Hi,
If you have 1 slave with say, 2 concurrent tasks being processed and one of the tasks crashes out. If you have to end task the slave as it becomes un-responsive, then even when you set that particular slave to “set to offline” in the monitor, the task render status displays just a single “,”, indicating the separation between tasks. However, nothing is still running on this slave. Not sure if much can be done in this case, but thought it worth mentioning. Perhaps we should be discussing why the slave doesn’t recover from one of its concurrent tasks taking a dive and the other tasks should be thread safe?
Mike
Hey Mike,
When a slave is manually set to offline, the only thing updated in the slave info file is the status. That’s why the render status doesn’t change.
But you’re right, the more important issue is why the render thread responsible for the crashed task doesn’t exit. I’m pretty sure Deadline has checks to see if the thread has died, so my guess is that it’s deadlocked.
You wouldn’t happen to have a slave log from a session where this issue occurred, do you? I’m just curious if there would be any useful information there to help us figure out what’s going on.
Cheers,
- Ryan
I’ll send the logs as a support ticket as it contains some sensitive info.
Mike