I have installed Deadline 10.2.0.10 on M1 computers. On every client the launcher and the worker have problems accessing the connection.ini file in the repository. Here is the log file:
On the mac where the mongod is running there is also an access error. The repository has read and write access for all users.
2023-02-20 11:09:49: BEGIN - MacMini-RF02\admin
2023-02-20 11:09:49: Operating System: macOS 12.6.1
2023-02-20 11:09:49: CPU Architecture: x86_64
2023-02-20 11:09:49: CPUs: 8
2023-02-20 11:09:49: Video Card: Apple M1
2023-02-20 11:09:49: Deadline Launcher 10.2 [v10.2.0.10 Release (3b87216c7)]
2023-02-20 11:09:49: ERROR: DataController threw a configuration exception during initialization: Could not read the connection.ini for the given repository path “/Volumes//Database/Deadline/Repository10_2" because:
2023-02-20 11:09:49: Access to the path '/Volumes//Database/Deadline/Repository10_2/settings/connection.ini’ is denied. (System.UnauthorizedAccessException) (Deadline.Configuration.DeadlineConfigException)
2023-02-20 11:09:49: Error encountered when connecting to repository /Volumes//Database/Deadline/Repository10_2: Object reference not set to an instance of an object. (System.NullReferenceException)
2023-02-20 11:09:49: at Deadline.Launcher.Launcher.ChangeRepository(RepositoryConnectionSettings updatedSettings)
2023-02-20 11:09:49: at Deadline.Launcher.Launcher.CheckForConnectionUpdate(Boolean verbose)
2023-02-20 11:09:49: WARNING: PythonSync (Python 3) upgrade failed with the following error:
2023-02-20 11:09:49: Object reference not set to an instance of an object.
2023-02-20 11:09:49: WARNING: PythonSync (Python 2) upgrade failed with the following error:
2023-02-20 11:09:49: Object reference not set to an instance of an object.
2023-02-20 11:09:49: Launcher Thread - Launcher thread initializing…
2023-02-20 11:09:49: Launcher Thread - opening remote TCP listening port 17000
2023-02-20 11:09:49: Launcher Thread - creating local listening TCP socket on an available port…
2023-02-20 11:09:49: Launcher Thread - local TCP port bound to: [::1]:54932
2023-02-20 11:09:49: Launcher Thread - updating local listening port in launcher file: 54932
2023-02-20 11:09:49: Launcher Thread - Launcher thread listening on port 17000
2023-02-20 11:09:50: ERROR: DataController threw a configuration exception during initialization: Could not read the connection.ini for the given repository path "/Volumes//Database/Deadline/Repository10_2” because:
2023-02-20 11:09:50: Access to the path ‘/Volumes/***/Database/Deadline/Repository10_2/settings/connection.ini’ is denied. (System.UnauthorizedAccessException) (Deadline.Configuration.DeadlineConfigException)
2023-02-20 11:09:50: Launcher Thread - No database connection - cannot update Repository Options
2023-02-20 11:09:50: Launcher Thread - Remote Administration is now disabled since there is no database connection.
I have found this thread: OSX Catlina issues
I have added bin/sh to security drive access list and also set the write permissions of Benutzer/geteilt/Deadline
to read and write for the logged in Deadline user. That seems to do the trick.
Are there any plans of an Deadline release for Mac M1 nativley without Rosetta?
So here is an update. I have started the deadline worker with -nogui to make sure it is not ui related.
The worker runs for several minutes, sometimes 30minutes sometimes even for hours, but then suddenly the spin wheel comes up and the worker gets stalled. The gui is open until you force quit the worker. BTW in the Deadline logs there are no errors. Any idea?
Does this happen while the Worker is running a render? Or just while it’s sitting idle?
It could be that there’s something going oddly but only at the verbose logging level - to enable that in the Monitor under Tools → Configure Repository Options → Application Data enable Verbose Worker logging.
See if you can induce that failure again and let’s see what kind of logs we get back.
