AWS Thinkbox Discussion Forums

Job/task re-queueing/suspension is a little lazy

We currently use Rush as our queue manager. While it has more than its fair share of detriments, one of the nice things is that, because it simply executes commands in a managed shell, operations like killing or re-queueing frames are quick and merciless.

The current behavior I’m seeing in Deadline 6 is that, if a running job is re-queued, the slaves don’t notice for a good 10 to 15 seconds. So my question is, is it too late in the game to request that this be either tweaked to be more responsive, or added as a Slave setting in the Repository Options?

Also, right-clicking a task currently presents no options other than Copy… I’m assuming populating this menu is still on the to-do list?

Thanks,
-Nathan

We’re going to reduce the interval at which the slave checks the state of its current task. It was set to 30 seconds, but in beta 4, it will be a random value between 5-10 seconds. That should make it more responsive. We may expose this interval as a Repository option, but for now we’re keeping it hidden.

You’re right, the task right-click menu still needs to be populated, and it’s on the todo list.

Cheers,

  • Ryan

Perfect, thanks for the info.

Privacy | Site terms | Cookie preferences