Hi,
Can’t get AE works today :S since today most of the time nodes are failing 2-3 times on the first frames and then works but now I got only my WS that works. I’m using last AE and laste Deadline (beta) here is the log > share.mattguetta.com/1fukP/3kOqodYr
Error is on l.40 - Translation: aerender ERROR: An existing connection was forcibly closed by the remote host.
Thanks
G
If I remember right, this problem might be caused by licensing. There isn’t much else that AE would try to connect to. The error is usually caused by a firewall.
Are you using the “ae_render_only.txt” trick to have AE render without a license? That may be worthwhile to try out.
Also, I mentioned this earlier in the week on another thread, but Deadline only really catches errors in English mode. Though many of our development staff grew up speaking french at home or in school, translation for the plugins would be hard to maintain.
FIrst thing I’ve try is remove firewall but it didn’t help. I’ve also try the force english parameter but still not working.
All nodes have the file, but this is strange as render works and then break when I requeue some frames. And now nothing works, even a simple file without plugin.
Now I got all working 
Problem was the Red Giant Suite > forums.thinkboxsoftware.com/vie … ant#p24089
While I'm here, I can give the fix to the above error since its was a pain to track down. Its related to Red Giant's Magic Bullet Looks, when the plugin is called After Effects crashes/closes. The fix is to set environment variable KMP_DUPLICATE_LIB_OK=TRUE if you want your program to continue in this case. Its not related to Deadline one bit, After Effects in GUI mode acts the same way. And today all breaks again :S Drives me crazy…
Hi,
We’ve had real problems since Red Giant’s latest set of updates because of the GPU requirements. Before we were alright with rendering on the standard CPU blades with onboard graphics, but following the updates we were experiencing the exact issues you are (Magic Bullet Looks 3 was the culprit) because the onboard is below their minimum spec, maybe you’re experiencing something similar? There’s some details on their site:
redgiant.com/support_old/video-cards/
We’ve still not fully resolved it, we rolled the plugins back to the previous version that doesn’t require GPU as a temporary fix. We’re currently experimenting with running multiple slaves on a dual CPU, dual GPU workstation to try to fix this and get around After Effects poor multi-threading.
Nick
Hi,
Probably related to this plugin but what I don’t understand is why does it works sometimes ?
And why does renderer is not wokring even without the plugin ?
I’ve notice there are errors logs on Windows side too:
[code]-
Weedy > Did you get the errors on node with GPU too ? Because I have one with a 680 and it has the same error than one that use intel HD.
Thanks
Hi,
Initially I tried putting a Quadro 2000 in one of the rendernodes but that didn’t help because the the card was too old. We’re using a mix of 960,970 and 1060 cards now and I’ve not had any problems yet, we are having the weird 0% hanging issue but I don’t think that’s related to the plugins.
Nick
Renamed the title of this thread so hopefully more people can find it.
Hi guys,
Great news ! After lots of testing I found the error and get the whole farm (gpu and non-gpu node) working with last RG version. The key is a settings in the scene > Ctrl+Shift+Alt+K > Choose Mercury Engine
The thing is that if you set this while working you won’t use your GPU so you have to move to Mercury only for rendering.
Edwin, do you think you can add a setting for this in the Deadline Submitter ? So we keep GPU while working but turn off for rendering.
Thanks,
G
Hmm. I can ask around here. I don’t see much of a technical issue assuming the API allows us to do this. If it’s going to be difficult, it will be how we want to handle this. We may put it in as a plugin option.
Plugin option sounds perfect to me
I’ve been running into this issue trying to render an AE project w/ a bunch of plugins.
I tried removing RG Denoiser III, removing Saber, adding the “ae_render_node_only.txt” file, adding that environment variable, and setting the project to use Mercury Software Only for rendering. Two of my workers are giving the “an existing connection was forcibly closed” error message, and my own workstation is giving a “After Effects returned non-zero error code: -1073741819.” error message.
Any other thoughts on how to resolve this issue?
We know that particular error code is equivalent 0xC0000005, which represents a Memory Access Violation error. So After Effects is either running out of memory, or memory has become corrupt.
The work-around for that we’ve used is to change the Plugin python file AfterEffectsPlugin which can be found in [Repository]/plugins/AfterEffects with the code in the docs. Look for “CheckExitCode” function in the AfterEffects.py and change the code after “Else” Statement:
else:
if exitCode == -1073741819:
self.LogInfo( "Ignoring exit code -1073741819" )
elif exitCode != 0: # if exit code non-zero, then fail the job
self.FailRender( "After Effects returned non-zero error code: %s. Check the render log." % exitCode )
Copy that python file to this path “C:\DeadlineRepository10\custom\plugins\AfterEffects”. And try running a test render to check you are getting the same error. Please follow the documentation link for more information on AE’s Error Messages and Meaning.
Fixing that might fix up the issues you’re seeing on the other machines. Let us know how it goes!
-Justin
It did fix the errors I was seeing with my own workstation, but the other machines are still reporting the same “existing connection was forcibly closed” error message.
Right on, halfway there!
Easiest check is to restart your Workers (not the machines, just the application). Just to force them to grab the new file in case they haven’t yet.
If that doesn’t do the trick, could you share the whole job report where you’re seeing that error? If you don’t want to go through the work of making it internet-safe beforehand, feel free to send it to us in a new ticket at support.thinkboxsoftware.com. Just mention a link to this forum thread, for the sake of continuity.
This is a known issue when using Red Giant with Microsoft RDP.
If you are using RDP session to login the machine I would suggest to reboot after you have worked with it. Changing to VNC might help you in this case.