@Kevin, yes, there is timeout mechanism to avoid being blocked by a record that is in processing by a dead worker.
BTW, in the patch I am working on I chose a less intrusive approach avoiding background workers, by just on-the-fly ordering enforcement in the same sync API call context. The task queue is still implemented in DB in the way we discussed above.
@Kevin, yes, there is timeout mechanism to avoid being blocked by a record that is in processing by a dead worker.
BTW, in the patch I am working on I chose a less intrusive approach avoiding background workers, by just on-the-fly ordering enforcement in the same sync API call context. The task queue is still implemented in DB in the way we discussed above.