desktop
desktop copied to clipboard
libva sporadic desctop client crash
I confirm (by marking "x" in the [ ] below: [x]):
- [x] This is not a troubleshooting question. Troubleshooting questions go here: http://www.mattermost.org/troubleshoot/.
- [x] This doesn't reproduce on web browsers (such as in Chrome). If it does, issue reports go to the Mattermost Server repository.
- [x] I have read contributing guidelines.
Summary Latest Linux Desktop App dies with
libva error: vaGetDriverNameByIndex() failed with unknown libva error, driver_name = (null)
[108487:1203/213321.397395:ERROR:sandbox_linux.cc(374)] InitializeSandbox() called with multiple threads in process gpu-process.
Environment
- Operating System: Fedora 35
- Mattermost Desktop App version: 5.0.2 but was observed on all 5.x.x
- Mattermost Server version: 6.0.1
Steps to reproduce crashes sporadically :-(
Expected behavior works
Observed behavior crashes
@shurrman I have seen this once or twice, does it happen when you're doing something or is it completely random?
@devinbinnie I faced it several times, and I think several times I even had another app active/in front of me. Then I was like: "ok, let me see what if going on in our MTM channel..." - and then I see that the app is gone.
@shurrman Looks like this is an issue in Electron: https://github.com/electron/electron/issues/30902
One commenter mentioned that changing off of the nouveau
driver helped them, not sure if you're using that yourself?
Also looks like GPU acceleration could affect that, you can try turning that off in the Settings window and see if that helps.
I am not using nouveau Turned off GPU acceleration, sad that MTM is not able to use it :-(
I am not using nouveau Turned off GPU acceleration, sad that MTM is not able to use it :-(
Let me know if it reproduces with GPU acceleration off.
@devinbinnie just got it again. Acceleration is turned off: [152392:1207/183235.064606:ERROR:gpu_init.cc(441)] Passthrough is not supported, GL is swiftshader I know exactly what I was doing at that time - however, I was unable to reproduce, doing same things again
Okay thanks, we can eliminate that as a workaround. I'm not sure what else could be causing it, it's likely an issue with the Electron internals at this point.
Looks like the upstream issue was closed, so it's unlikely this will be fixed. Hopefully it's working with the current version of Electron. Closing as stale.