Tasks: Reducing the number of SKIPPED tasks¶

For tasks with a user-specified schedule, the tasks run as follows:

Before the change:

A task runs based on the user-specified schedule, for example, every 1 minute. The task completes with a SKIPPED state if there is no new data in the stream within the specified time interval (for example, 1 minute). In this case, the SCHEDULED_FROM column of the TASK_HISTORY view is SCHEDULE.

After the change:

A task still follows the user-specified schedule but only executes when there is new data in the stream. For example, a user has a task with a schedule to run every minute (SCHEDULE = ‘1 m’) and a stream_has_data WHEN condition. The task checks the stream a minute after the last task start time, and only executes when there is new data. In this case, the SCHEDULED_FROM column of the TASK_HISTORY view is TRIGGER. The task is executed approximately every 12 hours to prevent the stream from going stale.

If you set up monitoring or alerts to check your task run states, we recommend adjusting your monitoring to account for this behavior change because the tasks in the SKIPPED state are significantly reduced.

Ref: 1710