Enabling Performance Threshold Monitors
It is very difficult to set performance threshold monitors that are suitable for most environments; therefore, most performance threshold rules are initially disabled, and the threshold values are configured so that they do not trigger state changes. The following table lists the performance threshold monitors that we recommend you enable. Before you enable a performance threshold monitor, you should baseline the relevant performance counters, and then apply the appropriate overrides to define and enable a suitable threshold for your environment.
| Monitor Name | Target | Elements to be Configured |
|---|---|---|
Cluster CPU Usage monitor |
Windows HPC Server 2008 R2 active head node |
Low threshold – The threshold value for the warning state. High threshold – The threshold value for the critical state. |
Cluster Disk Throughput monitor |
Windows HPC Server 2008 R2 active head node |
Low threshold – The threshold value for the warning state. High threshold – The threshold value for the critical state. |
Cluster Network Usage monitor |
Windows HPC Server 2008 R2 active head node |
Low threshold – The threshold value for the warning state. High threshold – The threshold value for the critical state. |
CPU Utilization monitor |
Separate monitors are targeted to the Windows HPC Server 2008 R2 head node, compute nodes, broker nodes, and workstation nodes |
Threshold – The threshold value for the warning state. |
Memory Utilization monitor |
Separate monitors are targeted to the Windows HPC Server 2008 R2 head node, compute nodes, broker nodes, and workstation nodes |
Threshold – The threshold value for the warning state. |
Daily Job Queue Time monitor |
Windows HPC Server 2008 R2 job scheduler |
Low threshold – The threshold value for the warning state. High threshold – The threshold value for the critical state. |