Fusion Studio do not correct close

I did upgrade Linux Mint from 21.3 to 22. Since them Fusion Studio 19.01 do not correct ending and I couldn't start it again. If I close Fusion the three processes are always exist (see below). I reinstalled Fusion Studio but it shows the same behavior.

GUI Thread -> parent process
and twice fuscript -> child processes (see attachment)

If I kill the process "GUI Thread" then I could start Fusion again.
Why do this processes not correct close after I close Fusion Studio?

Regards, Pius

Is there nobody here who has the same problem and possibly a solution? Or what is Black Magic's answer?

I saw something similar lately... But I'm on mac.

Either at launch it first bounces forever in the dock without going to the splash screen > I kill and try again and usually it's ok then; or I have to kill it in the end to totally quit it.

Weird; given I know I can "finally" access I didn't report but this is obviously not normal indeed

You're not alone!

Today I started noticing the same behaviour (after updating to Fusion 19.1.2, but I guess it also happened with earlier versions). The GUI Thread process keeps running forever.

I'm on Mint 22, too.

Starting from the terminal gives this:

Code: Select all

/opt/BlackmagicDesign/Fusion19$ ./Fusion
sh: 1: python2: not found
Attribute Qt::AA_ShareOpenGLContexts must be set before QCoreApplication is created.
DNX SDK API version 2.7.3-27-r
libDeckLinkAPI.so: cannot open shared object file: No such file or directory
Unable to read VR Path Registry from /home/steffen/.config/openvr/openvrpaths.vrpath
FFmpeg AVCodec Ver: 60.31.102
11:43:22,228 | INFO  | GPUDetect          | Starting GPUDetect 1.2_6-a2
11:43:22,294 | DEBUG | GPUDetect          |   - System Info: 66 ms.
11:43:22,294 | DEBUG | GPUDetect          |   - OpenCL: 1 detected in 0 ms.
11:43:22,294 | DEBUG | GPUDetect          |   - CUDA: 1 detected in 0 ms.
11:43:22,306 | DEBUG | GPUDetect          |   - Xorg: 1 detected in 11 ms.
11:43:22,323 | DEBUG | GPUDetect          |   - NVML: 1 detected in 17 ms.
11:43:22,323 | DEBUG | GPUDetect          |   - GPU Matching: 1 matched in 0 ms.
11:43:22,323 | INFO  | GPUDetect          | Done in 95 ms.
11:43:22,323 | INFO  | GPUDetect          | Detected System:
11:43:22,323 | INFO  | GPUDetect          |   - OS: Linux Linux Mint 22
11:43:22,324 | INFO  | GPUDetect          |   - Model: SchenkerTechnologiesGmbH GM5MxxW E20
11:43:22,324 | INFO  | GPUDetect          |   - System ID: 997377e6ec9b45ebaecb336560a418da
11:43:22,324 | INFO  | GPUDetect          |   - CPU: Intel Core i7-10875H, 16 threads, x86-64
11:43:22,324 | INFO  | GPUDetect          |   - RAM: 4.6 GiB used of 62.7 GiB
11:43:22,324 | INFO  | GPUDetect          |   - NVIDIA GPU Driver: 560.35, supports CUDA 12.6
11:43:22,324 | INFO  | GPUDetect          | Detected 1 GPUs:
11:43:22,324 | INFO  | GPUDetect          |   - "NVIDIA GeForce RTX 2070 Super" (gpu:f6fdd2ca.f91e8f0c) <- Main Display GPU
11:43:22,324 | INFO  | GPUDetect          |     Discrete, 268 MiB used of 7.2 GiB VRAM, PCI:1:0.0
11:43:22,324 | INFO  | GPUDetect          |     Matches: CUDA, NVML, OpenCL, XOrg
11:43:22,324 | INFO  | GPUDetect          | Detected 1 monitors:
11:43:22,324 | INFO  | GPUDetect          |   - "Monitor" <- Main Monitor
11:43:22,324 | INFO  | GPUDetect          |     2560x1440, connected to "NVIDIA GeForce RTX 2070 Super" (gpu:f6fdd2ca.f91e8f0c)
log4cxx: No appender could be found for logger (DVIP).
log4cxx: Please initialize the log4cxx system properly.
FusionScript Server [48531] Started
Host 'Fusion' [48478] Added
sh: 1: python2: not found
Host 'FusionServer' [48548] Added
QCursor: Cannot create bitmap cursor; invalid bitmap(s)
Host 'Fusion' Removed
Host 'FusionServer' Removed
FusionScript Server [48531] Terminated


After that I have to manually kill the GUI Thread.

I started noticing it after Fusion didn't close after closing an open project.

