Configuring the initial sensor data poll delay for long periods

This section provides information on how to configure the initial sensor data poll delay for long periods.

Some performance monitoring sensors might invoke heavy-load tasks like executing the tsdf command. The scheduled period of such sensors can be as high as once per hour or once per day, and the heavy-load tasks can force the Performance Monitoring Tool (PMT) to change its configuration frequently. When the PMT configuration changes frequently, the system triggers the sensors to restart at every instance. Once the sensors restart, they might continue to run more often than the defined period value. This causes system overload issues, which can tamper the overall system performance.

In order to mitigate the impact of system overload, the PMT supports a Boolean flag, delay_initial_poll. This flag delays the initial invocation of sensors depending on its period value and is set to true by default. The metrics of those sensors, where the delay_initial_poll is set to true, experience a delay and are not immediately available after a sensor's restart. However, if the delayed invocation of sensor is not preferred, then the delay_initial_poll flag is set to false.

If the delay_initial_poll flag is set to true, then it delays the very first poll of sensors as follows:

• If the period is greater than or equal to 24 hours, then there is a delay of three hours until the first poll of the sensor is done.

• If the period is greater than or equal to 12 hours, then there is a delay of two hours until the first poll of the sensor is done.

• If the period is greater than or equal to 3 hours, then there is a delay of one hour until the first poll of the sensor is done.

• If the period is greater than or equal to 1 hour, then there is a delay of one-third of the period until the first poll of the sensor is done.

Here, period refers to the configured time period for a sensor.