AWS Thinkbox Discussion Forums

Worker randomly crashing on MacOS M2 Ultra

Hi!

I installed yesterday Deadline client to a new Mac Studio (M2 Ultra chip) and I’m experiencing stability issues with the Workers. They would randomly freeze, stall or just crash and the only solution if to force exist and re-launch the worker. I have a few workers running on PC (intel) without any problem.

I have tried Deadline 10.2.1.0 without success. I updated my repo and client to the latest 10.3.0.9 and still experience the same problem. I also made sure Rosetta was installed and I’m pretty sure it works because I’m running After Effects with the Rosetta mode enabled.

I know I’m not the only one in this situation because I found quite a few threads on this forum describing the same problem; but I was wondering if there’s any setting I’m not aware of?

I’ve attached the Apple Crash log as well as the worker log (even tho it’s quite empty and not usefull).

Thank you!

deadlineslave-NH-MacStudio01s-Mac-Studio-2023-08-23-0000.zip (741 Bytes)
Deadlineworker_AppleLog.zip (262.5 KB)

same thing , both monitor and slave stalls on M2 Studios
restarting the slaves and monitors work but only for a non set amount of time,

no replies from Deadline team on this… very frustrating

Could you try running the Worker in nogui mode to try and get an Apple crash log from it?

To do that run /Applications/Thinkbox/Deadline10/bin/deadlineworker.exe -nogui. As long as there isn’t another Worker running on the machine the Worker should fire up and start working as normal. But without QT loaded (which may be the cause of the hang+crash) a log from a crash in nogui mode may get us another thread to pull on.

Thanks!

M1 Studio Ultra and same issue. Worker won’t run for more than 3 or 4 minutes without locking up.
Form_streamedlog.txt.zip (3.8 KB)

1 Like

I’m also experiencing this with latest Deadline release. Any resolution?

1 Like

Any updates on this? These worker stalls on Mac OS are still happening for me.

1 Like

We’ve got no news we can share yet. We’ve re-created the issue locally but I can’t comment on the roadmap.

Its now almost August and there doesnt seem to be a fix for this?
I see it happening consistently on 15 M2 Mac machines in a production enviroment.
The monitor app crashes after a while on the license/repo server
Nodes seem to crash more frequently when they have gone idle for a while

sudo sysctl -w net.inet.tcp.delayed_ack=0

this might be a possible fix, at least it lowers the frequency of workers stalling. this fixes macos VNC laggy and i suspect it might be the same with deadline. read more macos legacy tcp settings here

Hey Justin, any updates on this? I’m using m2 ultra, latest deadline worker, and Ventura OS. Worker freezes after 30-60 mins of running.

Privacy | Site terms | Cookie preferences