AppEngine monitoring

The following information describes how the failover service monitors the redundant engines.

In general, an engine has the following states:

• Start Up: Measured as the time required for all engine objects to be created, initialized and started.

• Execution: Measured as the time required for all engine objects to be executed in one scan cycle.

• Shut Down: Measured as the time required for all engine objects to be stopped.

The following parameters determine how much time the engine can be unresponsive during each of the above states.

Start Up and Shut Down

If you shut down the platform from the Operations Control Management Console and the platform shuts down improperly, you may need to increase the amount of system RAM.

Execution

The EngineFailureTimeout attribute determines how long that engine has to inform the bootstrap that it is executing. If the timeout period elapses, a failover to the backup engine occurs.

Setting this attribute value too low causes the redundant partner to overreact when CPU usage is high. Setting the value too high can delay a failover to the backup engine.

The timeout period should be set to a long enough interval to accommodate the completion of any engine-related actions that might prevent or delay the bootstrap notification (for example, cascade deployment of objects with a large number of scripts). The minimum timeout the system allows at run time is 55,000 ms (default value is 30,000 ms). If a shorter timeout is configured, the configured value is ignored and the object will use the 55,000 ms minimum when it is deployed.