I seem to run into some issues submitting AE renders via the AE submit script.
I have used the installer to put it in the right place but get the message that the “SubmitAEToDeadline.jsx” can’t be found in the repository while it’s there.
when I copy both this script and the “DeadlineAfterEffectsClient.jsx” into the AE scripts folder I manage to get it working…
I first run the “DeadlineAfterEffectsClient.jsx” and then the “SubmitAEToDeadline.jsx” and am able to submit a job.
ok, I’ll have a look at the beta as i’m running into some weird stuff.
=======================================================
aerender Error: After Effects error: File exists but couldn't be open for reading: Z:\Focus4\02_Ocean\RNDR\Layers\Ocean_v0.32\Ocean_v0.32_.0000.png.
at Deadline.Plugins.ScriptPlugin.RenderTasks(String taskId, Int32 startFrame, Int32 endFrame, String& outMessage, AbortLevel& abortLevel)
This just happened a few times (i’m guessing it’s limited fileserver (NAS) reachability ?)
I also found 4 corrupted files (using .png at the moment to preserve some space)
On top of that I had 600 frames, rendering in sets of 6 frames over 10 machines.
It’s a ocean surface that has a few effects applied, but in general all very similar (just waves going on).
Some frames sets render in a few minutes, others took up to an hour. I noticed the longer it ran the slower it got.
Then i suspended the job and changed the settings to “local rendering” and the rest of the queue was put out in under 5 minutes per set of 6 frames.
Is this purely a traffic jam on the network or can it also be something else?
If it is network related, anyone have suggestions on how to limit this?
So based on what sounds like network congestion in some aspects, but I would be interested to see what it was doing when it slowed up, if you can grab the slave log from the respective machines when this happens.
A small update… I haven’t had a chance to upgrade the farm but have found more about the slow down part.
The missing frames are definitely attributed to the network congestion (and partially shitty file handling by adobe stuff)
Also while encoding via AME I often have ‘offline footage’ that are actually there, but the files are not reached in time to be encoded.
I fear I’ll have to upgrade some infrastructure to a bit more professional / industrial setup to counteract this.
(now I just have one single NAS serving everything over a gigabit network to regular win7 machines)
Besides that I believe Adobe should fix their file handling a bit better (not to blame, but it seems notoriously bad at it.)
another small update… I know it might be useless as it’s an older version of deadline, but figured to mention it anyway.
I noticed that when I select ‘reload plugin’ the rendertimes do not increase over time.
on average the frames take about 3-4 minutes to render. When it’s not reloading AE it slowly increases depending on the amount of frames rendered by that slave.
This could be an AE thing, but figured to mention it
yeah… plus I didn’t have my mapped drives properly configured.
It was using the NAS name, not ip address, which might not have helped to prevent some congestion…
Ok… just to confirm. Using IP’s instead of the NAS name did the trick.
I’m using now 3-4 slaves to run AE on 11 machines and haven’t lost a frame since.
And the multiple slaves per node thing is awesome as it with AE it speeds things up tremendously
There’s some wierd quirks but will first upgrade to the latest DL until I post them around here
