Coffee Break: Azure Stack HCI Management Pack Deep Dive

In the latest SCOM Coffee Break we took a deep dive into the new Azure Stack HCI Management Pack (MP) from the Microsoft SCOM Product Team. Bhavna Appayya, from the Microsoft SCOM Product Team, walked through the features and some hints and tips you’ll need as you dive into using this new MP.

But first, let’s quickly explain what Azure Stack HCI is.

What is Azure Stack HCI?

Traditionally, you would have had a bunch of your machines running in Azure and another bunch of your machines on-prem, running on physical hardware. These groups of machines would have been entirely separate and managed independently of each other. In addition, your on-prem VMs would also be managed entirely separately. This meant that, ultimately, there was no easy route to cloud for you.

With Azure Stack HCI, you host your workload where you like. Your on-prem machines are managed as your cloud ones are holistically, all together, as if they were cloud machines (and so are hyper converged infrastructure, or HCI). The idea is, that this makes cloud transitions and the hybrid scenario a lot easier and it is easier to cloud burst your workloads as needed. This is good news as the world is shifting toward cloud these days.

The management pack we’re going to be talking about today is all about monitoring this scenario.

The Azure Stack HCI Management Pack

First, here are some of the highlights of Microsoft’s new Azure Stack HCI management pack.

Azure Stack HCI as a platform is an industry shift for Microsoft. Microsoft is seeing more and more on-prem data centers shifting to HCI from traditional environments they are expecting that customers will be interested in using the Azure Stack HCI OS on their host servers.

The OS combines compute and storage into a single box, which means more cost savings for users. The underlying hardware is still the same, but the common targets running on HCI will be the ones running, for example, on a Hyper-V host.

One thing to keep in mind is the host software such as, SQL Server or web server software doesn’t run directly on top of Azure Stack HCIs OS. This will run in the guest VMs running on top of the HCI fabric. And that’s why the goal of this Azure Stack management pack is tweaked towards some of these scenarios.

The management packs that Microsoft created for HCI support versions 21H2 and 20H2 of Azure Stack HCI OS. They are designed to work for SCOM 2019 UR3 or later versions.

The new management packs were bundled with the following capabilities:

You don’t have to import all the management packs. So if, for example, you do not have SDN workloads that you want to monitor in your HCI landscape and you only S2D or cluster related scenarios for HCI clusters, then you can choose to import only that particular management pack.

Base OS MP Scenarios

The first MP in the bundle is the Base Azure Stack HCI OS. This is the core OS MP for Azure Stack HCI and monitors.

This lets you monitor the performance, health, and availability of the Azure Stack HCI server operating systems. With this MP, aspects of critical and required services – like the disk managers, servers, workstations, storage drivers, network adapters, etc. – are checked for connection health and availability.

Microsoft has also included the standard performance indicators for processors. The memory related counters and logical disks are also checked. The focus is on logical disks here, but it can be tweaked to do some monitoring support for physical disks as well if required.

You can see in the above image that an Azure Stack HCI folder is created on the left side of the screenshot with different folders underneath that. The first folder that is expanded shows all the dashboards that are available in the base OS MP for Azure Stack HCI.

Hyper-V MP Scenarios

The next MP is the Hyper-V management pack. This monitors the critical services, like networking and storage components, to ensure that the health of the Hyper-V role is good.

This management pack also allows you to monitor events that result from errors on virtual machines. Additionally, you can monitor the configuration of hosts that are using the Microsoft RemoteFX technology. Plus, you will be able to monitor the availability and configuration of hosts that are using the Microsoft Hyper-V replication technology. So, virtual machine replication related scenarios are also supported for HCI.

Under the root Azure Stack HCI folder, there is a subfolder for the Hyper-V management pack, if you decide to import it install it and all the dashboards that you can see in the Operations console as well.

S2D MP Scenarios

Then there is the S2D MP – the Storage Spaces Direct management pack. This is for monitoring storage pool and storage node-related scenarios. Again, this lets you monitor the health; the performance data is collected; and all the alerts on the fault event, for example, for the storage pool would be available too.

The focus here is monitoring physical disk or servers from the cluster, in this case. Plus, you can monitor the health and performance counters of cluster shared volumes in a cluster, plus other things like cluster quorum, and networks. 

There are several dashboards available for each of these under the Storage Spaces Direct folder in SCOM.

SDN MP Scenarios

The fourth MP in the bundle is the software defined network (SDN) management pack. This management pack helps you monitor the SDN infrastructure components and tenant resources on computers that are running Azure Stack HCI.

Microsoft added support for more than one network controller in this MP. So, the network controller monitors now include checks to be able to support more than one network controller and it could have all the availability related checks for, as an example, the ports that are needed by them.

