself. We want to avoid monitoring black holes by putting the entire VM in maintenance mode. Open to any other suggestions on how to suppress the CPU Analysis alerts on a scheduled basis while AV scans run.

thanks!

[Alec King]

Hi nic,

One method could be, to use a script to apply an override MP, instead of applying maintenance mode.

By setting a scheduled task to import an MP, containing a single override for Veeam VM CPU Monitor, then you can have granular control over that single monitor without disabling the whole VM object.

However, importing override MP will take longer to take effect than setting maintenance mode.
The MP must be imported in SCOM Management Server, then delivered to all relevant agents, then loaded by the agent, causing additional network traffic, agent cpu usage, and overall additional load on SCOM. The same applies in reverse when running another Task to delete the Override MP.

However since Microsoft do not have the feature of scheduled overrides, above is the only method to achieve your aims.

Hope that helps,
Alec

[nic]

Thanks Alec,
We're hoping to avoid the churn of enabling/disabling monitors with that kind of frequency. I think we'll end up permanently overriding certain monitors in that case and use other metrics to get an accurate view of server health.

thanks!

[sergey.g]

Hi Nic,

Have another idea for you. Do you put these CPU objects in Mmode as a group or by a certain event? If you are using a group, then it's super-easy (well, for me, but I can help) to create a group of VMs which contain in-guest objects which contained in VMs which are in the Mmode group. Then you just apply overrides for such a group to increase thresholds for CPU analysis (or numsamples paramter to set how long AV scan can eat CPU). In this case there will be no overrides back and forth, VMs will by just either put or removed from a group and based on that they are going to be either in our out of overrides context. Even if you don't use a group, I think I can help creating a script-based group discovery which puts VMs in the group.

The downside of group discovery - it will take time for a VM to get in the group, so you should have agressive discovery interval for it.

So, let me know more about your scheduling script - we may be able to built in something for VMs.

Thanks.