Monitoring service

The following information shows you how to troubleshoot monitoring service problems.

IBM Spectrum Control shows an event for the discovery and control components for the following actions:
  • When you add a resource.
  • When you make resource status changes.
  • When the resource is not detected.
IBM Spectrum Control provides an alert generation on behalf of the performance monitoring service when the threshold changes and there is a collection failure. The database table snapshot for a certain resourfce type is based on the request that is made to the performance monitoring service.
For general diagnosis of the problem, check these items:
  • Make sure that the alert trigger condition is set up correctly.
  • Make sure that the changes did happen.
  • Check the alerts that are shown on the Alerts page in the IBM Spectrum Control GUI.
For specific diagnosis of the problem, check these items:
  • Locate the Device server trace log and search for possible exceptions that are generated from the monitor service.
  • If no errors and exceptions are found for the monitor service, it means that the monitor service is running without problems. Next, check to see whether some other component is causing the problem such as the Change Detection component.

Common usability problems

These items are some common user errors with the monitoring service:
Why do I not get the correct alert?
Check to see that you configured the correct trigger condition for the correct resource or resource type.
Why is overall performance slow?
Many database activities to form SNMP and Netcool®/OMNIbus events might affect overall performance if too many alerts get created. You can configure only the needed alerts to control the number of alerts.
Why am I losing an event or alert?
An event or alert might be lost because the required data is unavailable in the database or there are network communication problems. After the alert creation fails, the alert for this changed device is not re-created.
Why do I see unexpected alert definitions and unexpected alerts with the keyword Pre-defined in the name?
These alerts are shown when another application is subscribed to receive events from IBM Spectrum Control through the product API.
You can change the setting to hide Pre-defined alert definitions when you use the tpctool command: 
tpctool setdscfg -user user_ID -pwd password -url localhost:9550 
-property APIAlert.Visibility false

Logs and traces

When a problem occurs, you can set the trace level parameter san.eventFactoryTrace.level = INFO when you use the CLI command to get detailed trace information. Make sure that the created FabricAlert object is sent to the Data Server successfully.

The detailed trace information is saved in the following directory: install_directory\data\log\traceTPCDeviceServer.log

Use the service tool and repocopy tool to collect trace information for IBM Spectrum Control and the database.

When an expected alert is not presented, check the following items:
  • The trace log to make sure that the resource changes are reported to the monitor service. For example, search the log for the corresponding API call.
  • The correct alert population routine is used, and any other exception that is encountered during the process.
  • Make sure that the alert object is sent to the Device server.