3.3. Predefined Incidents and Problems

The Enterprise Manager Plug-in for Oracle VDI adds thresholds for a lot of the metrics it collects. The default severity level for predefined Oracle VDI incidents is Critical, with the exception of system service failures, which trigger a Fatal event. The plug-in comes with event-driven problem prevention and detection. The predefined incidents and problems eliminate a lot of initial configuration time. The administrator can customize all metric thresholds and events to suit the requirements of the Oracle VDI installation, as described in "Collection Intervals" in Section 4.3, “Metrics Collection”.

The plug-in provides predefined incidents and problems for all Oracle VDI targets, grouped in the following categories:

Table 3.2. Categories of Predefined Incidents and Problems

Category

Incidents and Problems

Availability

  • Critical system services down

  • Monitoring targets down

Utilization

  • Host or storage memory utilization above 90%

  • Host or storage top network utilization (read or write) above 90%

  • Host or storage file system utilization above 90%

Load

  • Host or storage top CPU utilization above 90%

  • Host or storage disk busy value above 90%

Error

  • One or more desktops unresponsive or with machine state error

  • The number of critical sessions for a host is 5 or greater


For more information about incident management in Oracle Enterprise Manager, and for instructions to customize events and configure notifications, refer to the Oracle Enterprise Manager Cloud Control Administrator's Guide, Chapter 3: "Using Incident Management".

The default threshold for hardware-related events is set to 90%. To avoid false positives, the default setting requires that events occur twice before an alert is sent. Events related to system service availability generate alerts immediately. This behavior can be customized entirely by the Oracle Enterprise Manager administrator, as described in "Collection Intervals" in Section 4.3, “Metrics Collection”.

If a value changes after triggering an event, no new incident is reported. The original alert remains unchanged and the user can drill down into the recently updated values.