Page 1 of 1

Random MR Node failure in MR v03

Posted: Tue Jun 03, 2014 10:54 pm
by animation_man
I am doing a progressive SL animation render on a networking of 6 different Mac computers. The fastest computer I am using did about 8 or 12 files just fine then it got stuck on a file while all my other computers did another few hundred frames. Because of this the rest of my Progressive render stopped and would not start the next Sample Level, so regretably I have to start all over. Any idea what would cause this, and is there anything I can do to prevent this in the future?

Is there a specific type of naming convention that MR prefers/hates?

Is there a name length I should stay under or characters I should avoid?

Re: Random MR Node failure in MR v03

Posted: Tue Jun 03, 2014 11:19 pm
by polynurb
where does it get stuck?
are you sending dependencies?

i usually do not use that function, but i sometimes observe some nodes getting randomly stuck when receiving dependencies.
you ca turn on "deep debug" in the node log to see more info, and maybe figure out better at what point it is getting stuck.

Re: Random MR Node failure in MR v03

Posted: Tue Jun 03, 2014 11:22 pm
by animation_man
Yeah, I am sending dependencies. Looks like it is getting stuck on sending the MXI file.

Re: Random MR Node failure in MR v03

Posted: Tue Jun 03, 2014 11:50 pm
by polynurb
animation_man wrote:Looks like it is getting stuck on sending the MXI file.
that is something else i noticed in the 3.0 release, which i have not seen in 2.0.
that sometimes either when clicking on "preview" or even when stopping a job, one node does not completely send the mxi.
i can see in the merge folder on the manager machine that one file is much smaller/incomplete and the complete mxi is still available in the nodes temp folder. (that has nothing to do with sending dependencies though, unless you a referring to an .mxi dependency like emitter texture)

when requesting preview one can just click on it again and it will work next time usually, but when stopping a job and it happens, it won't complete successfully.

Re: Random MR Node failure in MR v03

Posted: Thu Jun 05, 2014 1:06 pm
by Mihai
It would help to have some more precise info:

- does it always happen on the same frame
- does it always happen on the same computer
- have you run a very simple test scene to see if the same happens
- do these computers have enough RAM
- how much RAM is Maxwell using while rendering

These are some of the first things I would check.
did about 8 or 12 files just fine then it got stuck
So was it 8, or 12? What was the log saying in this Node? What do you mean exactly by "stuck"? It helps to be as precise as possible.

Re: Random MR Node failure in MR v03

Posted: Mon Jun 09, 2014 6:56 pm
by Bubbaloo
Keep in mind that the Monitor can freeze and it will appear that the rendering process has frozen too. Sometimes all it takes is to close and re-open Monitor.