Some VMware hosts appear as "not monitored" in SCOM

[Stoyan Chalakov]

Hello Community and Supporters,

I have pretty strange issue with the Veeam SC MPv8 on SCOM 2012 R2 UR6. I am currently doing PoC for the VMWare Monitoring with the Veeam SC MP for a big customer of mine. After installing and configuring the MP with two Collectors I noticed that some of the hosts (all in the same cluster) appear as not monitored (4 out of 16). The other components of those same hosts (VM Container, vSphere Host NIc Container, HBA Container, etc...) are shown as monitored and even the cluster itself is shown as "monitored" (only the State of the hosts is shown as "not monitored"). The Veeam UI shows them as "monitored" and the collectors are green (Available).

I already checked the forum threads and found only one with the same issue description, but unfortunately it hasn't been resolved:
http://forums.veeam.com/veeam-managemen ... 24557.html

What I have already done:

- I made sure the collectors are not overloaded. I have two collectors for 16 hosts and they are pretty powerful (8GB Ram, 4 vCPUs)
- I did a Rebuild of the Topology (Full)
- I moved the Jobs on only one collector to check if it somehow related to a specific collector.
- I have removed and re added the hosts for monitoring.
- I've restarted the the Health Service on both collectors.
- I've even restarted the servers themselves.
- "Act as proxy" is checked on all collectors.
- The collectors are shown healthy in both SCOM and the Veeam UI.

I would gladly open a case for this, but currently we are in a PoC phase and we have no support contract.

Many thanks in advance for all the help and hints.

Regards,
Stoyan

[Alec King]

Hello Stoyan,

Could you try to clear the SCOM cache on both Collectors, as in this KB article - http://www.veeam.com/kb1043

After this, wait several intervals (e.g. 45 minutes - 1 hour) for cache to rebuild. Then check status of those hosts again.

Let us know how it goes. Thanks!

Alec

[Stoyan Chalakov]

Wow, you guys are really great. Thank you for the fast and accurate reply.
It seems that this was exactly the cause for this. After rebuilding the HealthService cache the hosts now appear as monitored.
Alec, can I prevent somehow this from happening again?
Do I have to make additional changes to the HealthService? I already ran the task "Configure Health Service" shortly after installing the MP as it displayed an error, about it not being configured properly for the Collector.

I think my customer will be more than happy with both the product and the support.

Best Regards,
Stoyan

[Alec King]

Hi Stoyan,

Great that it's all working now!

I wanted to make one point - you CAN raise a Support Case, even when on an Evaluation/POC. Just go to Veeam Customer Support portal https://cp.veeam.com/ and raise a case using "Evaluation" as the current support entitlement. For sure we want all POCs to be successful, so our support team will help you with those!

Regarding the Health Service, if you ran Configure Health Service task, then that is all you need to do. Sometimes the Health Service gets in a bad state, and clear cache is the way to fix it. But it doesn't happen often.
Note - We do recommend a SCOM Management Server for Collector, especially in large environments - if the Collector Load% approaches 100%, then I would consider using SCOM Management Servers. They will be more scalable and stable.

Hope the rest of your testing is successful!

Cheers
Alec

[Stoyan Chalakov]

Hi Alec,

Thank You very much once again. Good to know that you offer support even when customers are evaluating.
All the testing so far shows great results and only confirms the opinion of the community regarding the MP.
Wish you a successfull rest of the week.

Regards,
Stoyan

[Alec King]

Thank you for the kind words Stoyan!
Good luck with the POC and if we can help - just let us know!

Cheers
Alec

[Stoyan Chalakov]

Hi Alec,

I need your opinion on this once again. As I mentioned my customer is doing a PoC for the VMware monitoring. The VeeaM MP has been approved for usage in the PROD environment and will be implemented there. So far so good.
Still I have the strange issue in the Acceptance env. Some random hosts (small number out of all) are again shown as not monitored (only the hosts, thir compoenents appear as monitored.) When I delete the HealthService cache on the collectors this issue is resolved for a couple fo days or a week. Afterwards it occurs again. As of my first writing here I have emptied the cache two times on the collectors. This helps every time, but for a limited amount of time.
I also made sure the HealthService on the collectors is properly configured (I executed the health service configureation task).

Do you think it could be related to the number of collectors? I have only two collectors for 16 hosts (Trial license), but they are good equiped. I also made sure they have sufficent resources to handle the load of the data collection.
Should I open a case about this?

Many thanks again!

Regards,
Stoyan

[sergey.g]

Hi Stoyan,

This has to be something exotic which I'm sure we should be able to fix.

Could you upload Operations Manager event logs from both collectors and from the management server somewhere and share the link? Feel free to send me a DM about that. Also in some situations, our support should be able to help you even without a contract, especially when you are in the trial mode, we'll include them in the loop if necessary. For now event logs from collectors should be enough to do the initial troubleshooting.

Objects become unmonitored when monitors fail in the health service on a collector side, so we need to understand why they are failing. It shouldn't be a performance issue because 16 hosts is a really small environment, you should be able to manage this with a single collector. However, sometimes there could be some exotic host or vcenter configuration which may cause performance issues.

So, I'm waiting for logs, there should be errors there or on the Management Server side, but let's go step-by-step, we'll include support if we can't find any obvious reasons on collector servers.

Thank you very much for keeping us in the loop with this PoC.

[Stoyan Chalakov]

Hi Sergey,

I posted the details about the requested log files on PM.

Thanks in advance!

Regards.
Stoyan