Reviewing service statistics

All services publish a set of statistics that can used to monitor the health of the system. They include the following for all services:

• Log statistics related to errors, notices, warnings, statistics, information and diagnostics logged. An event template on the number of errors logged in the last period can be useful to raise awareness that the service has encountered some issue that might need to be resolved.
• Pool statistics related to message and object memory pools. This information is highly technical and useful for troubleshooting by support.
• Process statistics related to CPU and memory usage. This information can be useful for determining if your services machine is handling your monitoring load.

The following is applicable to the Application Service. This information can be useful for determining how policies are performing in your environment.

• Policy Evaluation Stats
• Policy Executor Stats

By default the following occurs:

• The statistics are published every five minutes.
• The statistics are automatically associated with the SAMPLE history template for the published statistic's type definition unless it is found they are already associated to a history template.
• The statistics are published under the host named ServiceSet.

These defaults can be changed in services.cfg, as shown below.

[Stats_Service]
 service_stats_history_template=SAMPLE
 service_stats_host=ServiceSet
 seconds_to_report_stats=300

If service_stats_history_template has an empty value no history template is associated. It is recommended that history is collected on these statistics to assist in troubleshooting any performance related issues.
Chapter 9