Setting an Alert for a Performance Metric

For some services such as Oracle Cloud Infrastructure Compute Classic, Oracle Java as a Service (Oracle JaaS) or Oracle Database as a Service (Oracle DBaaS), service administrators can configure rules to generate alerts when metrics exceed or are under specified thresholds for a specific time period. Service administrators will receive the alert notifications.

A single metric is monitored over a specific time period vis-a-vis a defined threshold value. An alert is sent to the service administrators when there is a change in the metric behavior for a sustained period of time that is defined by you. The threshold state can be either true (satisfied) or false (unsatisfied) depending on the change in the metric behavior.

An alert is sent only when the threshold state changes as follows:

• Metric is not within the defined threshold for a sustained period of time and the threshold state changes from either true to false or vice versa.

• Metric is within the defined threshold for a sustained period of time and the threshold state changes from either true to false or vice versa.

To create alert rules, do the following:

1. Sign in to Infrastructure Classic Console.
2. Navigate to the appropriate service listing.
3. Click the service name to open the details page for the service.
   The Overview page is in focus.
4. Click the Monitoring Metrics tab and then click Show Thresholds/Alerts.
5. In the Alert Rules page, click Add.
6. In the Create Alert Rule dialog box, specify the following:
   1. Channel: Email. This is selected by default and is read-only. You’ll receive alert notifications via email.
   2. Instance: Select an instance from the list to monitor.
   3. Criteria: Select the metric to monitor. Select the average, the operators (>, <, <=, >=) and then enter an amount in the box. This is the threshold value. The amount must not be less than zero.
   4. Duration: Select the time period for which you want to monitor the metric.
   5. Period Count to Satisfy: Enter the number of consecutive times that the metric must meet the criteria you specified above to trigger an alert. The count must be greater than or equal to 1.
   6. Period Count to Unsatisfy: Enter the number of consecutive times that the metric must be within the threshold limit to trigger an alert. The count must be greater than or equal to 1.
   7. Click Done.

Example 5-2 Configuring an Alert Rule to Monitor CPU Usage

If you specify the following criteria to monitor the CPU usage levels:

• Criteria = CPU Percentage (%) Average >=80

• Duration = 10 minutes

• Period Count to Satisfy = 3

• Period Count to Unsatisfy = 2

Then,

An alert is sent once when the metric:

• Crosses the specified threshold during 3 consecutive 10-minute periods and the threshold state changes from false to true.

• Remains below or is within the threshold limits for 2 consecutive 10-minute periods (after a corrective action is taken) and the threshold state changes again from true to false.

Another alert is not sent:

• If a metric remains above the threshold for more than 3 consecutive 10-minute periods and there is no change in the threshold state.

• If a metric remains below or is within the threshold for more than 2 consecutive 10-minute periods and there is no change in the threshold state.

The table below summarizes when an alert is triggered based on the specified criteria.

Time Period	Metric (CPU Usage % >=80)	Threshold Value (Above/Below/Equal)	Threshold State (True/False)	Alert Sent (Y/N)
1:00	Threshold Created	0	False	N
1:10	50	Below	False	N
1:20	80	Equal	True	N
1:30	85	Above	True	N
1:40	90	Above	True	Y (Avg>=80 for 3 consecutive 10-minute periods)
1:50	81	Above	True	N
2:00	60	Below	False	N
2:10	50	Below	False	Y (Avg<80 for 2 consecutive 10-minute periods)
2:20	80	Equal	True	N
2:30	60	Below	False	N
2:40	81	Above	True	N

Here is an illustration of the above example:

The alert rules are listed in the Alert Rules page. To delete a rule, click Remove from the Action menu.