There is gateway monitoring included so things like virtual network monitors, gateway pool monitors, and virtual gateway monitors are all supported in the HCI related synopsis.

One thing to keep in mind here is that Microsoft recommends that, if you use the Azure Stack HCI SD MP and the Windows Server SDN MP in the same environment, you upgrade or install the latest version of the Windows Server SDN MP (Microsoft has published an update for that). 

Cluster Management MP Scenarios

Finally, you also get the cluster management pack. This MP covers monitoring aspects where possible issues are identified with starting the cluster servers or connectivity between cluster nodes. This includes things like the AD DS settings that could affect the cluster. For example, when there are permissions needed by the computer account that is used by the cluster. Alerts for that are taken care of as well. In addition, if there are issues with the domain name system, they will be flagged. The MP also covers monitoring availability of a cluster resource and storage of the cluster.

You will see, in the cluster management folder, all the dashboards that are included in this particular management pack for HCI.

Demo

Watch the 5-minute demo video of the Azure Stack HCI MP given by Bhavna Appayya from the Microsoft SCOM Product Team:

Hints and tips

There are a few hints and tips to keep in mind when you import the Azure Stack HCI MP, some prerequisites, and a few known issues.

First, when you import the Azure Stack HCI MPs into your environment, make sure that the common library management pack is imported prior to importing the main MSI for the particular entity you decide to use. This is prepackaged and taken care of, so you don’t have to worry too much about going out of your way and finding the MP, but it’s worth noting.

In terms of the order in which you want to import the MPs, you can just follow the built-in order that Microsoft has for the downloadable files. Just follow the instruction on the screen as it prompts you.

There are two scenarios for which you may use the Azure Stack HCI management pack. One could be that you have an existing Windows Server related setup in your environment and the HCI landscape, side-by-side. In this case, it could be that you have Windows Server Management Packs, for example for SDN, as well as the Azure Stack HCI management pack for SDN on the same SCOM installation. If that’s the case, there are a couple of things to keep in mind.

If you don’t have any Windows Server management packs installed in your SCOM landscape where you decide to use the HCI management packs, then you can skip the override management pack which would need to be imported. Everything should work fine.

But if you do have existing Windows Server Management Packs, or have the Storage Spaces Direct MP or Hyper-V MP, then you need to import the override management pack that is bundled with the Azure Stack HCI features.

However, what could then happen is, while the overwrite management packs do remove the Azure Stack HCI resources from appearing in the Windows MP discoveries in the future after they are installed, some of your older inventory could still appear. If you want to clean that data, we recommend using the PowerShell commandlets so that they will be automatically removed from the earlier discovered inventory as well.

Microsoft is aware of two known issues that seem to come up in some cases with the Azure Stack HCI management pack. In some scenarios, in the Base OS MP and sometimes in the cluster MP, the Azure Stack HCI Computers appear in Windows Server Management Packs views. Microsoft knows the root cause of the issue and is working towards fixing it.

The second known issue is that workflows are duplicated on Azure Stack HCI computers. This happens because both the Azure Stack HCI and Windows Server workflows will run on all Azure Stack HCI resources. In some cases, this could create duplicate alerts. Again, this is a known issue that Microsoft is working towards addressing in the next release or update that will come out for the HCI MP.

Q&A

At the end of the Coffee Break we had a Q&A session. Here are some of the audience questions and their answers.

Does the agent have to be installed on all the nodes?

Yes, in the same way that it works for all standard management packs.

Does GM have to be installed inside the cluster? Or can it monitor a cluster that runs on Azure Stack HCI without being inside it?

Yes, it can run without being installed inside the cluster. It would just need to be connected to the cluster.

Will Azure stack HCI MP work with SCOM 2016?

Microsoft has not tested it on SCOM 2016. They have only tested it with 2019 and 2022. You could try it out though and it may work. But at least in terms of support, Microsoft won’t be supporting the HCI MP for SCOM 2016. Maybe it’s time to upgrade.

Will the MPs work for S2D running on server 2019?

We have a separate Windows Server S2D MP and that will work on 2019. If you want to monitor
HCI while you have the Windows Server S2D MP installed on the same SCOM instance, that should work, yes.

How do you discover the cluster?

All the standard discoveries in the MPs exist. There is a standard MP file that contains the definitions and discoveries used for the Azure Stack HCI cluster service management, so Microsoft has taken care of that aspect.

Conclusion

So that was our deep dive into the Azure Stack HCI management pack, with huge thanks to Bhavna Appayya.

Keep an eye out for the next Coffee Break session and join in. Subscribe to SCOMathon for all the updates so you don’t miss all things SCOM.