Forum Discussion

Ryan_Rippo's avatar
7 years ago

SQL Cluster Alerting

Currently we are managing a customer with multiple SQL Clusters.  Currently we don't have a way to effectively monitor against the cluster.  LM throws alarms on the passive node in the cluster because services are not started.  That is obviously by design because it is the passive node.  We are trying to figure out a way to monitor the cluster as a whole and not the individual servers in the cluster.  Does anyone have similar issues or any experience with this?

Thanks!

Ryan

8 Replies

Replies have been turned off for this discussion
  • Ryan,

    I have a group for cluster nodes, and clusters. I add in the clusters using their DNS name so even if it fails over to another node, it still shows as up. I then apply the property ClusterNode to the nodes and WindowsCluster to the cluster name. I have modified the datasources for monitoring windows cluster resources, groups, and nodes to only apply to devices with the category of WindowsCluster. On my groups, I disable all datasources that may be duplicated between them This works perfectly for all clustering, exchange and sql, but not for Always-On clusters as services need to be running on both sides. To correct this, I manually add the sql monitoring to each of the nodes.

    This may sound a little confusing, but it was pretty straightforward to setup. Happy to help if this sounds like what you are looking for. It does cost an extra device per cluster though...

  • Sarah_Terry's avatar
    Sarah_Terry
    Icon for Product Manager rankProduct Manager

    Hi Ryan,

    We're currently working on a feature which will provide more service-oriented monitoring / alerting capabilities, which should work well for your use case. We don't have any specifics re timing yet, but we'll keep you updated as we get closer.

    Thanks,

    Sarah

  • @Ryan Rippo I created some tickets to investigate the best way to tackle your cluster monitoring.

    For the time being, have you tried what @Justin suggested ? Its pretty straight forward and accomplishes what you are intending to do.

    If you have any questions, feel free to reach out.

  • 20 hours ago, Justin said:

    Ryan,

    I have a group for cluster nodes, and clusters. I add in the clusters using their DNS name so even if it fails over to another node, it still shows as up. I then apply the property ClusterNode to the nodes and WindowsCluster to the cluster name. I have modified the datasources for monitoring windows cluster resources, groups, and nodes to only apply to devices with the category of WindowsCluster. On my groups, I disable all datasources that may be duplicated between them This works perfectly for all clustering, exchange and sql, but not for Always-On clusters as services need to be running on both sides. To correct this, I manually add the sql monitoring to each of the nodes.

    This may sound a little confusing, but it was pretty straightforward to setup. Happy to help if this sounds like what you are looking for. It does cost an extra device per cluster though...

     

    Thanks @Justin - Great feedback and commentary!

  • That said, I do not see any support documentation around this

    "ClusterNode"

  • 5 minutes ago, WillFulmer said:

    That said, I do not see any support documentation around this

    "ClusterNode"

     

    @WillFulmer once we conclude our research, we will post up support documentation. 

    Thank you

  • On 12/21/2017 at 4:50 PM, Justin said:

    Ryan,

    I have a group for cluster nodes, and clusters. I add in the clusters using their DNS name so even if it fails over to another node, it still shows as up. I then apply the property ClusterNode to the nodes and WindowsCluster to the cluster name. I have modified the datasources for monitoring windows cluster resources, groups, and nodes to only apply to devices with the category of WindowsCluster. On my groups, I disable all datasources that may be duplicated between them This works perfectly for all clustering, exchange and sql, but not for Always-On clusters as services need to be running on both sides. To correct this, I manually add the sql monitoring to each of the nodes.

    This may sound a little confusing, but it was pretty straightforward to setup. Happy to help if this sounds like what you are looking for. It does cost an extra device per cluster though...

    And to this note: 'It does cost an extra device per cluster though...'

    Monitoring say a 2 node Windows Server File Cluster = 4 devices
    - Node1
    - Node2
    - Failover Cluster Name
    - Clustered Resource Name (File Server, SQL, etc)

     

     

  • It is not necessary to add in the cluster resources as devices, they are monitored instance under the cluster named device.