A Great Management Pack
We need a brand new Management Pack that sets the reference for fabric monitoring. No more OpsMgr SDK dependencies to take into account managed service provider scenarios.
Ben Hummerstone commented
Additionally, when creating the OpsMgr connector, the OpsMgr Run As account is created "less secure" by default, which can generate warnings if monitoring Domain Controllers as the account cannot log in locally. It would be helpful to automatically limit this to the required computers, or at least update the documentation to assist.
Here is some feedback on the MP:
There are some monitors that cannot be reset at all
Some of my Hyper-v Hosts are reported as not being monitored. The reason for that is because the top level object is shown as not monitored although all objects beneath are being monitored. This happens only to Hyper-V hosts and VMM server itself.
There are some of the Views that show no objects at all - IP Address Pool performance, Mac Address Pool Performance. There are views that show health of object but these objects are not monitored and there are not monitors to enable for them - Mac Address Pool Health, IP Address Pool Health.
On TechNet there is article for monitoring Storage Space SOFS cluster with VMM MP but there is no other information how to enable or change any monitors. I tried searching those monitors but couldn't found any related to this scenario.
Most of the monitors do not work properly - example I have ASR Provider which is currently unhealthy and cannot connect to Azure but VMM shows it as healthy.
The MP does not show compliance of Virtual Network adapters and etc.
Overall the experience with the MP is not good.
Alexey Zhuravlev commented
I think I can. The current OpsMgr-VMM MPs are using the direct SDK calls for discovery, for passing all monitoring data etc. This design makes impossible to manage almost anything: you cannot override discoveries, you cannot fix the things yourself, and it's very hard to customize it. New management pack should be designed exactly like any other management pack - no direct SDK calls, everything is done via OpsMgr agent and 'mechanics'.
Hope this helps
Can you elaborate what you mean with new MP? What do you want to see monitored?