Forum Discussion

DavidRa's avatar
5 months ago
Solved

Azure Stack HCI resources don't have storage, memory, disk or cluster metrics

We have a customer with an Azure Stack HCI cluster deployed a few months ago. For those not familiar, this is basically a customised Windows Server core environment that runs Hyper-V VMs and some Azure-specific workloads on-premises.

The virtualised workloads are all added as resources using a locally-deployed collector (on the Windows jumphost, if it matters) and they all show CPU, Disks, Interfaces, Processes … everything you’d expect for Windows hosts.

We’ve added the two nodes as Resources, but we don’t see any detailed metrics - only Host Status (DNS), HTTP and Ping. I also added the FQDN for the cluster management point / VNN, and it has the same minimal detail as the individual cluster nodes.

There are quite a few valid/correct properties recorded for the systems - some, for example:

  • system.domain (customer AD domain)
  • system.ips (all IP addresses for all interfaces)
  • system.model (correctly identifies vendor and server model, presumably from WMI)
  • system.sysinfo (“Microsoft Azure Stack HCI”)
  • system.sysname (hostname)
  • system.systemtype (“x64-based PC”)

Is there something else I need to do to have this system monitored? I’m rushing because we nearly had a CSV run out of space - we thought it was monitored, and we were wrong.

  • So an update, adding the string “Windows” to the category did accomplish most of what I needed. A colleague pointed out that we could also override the sysinfo property, but that seems like a surefire way to shoot ourselves in the foot later.

    So the last remaining question would be “how do I flag a request to the LM team to get ‘Azure Stack HCI’ recognised as Windows automatically”? I can’t be the only one with an AzStack environment to monitor.

    As what @Joe Williams have suggested, while in LogicMonitor portal, click on Support menu > Contact Support > Feedback.

    Our Product team is constantly monitoring customer’s feedback and reassess market demand for this  to be added to “official” LogicModules. 

9 Replies

  • Lim's avatar
    Lim
    Icon for LM Champion rankLM Champion

    @DavidRa To solve the problem of isWindows(), you can consider creating a PropertySource "addCategory_Microsoft Azure Stack HCI" with the following:

    Applies To
    system.sysinfo == "Microsoft Azure Stack HCI" && !hasCategory("Azure Stack HCI") 

    PropertySource Script (Embedded Groovy Script)
    println "system.categories=Windows,Microsoft Azure Stack HCI"

  • @DavidRa While in LogicMonitor, click on Support, then there is a Contact Support option. Click that, then Feedback. That is the way to have items like that sent over.

  • Lim's avatar
    Lim
    Icon for LM Champion rankLM Champion

    So an update, adding the string “Windows” to the category did accomplish most of what I needed. A colleague pointed out that we could also override the sysinfo property, but that seems like a surefire way to shoot ourselves in the foot later.

    So the last remaining question would be “how do I flag a request to the LM team to get ‘Azure Stack HCI’ recognised as Windows automatically”? I can’t be the only one with an AzStack environment to monitor.

    As what @Joe Williams have suggested, while in LogicMonitor portal, click on Support menu > Contact Support > Feedback.

    Our Product team is constantly monitoring customer’s feedback and reassess market demand for this  to be added to “official” LogicModules. 

  • After quite a few hours, I think I may have found something relevant. However, I do not work with LM much so I could be wrong.

    It seems to me that a lot of the data gathering revolves around modules that are filtered to devices for which “isWindows()” is true. And my reading of the documentation suggests that this is only if sysinfo contains “windows” or “hyper-v”, or categories contains either of those terms.

    If so, can I assign windows to the categories manually (would it be automatically removed)? Or do I update the AppliesTo for everything (that seems to be very bad at any reasonable scale)? Is there something else I can do (like get the isWindows() function updated to include Azure Stack HCI)?

  • So an update, adding the string “Windows” to the category did accomplish most of what I needed. A colleague pointed out that we could also override the sysinfo property, but that seems like a surefire way to shoot ourselves in the foot later.

    So the last remaining question would be “how do I flag a request to the LM team to get ‘Azure Stack HCI’ recognised as Windows automatically”? I can’t be the only one with an AzStack environment to monitor.

  • I’ll keep an eye out for the feature request.  After you’ve made this adjustment, what else, Azure Stack related, would you like to be able to monitor that the Windows modules don’t cover?

    Thanks

  • I ventured into this myself, and found I could simply just addCategory HyperV and Windows by copyin the built in funtions, and LM just picks up my Azure Stack HCI cluster and starts monitoring like any other HyperV with Windows server, so all cluster checks, windows monitoring, storage, cpu, mem, etc. is up and running, without the need for changing logicmodules. 

    I do belive that built in isWindows should be updated to include a match when seeing Azure Stack HCI and seeing it as just another windows OS. 

  • I haven’t tried this but I wonder if you can override the isWindows() via AppliesTo Functions. The documentation doesn’t seem to say what happens if you attempt that or if it’s blocked.

    I don’t think there is any functionality to remove categories from devices, but I also don’t think isWindows() will look at categories to detection. But if there are some DataSources that do look at categories, you should be able to add them manually if you want.

  • I don’t have one of these to experiment with, but it might be worth looking at this article. (https://www.logicmonitor.com/support/monitoring/applications-databases/windows-server-failover-cluster-on-sql-server-monitoring) If the azure HCI cluster uses the same kind of cluster mechanics as Windows Server Failover Clusters, there are some system properties here which might allow for the application of WSFC logicmodules. I don’t know if they will work, but it’s worth checking out. (If it doesn’t work, just remove the properties)