Return the number of seconds this worker needs hold a lease on the task for while it performs work. For most tasks you can leave this at null, which will give you a default lease (currently 2 hours).

For tasks which may take a very long time to complete, you should return an upper bound on the amount of time the task may require.

Return the maximum number of times this task may be retried before it is considered permanently failed. By default, tasks retry indefinitely. You can throw a PhabricatorWorkerPermanentFailureException to cause an immediate permanent failure.

Return the number of seconds a task should wait after a failure before retrying. For most tasks you can leave this at null, which will give you a short default retry period (currently 60 seconds).

Set this flag to execute scheduled tasks synchronously, in the same process. This is useful for debugging, and otherwise dramatically worse in every way imaginable.

Queue a task to be executed after this one succeeds.

The followup task will be queued only if this task completes cleanly.

Get tasks queued as followups by queueTask().

Schedule any queued tasks, then empty the task queue.

By default, the queue is flushed only if a task succeeds. You can call this method to force the queue to flush before failing (for example, if you are using queues to improve locking behavior).

Awaken tasks that have yielded.

Reschedules the specified tasks if they are currently queued in a yielded, unleased, unretried state so they'll execute sooner. This can let the queue avoid unnecessary waits.

This method does not provide any assurances about when these tasks will execute, or even guarantee that it will have any effect at all.