GPU process isn't usable. Goodbye.

[tobybatch]

Hi first boot on Raspverry PI 3b+

Left it for hours, all looks good, npm finishes but it goes in to a docker boot loop.

pi@magicmirroros:~ $ docker attach mm
[54:0603/124457.729459:FATAL:zygote_linux.cc(155)] Check failed: . : Operation not permitted (1)
[53:0603/124457.740062:FATAL:zygote_linux.cc(155)] Check failed: . : Operation not permitted (1)
[03.06.2023 12:45.01.407] [LOG]   Starting MagicMirror: v2.23.0
[03.06.2023 12:45.01.432] [LOG]   Loading config ...
[03.06.2023 12:45.01.443] [DEBUG] config template file not exists, no envsubst
[03.06.2023 12:45.01.461] [LOG]   Loading module helpers ...
[03.06.2023 12:45.01.466] [LOG]   No helper found for module: alert.
[03.06.2023 12:45.01.526] [LOG]   Initializing new module helper ...
[03.06.2023 12:45.01.529] [LOG]   Module helper loaded: updatenotification
[03.06.2023 12:45.01.532] [LOG]   No helper found for module: clock.
[03.06.2023 12:45.02.465] [LOG]   Initializing new module helper ...
[03.06.2023 12:45.02.471] [LOG]   Module helper loaded: calendar
[03.06.2023 12:45.02.477] [LOG]   No helper found for module: compliments.
[03.06.2023 12:45.02.481] [LOG]   No helper found for module: weather.
[03.06.2023 12:45.02.564] [LOG]   Initializing new module helper ...
[03.06.2023 12:45.02.565] [LOG]   Module helper loaded: newsfeed
[03.06.2023 12:45.02.568] [LOG]   All module helpers loaded.
[03.06.2023 12:45.02.613] [LOG]   Starting server on port 8080 ... 
[03.06.2023 12:45.02.635] [WARN]  You're using a full whitelist configuration to allow for all IPs
[03.06.2023 12:45.02.691] [LOG]   Server started ...
[03.06.2023 12:45.02.695] [LOG]   Connecting socket for: updatenotification
[03.06.2023 12:45.02.697] [LOG]   Starting module helper: updatenotification
[03.06.2023 12:45.02.699] [LOG]   Connecting socket for: calendar
[03.06.2023 12:45.02.701] [LOG]   Starting node helper for: calendar
[03.06.2023 12:45.02.704] [LOG]   Connecting socket for: newsfeed
[03.06.2023 12:45.02.706] [LOG]   Starting node helper for: newsfeed
[03.06.2023 12:45.02.708] [LOG]   Sockets connected & modules started ...
[03.06.2023 12:45.03.539] [LOG]   Launching application.
[51:0603/124503.685900:ERROR:gpu_process_host.cc(985)] GPU process launch failed: error_code=1002
[51:0603/124503.730946:ERROR:gpu_process_host.cc(985)] GPU process launch failed: error_code=1002
[51:0603/124503.750552:ERROR:gpu_process_host.cc(985)] GPU process launch failed: error_code=1002
[51:0603/124503.763975:ERROR:gpu_process_host.cc(985)] GPU process launch failed: error_code=1002
[51:0603/124503.775088:ERROR:gpu_process_host.cc(985)] GPU process launch failed: error_code=1002
[51:0603/124503.786113:ERROR:gpu_process_host.cc(985)] GPU process launch failed: error_code=1002
[51:0603/124503.786276:FATAL:gpu_data_manager_impl_private.cc(440)] GPU process isn't usable. Goodbye.
/opt/magic_mirror/node_modules/electron/dist/electron exited with signal SIGTRAP

The tiniternet suggest adding --no-sandbox to the start up I'm just not sure where to add it. Can you help please?

I'm very docker ready and *nix sysadminy, I'll raise a PR to help if I can fix it.

[khassel]

can you first check which os you are running with cat /etc/os-release and check if you have the correct docker-compose.yml content?

• for bullseye use the content of $HOME/magicmirror/run/rpi.yml
• for buster use the content of $HOME/magicmirror/run/rpi_buster.yml

Before please do a git pull in $HOME/magicmirror to get the newest files.

If this does not help there was one similar error reported here without solution.

You can set the electron options and electron switches in the $HOME/magicmirror/mounts/config/config.js file. This is described here with links to the possible options/switches.

[tobybatch]

Sorry for the delay. I ended up installing default os and setting lightdm to autorun MM so I've dug out another 3b+ and a card, and I can replicate the issue.

pi@magicmirroros:~ $ cat /etc/os-release
PRETTY_NAME="Raspbian GNU/Linux 10 (buster)"
NAME="Raspbian GNU/Linux"
VERSION_ID="10"
VERSION="10 (buster)"
VERSION_CODENAME=buster
ID=raspbian
ID_LIKE=debian
HOME_URL="http://www.raspbian.org/"
SUPPORT_URL="http://www.raspbian.org/RaspbianForums"
BUG_REPORT_URL="http://www.raspbian.org/RaspbianBugs"

pi@magicmirroros:~/magicmirror $ git pull 
Already up to date.

Assuming that ~/magicmirror/run/docker-compose.yml is the one being used by the OS:

pi@magicmirroros:~/magicmirror/run $ diff $HOME/magicmirror/run/rpi_buster.yml rpi_buster.yml

I added electronSwitches:["disable-gpu"], to the config.js and rebooted, no luck.

pi@magicmirroros:~ $ docker ps -a
CONTAINER ID   IMAGE                          COMMAND                  CREATED        STATUS          PORTS     NAMES
00f77aaf00ff   karsten13/magicmirror:latest   "/usr/bin/tini -- ./…"   46 hours ago   Up 18 seconds             mm
pi@magicmirroros:~ $ docker ps -a
CONTAINER ID   IMAGE                          COMMAND                  CREATED        STATUS                         PORTS     NAMES
00f77aaf00ff   karsten13/magicmirror:latest   "/usr/bin/tini -- ./…"   46 hours ago   Restarting (1) 4 seconds ago             mm

Almost the same error, same trace:

pi@magicmirroros:~ $ docker attach mm

> [email protected] start
> DISPLAY="${DISPLAY:=:0}" ./node_modules/.bin/electron js/electron.js

[54:0609/100930.018204:FATAL:zygote_linux.cc(155)] Check failed: . : Operation not permitted (1)
[53:0609/100930.023373:FATAL:zygote_linux.cc(155)] Check failed: . : Operation not permitted (1)
[09.06.2023 10:09.33.719] [LOG]   Starting MagicMirror: v2.23.0
[09.06.2023 10:09.33.746] [LOG]   Loading config ...
.....
[51:0609/100936.025968:ERROR:gpu_process_host.cc(985)] GPU process launch failed: error_code=1002
[51:0609/100936.026124:FATAL:gpu_data_manager_impl_private.cc(440)] GPU process isn't usable. Goodbye.
/opt/magic_mirror/node_modules/electron/dist/electron exited with signal SIGTRAP

[khassel]

did you do the git pull because karsten13/magicmirror:latest is the wrong image for buster, you have to use karsten13/magicmirror:buster.