AWS Thinkbox Discussion Forums

Cinema 4D/Redshift - Task completes in seconds but no output

Deadline
#1

Hi there,

We’re rendering scenes through Deadline 10.1.23.6 with Cinema 4D R25 using RS 3.0.65 and every once in a while certain machines on the farm will pick up tasks, complete them in record time (like 10 seconds) but actually produce no output frame and the log looks like the below.

This one dodgy machine will then do this for all the tasks in that job it can get its hands on so artists think the job is complete but actually nothing has been rendered.

Anyone had anything similar? Seems like its just a Cinema thing

Thanks alot,
Will

2022-10-07 10:31:47:  0: Render Thread - Render State transition from = 'ReceivedTask' to = 'Other'
2022-10-07 10:31:47:  0: Plugin already loaded: Cinema4DBatch
2022-10-07 10:31:47:  0: Start Job timeout is disabled.
2022-10-07 10:31:47:  0: Task timeout is disabled.
2022-10-07 10:31:47:  0: Job already loaded: ONX2201_UNIQUESYSTEMS_DESIGNS_stairs_mc_v020 - beauty sh1
2022-10-07 10:31:47:  0: Plugin rendering frame(s): 169
2022-10-07 10:31:47:  0: Render Thread - Render State transition from = 'Other' to = 'Rendering'
2022-10-07 10:31:58:  0: Executing plugin command of type 'Render Task'
2022-10-07 10:31:58:  0: INFO: Render Tasks called
2022-10-07 10:31:58:  0: INFO: Pre Build Script
2022-10-07 10:31:58:  0: INFO: Validating the path: 'F:/FD-LDN/ONX2201_UNIQUE_SYSTEMS/DESIGN/Marcus/Cinema4D/./render'
2022-10-07 10:31:58:  0: INFO: Rendering main output to network drive
2022-10-07 10:31:58:  0: INFO: Validating the path: 'F:/FD-LDN/ONX2201_UNIQUE_SYSTEMS/DESIGN/Marcus/Cinema4D/./render'
2022-10-07 10:31:58:  0: INFO: Rendering multipass output to network drive
2022-10-07 10:31:58:  0: INFO: 
2022-10-07 10:31:58:  0: STDOUT: Running Script: C:\ProgramData\Thinkbox\Deadline10\workers\chinook-02\jobsData\633fef48bd8b2b12688617e0\thread0_temp1igKC0\c4d_Batch_Script.py
2022-10-07 10:31:58:  0: STDOUT: Rendering frame 169 at <Fri Oct  7 10:31:58 2022>
2022-10-07 10:31:59:  0: INFO: Script Ran Successfully
2022-10-07 10:32:00:  0: INFO: Finished Cinema 4D Task
2022-10-07 10:32:00:  0: Done executing plugin command of type 'Render Task'
2022-10-07 10:32:00:  0: Render time for frame(s): 12.653 s
2022-10-07 10:32:00:  0: Total time for task: 23.513 s
2022-10-07 10:32:01:  0: Saving task log...

=======================================================
Details
=======================================================
Date: 10/07/2022 10:32:01
Frames: 169
Job Submit Date: 10/07/2022 10:20:08
Job User: marcus
Average RAM Usage: 4033688832 (2%)
Peak RAM Usage: 4033716224 (2%)
Average CPU Usage: 0%
Peak CPU Usage: 0%
Used CPU Clocks (x10^6 cycles): 0
Total CPU Clocks (x10^6 cycles): 676412

=======================================================
Worker Information
=======================================================
Worker Name: CHINOOK-02
Version: v10.1.17.4 Release (d3559fe75)
Operating System: Windows 10 Pro
Running As Service: Yes
Machine User: ldn-renderslave
IP Address: 10.151.68.101
MAC Address: AC:1F:6B:8A:E4:8A
CPU Architecture: x64
CPUs: 40
CPU Usage: 8%
Memory Usage: 19.4 GB / 254.7 GB (7%)
Free Disk Space: 171.736 GB 
Video Card: NVIDIA GeForce RTX 2080 Ti
#2

Hey!
I had this too in the past. Unfortunately I can’t remember the solution to the problem, however I remember that I’ve solved it back then by giving the job a Minimum Task Render Time in Job Properties - Timeouts
If it finishes in under a minute, count it as error. That way these frames should eventually be picked up by other nodes and produce actual results.

#3

Hello @FD_Tech
Seem like you are running Deadline Launcher as a service and the output path is a Windows network drive (letter). You will need to map this letter drive in service context for the user running the Deadline Launcher.

Here’s how to do that.

#4

Hey both,

Thanks for your replies :slight_smile:

@Thamer_Florian yeee thats our ususal workaround, works well but want to avoid it, thank you though, Caleigh says hi :slight_smile:

@zainali ye we are mapping them using the Mapped Drives configuration in the Repository settings? That works well - do you think it’s linked to that?

Cheers,
Will

1 Like
#5

@FD_Tech

You are very welcome. Yes I think, when running as a service the drive letters need to mapped for the service account. Deadline can do it for you if you configure the settings in the Repository Options, look at the link I shared.

#6

Hey @zainali

yee we’re doing that already, thats all setup and working fine for other jobs.

Anything else it could be?

Cheers,

#7

Make sure the machines are updated as well as the repo :wink:

#8

We have this same random occurrance using the standard renderer. No other job has done this. When opening the scene on another machine to test, C4d reported missing assets. Still waiting to test with those missing assets bundled and fixed, however might be worth you checking if it’s the same thing happening.
10.1.18 did report missing assets and failed the renders iirc, however 10.1.23 just starts and completes the task in 10 seconds with no output.

#9

If you are using the Network drives it could be that some machines are not able to access the network locations due to network issues.

Another issue could be that the limitation of sharing protocol or OS of the host. It can limit the number of connections made simultaneously to the drive to read assets.

I do not think this a Deadline issue but to troubleshoot further easy way is to isolate the issue from Deadline. You will need to submit the same job by using C4D non batch plugin. To do that just uncheck the box which says “Use Batch Plugin”. Then submit, let it fail and then follow here: Troubleshooting — Deadline 10.1.23.6 documentation

Share your findings.

To troubleshoot even more deeply, you can run some network and storage test. We can talk about those after the results of isolation test.

Privacy | Site terms | Cookie preferences