It happens during idle. I can reproduce this error several time. Just the time until it crashes is different every time. I have attached a log with verbose enabled.
Logs_MacOSVentura_M2_VerboseWorker.zip (5.4 KB)
@Justin_B I’m also having the same issue with workers randomly hanging.
I hope your able to look into this for us.
Logs_MacOSVentura_M1Max_VerboseWorker.zip (349.6 KB)
@Justin_B I’ve just been testing the latest Deadline Client Version: 10.2.1.1 Release (094cbe890)
and the problem persist. Is there anyone at amazon/thinkbox that can investigate this issue?
looking at the logs you have shared on 11th March, looking at the crash logs it seems to be an issue with Qt5:
43 QCoreApplication::exec() + 130 (QtCore + 2021170) [0x1ac11a732] 1-43
43 QEventLoop::exec(QFlags<QEventLoop::ProcessEventsFlag>) + 430 (QtCore + 2000222) [0x1ac11555e] 1-43
43 ??? (libqcocoa.dylib + 224081) [0x1af4eeb51] 1-43
43 -[NSApplication run] + 586 (AppKit + 196595) [0x7ff8079d6ff3] 1-43
43 -[NSApplication(NSEvent) _nextEventMatchingEventMask:untilDate:inMode:dequeue:] + 1219 (AppKit + 252344) [0x7ff8079e49b8] 1-43
43 _DPSNextEvent + 909 (AppKit + 256823) [0x7ff8079e5b37] 1-43
43 _BlockUntilNextEventMatchingListInModeWithFilter + 70 (HIToolbox + 194339) [0x7ff80e7af723] 1-43
43 ReceiveNextEventCommon + 679 (HIToolbox + 195046) [0x7ff80e7af9e6] 1-43
43 RunCurrentEventLoopInMode + 292 (HIToolbox + 195542) [0x7ff80e7afbd6] 1-43
43 CFRunLoopRunSpecific + 560 (CoreFoundation + 511647) [0x7ff804987e9f] 1-43
43 __CFRunLoopRun + 1387 (CoreFoundation + 514692) [0x7ff804988a84] 1-43
43 __CFRunLoopServiceMachPort + 145 (CoreFoundation + 520235) [0x7ff80498a02b] 1-43
43 mach_msg + 19 (libsystem_kernel.dylib + 6537) [0x7ff804870989] 1-43
43 mach_msg_overwrite + 723 (libsystem_kernel.dylib + 34586) [0x7ff80487771a] 1-43
43 mach_msg2_trap + 10 (libsystem_kernel.dylib + 5794) [0x7ff8048706a2] 1-43
*43 ??? (kernel.release.t6000 + 782844) [0xfffffe000840f1fc] (suspended) 1-43
As per the internal runbook on the same issue, Qt is also having this as a known problem:
Deadline 10.2.1.10 bundles Qt 220.127.116.11, and seems to be causing issues. For now, we do not have a resolution to this issue yet. Can you try starting the Monitor and Worker from command line if that does not help start it from the Finder. If that fails too then,
make sure that Rosetta 2 is installed and if it installed already re-install it. If you need to install Rosetta on your Mac - Apple Support
Thanks for taking a look the log. Like @Freiraum_FX_GmbH I’ve tried the -nogui option but still getting the worker crashing. So it’s possibly not related to Qt. I believe Maya and Houdini are all based on Qt for their GUI’s and have been working fine.
Are you able to do any deeper investigation, and reproduce the crash on your end?
@David_Deacon I was looking into this with the team and we were not able to figure out the issue following through the crash reports. The plan is to run the strace macOS equivalent command with the worker.exe to trace the process calls and find where the crash is happening. For this I would like to move this conversation internally to our ticket system as to keep any of your sensitive information away from public forums.
You would be able to install and run Dtrace on your macOS with the deadlineworker application to get the process calls made to the OS. Here is a documentation: DTrace Tools
Can you open a ticket with us on https://awsthinkbox.zendesk.com/ and attach the result from the dtrace command.