11
HTTP Server Event Tests

The HTTP Server event tests are provided for lights-out event monitoring and problem detection of the HTTP server.

Aside from the basic up/down event tests provided for all services administered in the Oracle Enterprise Manager console, there exists a library of event tests specific to the HTTP Server. The types of event tests associated with the HTTP Server are:

• UpDown event test
• Threshold event tests

The HTTP Server UpDown event test checks whether the Apache HTTP server being monitored is running.

The threshold event tests are triggered by user-specified metrics. The full descriptions of the individual event tests follow. The event tests are in alphabetical order.

Summary of HTTP Server Event Tests

The following table lists the HTTP Server event tests. The full descriptions of the individual event tests follow the table. The event tests are in alphabetical order.

Table 11-1 Applications Fault Management Event Tests

Event Test	Description
Bytes per Second	This event test monitors the number of bytes transferred by the HTTP Server per second. The rate is calculated depending on the frequency of the event. For example, if the event was registered with a frequency of 100 seconds, the event will calculate the rate by dividing the number of bytes served in that period by 100.
Requests per Second	This event test monitors the number of requests made to the HTTP Server per second. The rate is calculated depending on the frequency of the event. For example, if the event was registered with a frequency of 100 seconds, the event will calculate the rate by dividing the number of requests in that period by 100.
Total Servers	This event test monitors the number of servers that the HTTP Server has spawned to handle the incoming requests. The HTTP Server increases the number of servers when there is a greater number of requests to handle and reduces the number of servers when there are fewer requests. This action maximizes the resource utilization on the system.

Descriptions of HTTP Server Event Tests

The HTTP Server event tests are listed in alphabetical order.

Bytes per Second

This event test monitors the number of bytes transferred by the HTTP Server per second. The rate is calculated depending on the frequency of the event. For example, if the event was registered with a frequency of 100 seconds, the event will calculate the rate by dividing the number of bytes served in that period by 100.

Parameters

• Critical Threshold: Number of request per seconds that will trigger the Critical alert event. Default is 10000.
• Warning Threshold: Number of request per seconds that will trigger the Warning alert event. Default is 2000.
• Number of occurrences: Default is 3.

Output

None

Default Frequency

60 seconds

User Action

Check the access logs to determine a cause. If the condition persists, tune the system to handle a larger data output.

Requests per Second

This event test monitors the number of requests made to the HTTP Server per second. The rate is calculated depending on the frequency of the event. For example, if the event was registered with a frequency of 100 seconds, the event will calculate the rate by dividing the number of requests in that period by 100.

Parameters

• Critical Threshold: Number of requests per second that will trigger the Critical alert event. Default is 50.
• Warning Threshold: Number of requests per second that will trigger the Warning alert event. Default is 20.
• Number of occurrences: Default is 3.

Output

None

Default Frequency

60 seconds

User Action

Check the access logs to determine a cause. If the condition persists, tune the system to handle a larger number of requests.

Total Servers

This event test monitors the number of servers that the HTTP Server has spawned to handle the incoming requests. The HTTP Server increases the number of servers when there is a greater number of requests to handle and reduces the number of servers when there are fewer requests. This action maximizes the resource utilization on the system.

Note: On Windows NT, servers represent the number of threads while on UNIX systems a server correlates to a process.

Parameters

• Critical Threshold: Number of servers that will trigger the Critical alert event. Default is 80.
• Warning Threshold: Number of servers that will trigger the Warning alert event. Default is 30.
• Number of occurrences: Default is 3.

Output

None

Default Frequency

60 seconds

User Action

Check the access logs to determine a cause. If the condition persists, tune the Web Server to use resources more efficiently or upgrade the hardware.