Just to be sure I completely erased everything Fusion (all the configs in all of the 4 (!!) folders and subfolders.

The hang persists.

I also noticed I can no longer keep "use neural engine optimization" activated. I can activate it in the preferences but after closing and re-opening the prefs it's off again.

EDIT: In Resolve 19.1.2 everything it shiny. No hanging Fusion, neural engine optimization working.

I can confirm this with Fusion Studio since version 19.0 on windows, and persists in 19.1.1, but less frequently and not always. Sometimes it happens with Resolve Studio as well. Never happened before with 18.6.6 or earlier for me. Usually I have to use task manager to force quite, but that means some things don't get saved in the application, UI toolbar rearrangement etc. Looks like something is a glitch. I will try to install 19.1.2 later this week so will see.

Thanks for confirming. So it looks like it's happening in Windows, too?

The original poster and me we both use Linux Mint 22, but then I guess the rabbit hole goes even deeper.

Yes. Wondows 10. At first I though it was because originally in version 19.0 of fusion studio I had issues with USD not loading correctly, it was a hardware thing, which they fixed in later updates 19.0.3 I think. But the program was a bit buggy still, so I hoped it will be ironed out later, and by 19.1.1 it was mostly stable with not fully closing. It still happens about 60-70% of the time, but I was not able to figure out why it doesn't happen 100% of the time, so I'm unsure of the cause. Resolve has the same issue, but much less frequently. I will try 19.1.2 when I get a change in the next few days and report back.

I just uninstalled 19.1.2 and reinstalled 19.1.1

The "GUI Thread" is still hanging after closing Fusion BUT the "Use neural engine optimization" option is now working again.

When launched from the terminal, 19.1.1 reports fewer errors than 19.1.2:

Code: Select all

/opt/BlackmagicDesign/Fusion19$ ./Fusion
sh: 1: python2: not found
Attribute Qt::AA_ShareOpenGLContexts must be set before QCoreApplication is created.
DNX SDK API version 2.7.3-27-r
libDeckLinkAPI.so: cannot open shared object file: No such file or directory
Unable to read VR Path Registry from /home/steffen/.config/openvr/openvrpaths.vrpath
FFmpeg AVCodec Ver: 60.31.102
11:37:13,752 | INFO  | GPUDetect          | Starting GPUDetect 1.2_6-a2
11:37:13,817 | DEBUG | GPUDetect          |   - System Info: 64 ms.
11:37:13,817 | DEBUG | GPUDetect          |   - OpenCL: 1 detected in 0 ms.
11:37:13,817 | DEBUG | GPUDetect          |   - CUDA: 1 detected in 0 ms.
11:37:13,824 | DEBUG | GPUDetect          |   - Xorg: 1 detected in 6 ms.
11:37:13,841 | DEBUG | GPUDetect          |   - NVML: 1 detected in 17 ms.
11:37:13,842 | DEBUG | GPUDetect          |   - GPU Matching: 1 matched in 0 ms.
11:37:13,842 | INFO  | GPUDetect          | Done in 89 ms.
11:37:13,842 | INFO  | GPUDetect          | Detected System:
11:37:13,842 | INFO  | GPUDetect          |   - OS: Linux Linux Mint 22
11:37:13,842 | INFO  | GPUDetect          |   - Model: SchenkerTechnologiesGmbH GM5MxxW E20
11:37:13,842 | INFO  | GPUDetect          |   - System ID: 997377e6ec9b45ebaecb336560a418da
11:37:13,842 | INFO  | GPUDetect          |   - CPU: Intel Core i7-10875H, 16 threads, x86-64
11:37:13,842 | INFO  | GPUDetect          |   - RAM: 4.5 GiB used of 62.7 GiB
11:37:13,842 | INFO  | GPUDetect          |   - NVIDIA GPU Driver: 560.35, supports CUDA 12.6
11:37:13,842 | INFO  | GPUDetect          | Detected 1 GPUs:
11:37:13,842 | INFO  | GPUDetect          |   - "NVIDIA GeForce RTX 2070 Super" (gpu:f6fdd2ca.f91e8f0c) <- Main Display GPU
11:37:13,842 | INFO  | GPUDetect          |     Discrete, 410 MiB used of 7.2 GiB VRAM, PCI:1:0.0
11:37:13,842 | INFO  | GPUDetect          |     Matches: CUDA, NVML, OpenCL, XOrg
11:37:13,842 | INFO  | GPUDetect          | Detected 1 monitors:
11:37:13,842 | INFO  | GPUDetect          |   - "Monitor" <- Main Monitor
11:37:13,842 | INFO  | GPUDetect          |     2560x1440, connected to "NVIDIA GeForce RTX 2070 Super" (gpu:f6fdd2ca.f91e8f0c)
FusionScript Server [31690] Started
Host 'Fusion' [31635] Added
sh: 1: python2: not found
Host 'FusionServer' [31709] Added
QCursor: Cannot create bitmap cursor; invalid bitmap(s)
Host 'Fusion' Removed
Host 'FusionServer' Removed
FusionScript Server [31690] Terminated

+1 for Resolve Studio 19.1.2 Build 3 on Windows 11 (latest)

Happened in previous 19.x.x versions, too. Always have to kill via Taskmanager.

