[Delineator]

So I figure this is a long shot, but we are having a problem that stems from the basic functionality of how the network renderings work. As of right now, we have a 5 node farm, and my personal computer is running the manager and monitor.

Right now, people throughout our office submit jobs (mostly from SketchUp) to the farm. However, we've modified it so that when they submit a network render job their own render node does not automatically start. However, they are never actually "submitting" a job to the network, instead, their monitor automatically launches and submits a job to my manager. However, when queued up, that job is dependent on the monitor remaining open on the computer that submitted the job.

This seems inherently flawed, as that works should be off loaded so that computer can be shut down, restarted, etc. Whats the point of having submitting a job if each submitting computer and Network monitor has to remain up while their job renders?

Beyond a 3rd party queuing suite, is there a work around to this? Something similar to a "watch folder" or something along those lines?

[Bubbaloo]

What exactly happens when the monitor is closed (on the node that submitted the job)? Does the job disappear?

[Delineator]

It hangs. It doesn't disappear because MY monitor is always up (so it sees the job), but if the monitor is closed before the job completes OR even worse, before dependencies are sent (if there are other jobs in queue ahead of it), it will hang the whole farm, which is not ideal.

I've tried messing around with Deadline and maybe I'll give Qube a try, but on the surface it would seem neither offer super tight integration with Maxwell.

I really should learn scripting, there are so many thing I want to do but feel so restricted (for example a "watch folder" that automatically submits jobs to the network farm when MXS files are added)