Electron Crashing after update to 2.21.0 with PM2

sdetweil

@pattanner92 sorry two more

node -v
npm-v

u are running on 32 bit buster, which is ok
on some armv7l device, pi3, 4, or pi0w2

pattanner92

@sdetweil
node v16.16.0
npm v8.11.0

I am running on a Raspberry Pi B3+

sdetweil

@pattanner92 can u try running my upgrade script with force instead of apply

see
https://github.com/sdetweil/MagicMirror_scripts

pattanner92

@sdetweil
EDIT: I was digging while cleaning up pm2 so it didn’t display my old script. I noticed that the mm.sh script in ~/MagicMirror/installers was indeed missing export ELECTRON_DISABLE_GPU=1. Adding that back to the script seems to keep it stable. Still wondering if this is an Electron version issue.

No dice. Here are the errors I am seeing in the logs - I pulled only the errors to see if I caught them all. Here is what I am seeing at the exit event:

0|mm  | [25978:1009/112505.776458:ERROR:zygote_communication_linux.cc(275)] Failed to send GetTerminationStatus message to zygote
0|mm  | [25978:1009/112505.812741:ERROR:zygote_communication_linux.cc(275)] Failed to send GetTerminationStatus message to zygote
0|mm  | [25978:1009/112505.848727:ERROR:network_service_instance_impl.cc(470)] Network service crashed, restarting service.
0|mm  | [25978:1009/112505.880738:ERROR:gpu_process_host.cc(968)] GPU process launch failed: error_code=1002
0|mm  | [25978:1009/112505.897587:ERROR:gpu_process_host.cc(968)] GPU process launch failed: error_code=1002
0|mm  | [25978:1009/112505.917812:ERROR:gpu_process_host.cc(968)] GPU process launch failed: error_code=1002
0|mm  | [25978:1009/112505.942511:ERROR:gpu_process_host.cc(968)] GPU process launch failed: error_code=1002
0|mm  | [25978:1009/112505.959337:ERROR:gpu_process_host.cc(968)] GPU process launch failed: error_code=1002

And later in the restart process I saw these errors:

0|mm  | [09.10.2022 11:25.36.393] [ERROR] 
0|mm  | (node:26362) [DEP0005] DeprecationWarning: Buffer() is deprecated due to security and usability issues. Please use the Buffer.alloc(), Buffer.allocUnsafe(), or Buffer.from() methods instead.
0|mm  | (Use `electron --trace-deprecation ...` to show where the warning was created)

I did go in and check for the line export ELECTRON_DISABLE_GPU=1 in the startup script mm.sh and it is there. I did have two startup scripts (one that was created before your script was a a thing) and made sure it was there in both as I wasn’t sure which pm2 was using. The original is out in the home folder - I made a backup and then ran your pm2 setup script. While it seemed to reconfigure pm2 to use the script in ~/MagicMirror/installers it still crashes. Figured I would try that in case something was walking over itself.

Could this be an Electron version issue?

sdetweil

@pattanner92 the buffer thing is a warning, but not our code. nothing we can do

the gpu errors are concerning, because u have gpu use turned off.

do u know what driver is being used in /boot/config.txt, like this

dtoverlay=vc4-fkms-v3d