+1 for Resolve Studio 19.1.2 Build 3 on Windows 11 (latest)

Happened in previous 19.x.x versions, too. Always have to kill via Taskmanager.

That's shocking... so something is fishy here.

Just chiming in on this, I can not get the Neural Engine Optimizations to stay selected in 19.1.2. No matter what I try every time I open fusion and check the preferences, it is deselected.

Windows 10, Dual nvidia 3090 driver 566.14

Quick update: I installed 19.1.3 and nothing has improved.

It still leaves a hanging "GUI Thread" on Linux and the "Use neural engine optimization" option doesn't stay active.

Same on Windows, still have to kill DaVinci in Taskmanager.

Another update: Installed 19.1.4 and the GUI Thread is still hanging after closing Fusion.

But the "Use neural engine optimization" option now stays active

I didn't close for me either. Just updated to Resolve Studio 19.1.4 and doesn't close. But sometimes it does. Same is true for Fusion studio. Not quite sure yet, why. All other applications close, resolve and fusion are the only ones that stay not closed, but not always. Sometimes they do close as well. Hard to pin point a pattern.

Thanks for your reply.

In my case (Linux Mint) Resolve always closes fine and Fusion always has a hanging process lurking around like a zombie (GUI Thread).

MacOS here : Fusion 19.1.4 didn't close - even once - normally.
"Error 7: luascript"

MacOS here : Fusion 19.1.4 didn't close - even once - normally.
"Error 7: luascript"

Thanks for your report.

Now it's all 3 platforms (Windows, Linux and Mac) where unusual things happen during closing Fusion.

Haha, as chance would have it, it just conveniently shut down ok now (first time however)...

It was after a little and basic correction in a comp, as opposed to when I was torturing it for hours in a huge comp; could it be related to a script which cleans the cache or stuff ?

In my case I can just start Fusion, do nothing and then close it again. It closes but the one process "GUI Thread" keeps on running and consuming RAM.

And then after a few hours I can try to start Fusion again and nothing happens (because it's already running, as a zombie). Then I have to manually kill the "GUI Thread" and then I can use Fusion again.

Well, reading you guys scan your activity processes, I've been indeed having an ongoing root user "fusionserver" using 8,5 Mo RAM and still there, but after check, it's related to the app Fusion Render Nodes...
Are you sure it's not connected to the same on your machine(s) ?

In my case the "fusionserver" process keeps on running for a few seconds after quitting Fusion and then stops but the "GUI Thread" process keeps on running until I manually kill it or I restart the system.

Fusion 20 Beta 1: Same as before, GUI Thread keeps hanging.

EDIT: Additionally now Resolve 20 Beta 1 crashes during the neural engine optimization. Resolve has been working flawlessly before. Now it's getting worse there for me as well

It would seem that I also am having this issue. Fusion is not "reaping" fuscript. This would be a bug in the program if you ask me. Basically, not cleaning up things when program closes.

I even tried this as a script to see if I could force it to reap the process.

/opt/BlackmagicDesign/Fusion/fusion "$@" &
fusion_pid=$!
wait $fusion_pid

But this did not work. This should be reported as a bug. I do not see a place to report such things. For what it is worth, I run fusion from a script to set xwayland etc., so I have a terminal when I close it. If I close the terminal is cleans things up and there is no zombie fuscript. It is just a pain in the butt.

Also, I was thinking that since I do not have python 2.7 and it does not load it. Maybe it just does not like the fact that it is not loaded? Just a thought.

Cheers
Rocky Linux 9.5

Also, I was thinking that since I do not have python 2.7 and it does not load it. Maybe it just does not like the fact that it is not loaded? Just a thought.

Thanks for your reply. I'm on Linux Mint and was thinking about giving it a try under Rocky Linux (because it's "officially supported"). Good to know it's happening there as well.

When running Fusion from the terminal I also noticed a while ago that it complained about not finding Python2, so I installed Python 2.7.18 via pyenv. The warning / error message disappeared but Fusion (GUI Thread and fuscript) still kept on becoming zombies after closing it.