Disable monitoring of specific ESX host

[Biga]

We have Veeam 7. We monitor about 8 ESX hosts
I want to disable 1 ESX host from monitoring. It seems simple.
Just uncheck the ESX host in the Veeam MP Gui on the VMware Servers tab.
But still I get alerts from this host. I restarted the Veeam collector and Extension service, but that didn't work.
What am I missing?
Thanks

[sergey.g]

Hi,

Could you tell us which alert is still arriving after you unchecked the host in UI? If there was an issue with Collector<->Extensions server connection, Collector may still generate a topology with unchecked host in it. So first thing I would check is the topology.xml file in the Collector\Logs folder, check if not-monitored server is there or not.

Looking forward to your feedback.

Thanks.

[Biga]

The not-monitored esx host is not in the toplogy.xml file. The alerts I get are about the local storage on the server:

Alert: Veeam VMware: Datastore Latency Analysis
Source: VMFS02
Path: Not Present
Last modified by: System
Last modified time: 10/26/2015 4:05:14 PM Alert description: Datastore VMFS02 exhibited high max latency of 56 ms.
The latest highest latency values for related metrics are - Read Latency 56 ms reported by Host esxserver01.domain Write Latency 8 ms reported by Host esxserver01.domain
Current Datastore load is 10 total IOPS with a usage rate of 4 KB/s read and 97 KB/s write.
Top 5 Hosts for latency and IOPS to this datastore are below - Host Name - MaxLatency ms, (IOPS) esxserver01.domain - 56 ms, (10 iops)

Thanks

[sergey.g]

Hi Steve,

This is happening because this is a datastore alarm and datastores can be connected to many hosts in your environment and if unchecked host is affecting datastore performance we believe that you should be informed about it. Imagine that you have major storage latency issue with your virtual machines, but you don't see any issues on the backend storage because host which is creating excessive load is filtered out in performance and datastore is displayed as healthy although it's not. That's why we believe that we cannot filter out unchecked hosts in performance of other shared objects connected to the monitored parts of the infrastructure.

Thanks.

[Biga]

You're statement is correct if this ESX server was connected to or network storage(HP Lefthand). But it is a standalone ESX server with only local storage that we use for testing.
Testing the Veeam MP among other things. The VMFS02 volume is local disk on the ESX server
So I think disabling this host from monitoring, should also disable datastore alerts.
Thanks

[sergey.g]

Hi Steve,

I see your point and I think it's a valid assumtion. We also had similar thoughts for our current release and I think we've already fixed this in v8, hosts which are unchecked for monitoring should not have associated datastores in SCOM, but I'm afraid I need to double-check this. We've been adding this primarily for scalability, but I'm sure your case should perfectly fit as a use case for this functionality too. I'll update you with details.

Thanks

[Biga]

Hi Sergey,
I was about to open a support case. Is there a workaround in v7?
I could override the Datastore Latency Analysis monitor for the VMFS02 volume, but that's hiding the problem instead of solving it.
I don't really have the time to start a upgrade v8 project, because of other projects/activities. If it's an easy/quick upgrade, I can consider it.
I'll wait for your response.
Thanks

[sergey.g]

Hi,

Well there is a special overridable parameter for Veeam Stage 2 - Datastore discovery: ExcludeDatastoresByDatastoreName. You can specify a mask or individual datastores which should not be discovered, so you can put such datastores form unchecked hosts there.

We could do more flexible solution if we had a way to group such datastores, but I'm afraid in SCOM they look exactly the same as shared datastores or datastores connected to monitored hosts. The only difference could be the fact that a datastore from unmonitored host won't have VMs on it, but this could be also true for empty datastore connected to a monitored host, so we cannot use this as a group membership rule.

[Biga]

I created an override for Veeam Stage 2 - Datastore discovery. But if I had a second or more standalone servers with the same naming convention for the local disk, I couldn't use this option.
And if I would install another standalone ESX server, i have to remember not to name the local disks the same as in the override.
For now the override is a quick fix.
But would like to know if there is a better solution in v8.

[sergey.g]

Hi Steve,

I can confirm that in v8 when you remove a host from monitoring - all releated local datastores will be removed too.

Thanks.