fatal error when at 100%

[jarod.tulanowski]

when using the Db for my rendering I get the fallowing error. the sad part is I keep trying to use it even though it isnt stable. because without it It takes 7 hours to complete. with it it is 45 min.

mental ray has encountered a fatal error and the application will shut down the error is DB 10.n fatal 041500:interuption by exception code 0xc0000005 (access window)

this happens only when it gets finished and before the cleanup stage

I have 10 machines running windows xp pro 64 bit intel xeon E5530 @ 2.40 ghz, 12 gig of ram nvidia quadro FX 4800.

P.S. I do not use backburnner because not sure how to use it with DB. I heard it is stable though.

[jarod.tulanowski]

here is the screen capture

[Steve_Bennett]

Have you tried to setup a file to save the image to before rendering instead of using the rendered image window to save the image?

If so, does the image file created display correctly or is it corrupt?

Are you sure each machine has the same amount of RAM installed?

It looks as though the error is calling a specific node of the DB network as failing.

[jarod.tulanowski]

yes did both the save and not save. the problem occures before the save (clean up stage). because no file is created. all machines have the same ram 12 gig.

is there a way of tracking which node??

[Steve_Bennett]

the only way I know of to track this is run test renders with just one machine in the DB dialog and try different ones until it fails.

Also, how many machines are you running in the DB mode?

I heard recently from an Autodesk support webcast that running more than 8 in DB mode, mental ray can be problematic and unstable. For more than eight machines they recommend using back burner.

[jarod.tulanowski]

is that 8 pc's or 8 core processors. because we have quad core machines. and if i just need to use two machines than that is that. I will give the test mode a try thanks for the feedback.

[jarod.tulanowski]

Darn steve I was hopeful about your ideas. But with the test first with 2 licenced machines, then one both of them crashed with different node numbers. the first was 2.n then the second was 1.n. the wierd part is that I think the node is the number of machines not the indevidual machine. This is really bugging me because it speeds up my renderings sooooooo much. but to get to 100% and then crash just sucks

[Steve_Bennett]

is that 8 pc's or 8 core processors. because we have quad core machines. and if i just need to use two machines than that is that. I will give the test mode a try thanks for the feedback.
8 CPU sockets. Not boxes or cores.

[Steve_Bennett]

Darn steve I was hopeful about your ideas. But with the test first with 2 licenced machines, then one both of them crashed with different node numbers. the first was 2.n then the second was 1.n. the wierd part is that I think the node is the number of machines not the indevidual machine. This is really bugging me because it speeds up my renderings sooooooo much. but to get to 100% and then crash just sucks
Can you render a region, say the upper left quadrant at a pixel resolution to cover the area you want, then stitch the other quadrants together in PS once they are done?