Distributed frame rendering for maya/mr

[JasonA]

Hi all,

Im trying to render out a 15k single image as a distributed frame but when i submit the job I cannot get Qube to distribute the render across our available nodes, it always attempts to render it only on 1 blade.  I couldnt find specific instructions or details on this on the doc pdf, can anyone point me in the right direction?

[jburk]

Did you want to take advantage of mentalray satellite?  This is currently the only single-frame distributed renderer we support.

You must already have mentalray satellite correctly installed and running at your site for this to work; instructions on how to install and configure mray satellite are outside the scope of this forum.

If you do have mray satellite working, you can use the 'mray satellite' control in the 'submit maya' submission UI ("Renderer" section), which is available either from inside Maya in Qube->Submit Render Job, or from the QubeGuI under Submit->Maya Job.

[JasonA]

We do have maya unlimited with satellite installed.  When i submit a render through the mayabatch (mr), i see in the standard ouptut window that it says it got 8 satellite nodes, so we believe that satellite is correctly installed.

I tried submitting the job using Maya Job, and I specified 8 CPUs, and down near the bottom also set the option for satelite 8 cpus.  when i run the job, it fails before rendering and each worker reports back the same error:

Can't locate qb.pm in @INC (@INC contains: /usr/local/pfx/jobtypes /usr/local/pfx/qube//types /usr/local/pfx/qube//api/perl/qb/blib/arch/auto/qb /usr/local/pfx/qube//api/perl /usr/lib64/perl5/site_perl/5.8.8/x86_64-linux-thread-multi /usr/lib64/perl5/site_perl/5.8.7/x86_64-linux-thread-multi /usr/lib64/perl5/site_perl/5.8.6/x86_64-linux-thread-multi /usr/lib64/perl5/site_perl/5.8.5/x86_64-linux-thread-multi /usr/lib/perl5/site_perl/5.8.8 /usr/lib/perl5/site_perl/5.8.7 /usr/lib/perl5/site_perl/5.8.6 /usr/lib/perl5/site_perl/5.8.5 /usr/lib/perl5/site_perl /usr/lib64/perl5/vendor_perl/5.8.8/x86_64-linux-thread-multi /usr/lib64/perl5/vendor_perl/5.8.7/x86_64-linux-thread-multi /usr/lib64/perl5/vendor_perl/5.8.6/x86_64-linux-thread-multi /usr/lib64/perl5/vendor_perl/5.8.5/x86_64-linux-thread-multi /usr/lib/perl5/vendor_perl/5.8.8 /usr/lib/perl5/vendor_perl/5.8.7 /usr/lib/perl5/vendor_perl/5.8.6 /usr/lib/perl5/vendor_perl/5.8.5 /usr/lib/perl5/vendor_perl /usr/lib64/perl5/5.8.8/x86_64-linux-thread-multi /usr/lib/perl5/5.8.8 .) at /var/spool/qube/job/0/419/419_2.pm line 5.
BEGIN failed--compilation aborted at /var/spool/qube/job/0/419/419_2.pm line 5.



 

[jburk]

do you have the 'export_environment' flag set for the job?  That can cause the perl includes to get messed up and generate the "Can't locate qb.pm..." message.

[jburk]

Can you post the output from 'file /usr/local/pfx/qube/api/perl/qb.so'?  You may have a 32-bit qube install mixed with a 64-bit perl.

[JasonA]

do you have the 'export_environment' flag set for the job?  That can cause the perl includes to get messed up and generate the "Can't locate qb.pm..." message.

I didnt see that flag in the Maya Job rollout, so im not sure that this flag is set.

Im going to see if I can egt that ouput and post it, brb

[dball]

We use Qube to distribute the rendering of large print stills in Mental Ray standalone. Our method it to put the number of render nodes we require into the CPU's box. Then check the Satellite box and put the correct port number relevant to the version of mental ray you are running, ie. 7108 for Maya 2008. Most importantly put host.processors=1+ into the reservations box, this then uses all the cores of that render node.

However I wouldn't assume your satellite rendering is setup correctly based on the '8 satellite nodes' output. Check by running the render -mr -v 6 flags and using a maya.rayhosts file. You will see in the output if it's successfully connected to the slave machines.

Once satellite is running correctly (outside of Qube) the above procedure will work correctly for distributing renders.

Hope that helps.