Maya batch verses Maya batch in Qube

[mekrender]

Hi,

I have a mel script that we use when rendering with maya batch that behaves differently when the job is submitted through Qube.  I'm not clear on how Qube calls Maya batch each frame, and that would help me debug this.  Is there somewhere I can see a breakdown of how Qube calls Maya batch each frame and for the job, etc?

The mel script I'm trying to get to work on Qube has three parts.  The first is A "Pre Render Mel" that builds a temporary Maya project on the machine that is doing the rendering, if one already exists it cleans it out, and then sets the active workspace to this project.  The second is a "Pre Render Frame Mel" that copies any particle cache files needed for the frame to be rendered from the network mounted original project to the local temp project.  The last is a "Post Render Mel" that copies the final renders back to the original project.

When launched through Maya on any of the Workers on the farm the script works fine and though each frame's cache must be copied locally first it is much faster than forcing Maya to access large cache files across the network.  Through Qube though it acts as if the "Pre Render Mel" is being called each frame similar to the "Pre Render Frame Mel".  Is this to be expected even if only one Worker completes the job?

Also is there a mechanism to chunk out frames if using multiple workers so that each Worker does say 10 frames sequentially?

Any help greatly appreciated, as I am new to using Qube!

[jburk]

Check the 'stdout' pane for one of your subjobs; the command line run will be available in the job's stdout.

The difference you're seeing is that when the render is 1 frame long, which is the case for a single frame in batchrender mode on the farm with a chunksize of 1, then preRender and preFrame mean the same thing.  "preRender" is "do this before you render any frames", and "preFrame" is "do this before every frame".  So the preRender is called every time when the chunk running on the worker is only 1 frame long.

Even when 1 worker is processing the job, if the chunk only contains 1 frame, each call to "maya -batch" is only 1 frame long, and the worker has no knowledge that it ran a previous frame.

You can specify the chunk size via the 'execution' control of the 'Range' section, or specify the number of partitions, in which case Qube will divide your frame range into n equal-sized chunks.

[mekrender]

It all makes sense, and I've got the mel script working through Qube now.  I haven't gotten to the chunking yet but I think it should go well.  Thanks for your help!