Basic steps for installing and configuring UIM probes on the nodes of a cluster in order to monitor both shared and local resources.
The cluster probe enables failover support for profiles in the supported monitoring probes. Profiles of supported monitoring probes can be associated with different shared resource groups and follow that group even if it is transferred to another cluster node.
The cluster probe updates the changes that are made to the profile setup of the other monitoring probes also installed on the nodes of a cluster. When any monitoring probe profile is updated, the changes must be implemented on the probe that runs on the node which is in control of the associated resource group. The cluster probe automatically updates most of the changes of the monitoring probe profiles. However, general probe parameters must be updated manually on each of the nodes in the cluster.
A shared resource group is visible under the active cluster node in the navigation pane of the cluster probe's Configure GUI for the node that currently controls the shared resources. When one of the cluster nodes fails, the control of all of the shared resources is transferred to another functioning node. The shared resource group is then visible under the new functioning node in the navigation pane of the cluster probe's Configure GUI.
The shared resource group failover alarm does not clear until the resource group moves back to the original cluster node.
You can view the cluster node status when you select the cluster. However, you can view the related group metrics only under that particular cluster node to which the shared resource group belongs.
Consider the following points for creating monitoring profiles for shared resources.
- The cluster probe must be deployed on all nodes of the cluster
- All nodes in the cluster must have the same monitoring probes (and versions) with the same general configuration.
- UIM 8.51 or higher
- robot (controller probe): 7.93 or higher
- cluster: 3.70 or higher
1. Install the UIM robot on all cluster nodes
2. Install the cluster probe on all cluster nodes
3. Configure the cluster probe on the active node in the cluster
4. Install all UIM monitoring probes that you need on all of the nodes in the cluster (for example, cdm, ntservices, sqlserver, etc)
5. Configure each of the installed UIM monitoring probes for monitoring any resources that are not part of any of the shared resource groups on the monitoring probes on each node of the cluster. For example, configure memory, CPU, and C:\ drive (typically not part of a shared resource group) monitoring profiles for the cdm probe on all of the cluster nodes. Save all changes and restart the monitoring probe.
6. Configure the UIM monitoring probes to monitor the resources that are part of a shared resource group only on the node of the cluster where the corresponding shared resource group is running. For example, for the cdm probe, configure disk monitoring for any drive that is part of a shared resource which will be transferred to another node in the cluster when the currently active node fails over. Save all changes and restart the monitoring probe.
7. From the cluster probe's Configure GUI, create monitoring profiles for each of the shared resource monitoring profiles defined in each monitoring probe. You configure these cluster monitoring profiles on the cluster probe residing on the cluster node that is currently in control of the shared resources. These profiles are automatically uploaded to the other nodes when they gain control of the resource group.
Follow these steps:
a. Right-click on the Resource group where you want to create the profile and select New Profile. The New Monitor Resource dialog appears.
b. Specify the following information:
- Probe: specifies the monitoring probe name that you want to associate with the resource group. The description and configured profiles of the monitored probe are listed in the Description and Profile fields.
- Profile: enables you to select a monitoring profile of the defined probe, where the failover support is monitored.
c. Click OK
8. Once all shared resource monitoring profiles are created, save the changes and restart the cluster probe.
Notes on cdm deployment in a cluster
In general, when using the cdm probe in conjunction with the cluster probe, the shared disk profiles are set up initially through the cdm probe and THEN the cluster probe is configured to ensure that the shared cluster disk profiles are successfully activated on the new cluster node on a failover. Upon failover or failback, on the ACTIVE node, you should expect that Local and Cluster drives/filesystems persist - and are properly selected and monitored.
As of cdm v6.50, CDM discovers all the cluster disks automatically under their respective groups, whenever the cluster probe is configured AFTER the deployment of the cdm probe.
Note that moving forward, any new drives added on a given node must not only be 'discovered' by cdm but also 'selected/set to active' so everything works as expected and alarms are generated.
There is a new feature in cdm available in cdm 6.60HF1 (also announced in the release notes for cdm 6.70), where newly added disks are not only discovered but also automatically monitored (selected/active). This will be a new option for all cdm disks.
cluster with cdm deployment Steps
If the user wants to use the cdm probe with the cluster probe, note that the following steps should be followed:
- Deploy the cluster probe on all the nodes
- Configure the cluster probe on one of the nodes
- After the cluster probe is synchronized on all of the nodes, deploy the cdm probe on all of the cluster nodes
- Configure cdm probe on all of the cluster nodes (if needed)
If the cdm probe was already deployed before the deployment of the cluster probe, then the following steps should be followed:
- Deploy the cluster probe on all of the nodes
- Configure the cluster probe on one of the nodes
- After the cluster probe is synchronized on all the nodes, DO a cold start-> Deactivate/Activate the cdm probe on all of the nodes.
- Configure the cdm probe on all of the cluster nodes (if needed)