Black screen on HMDI output, content on remote browser

[sdetweil]

@chris_d interesting. I ssh into my mirror systems all the time and never have trouble.

do

pm2 stop all
cd ~/MagicMirror
npm start >somefile.txt 2>&1

wait til the black screen
ctrl-c in the ssh terminal

then nano somefile.txt, scroll down
see if there are any errors reported…

the package.json will set DISPLAY if not set

[chris_d]

@sdetweil said in Black screen on HMDI output, content on remote browser:

pm2 stop all
cd ~/MagicMirror
npm start >somefile.txt 2>&1

Errors are:
[2883:0119/213451.321388:ERROR:viz_main_impl.cc(186)] Exiting GPU process due to errors during initialization
[2916:0119/213451.509005:ERROR:gpu_memory_buffer_support_x11.cc(44)] dri3 extension not supported.

[sdetweil]

@chris_d in from to the npm start line do

export ELECTRON_DISABLE_GPU=1

and rerun…

[chris_d]

@sdetweil said in Black screen on HMDI output, content on remote browser:

export ELECTRON_DISABLE_GPU=1

No error now, but still a black screen

[sdetweil]

@chris_d on the black screen press ctrl-shift-i to open the developers window, then select the console tab and scroll up for any errors usually red text

[MikeBishop]

@sdetweil

I’m having what appears to be the same problem – HDMI is black (but apparently happy from the log output), while browser access is fine. I can’t get the dev tools to come up in Electron; perhaps unsurprisingly, the browser dev tools don’t show any errors that weren’t there on my old instance.

[sdetweil]

@MikeBishop are you also running on buster?

[sdetweil]

@chris_d try npm start dev to open dev window on startup

also can u try the npm start on the pi desktop, not over ssh?

[MikeBishop]

@sdetweil

Bullseye, looks like. Whatever’s in the newest RPi OS release, anyway.

For the moment, I’ve swapped back to the other SD card that’s working. System’s been a little flaky the last few days, figured I’d try a fresh install on a new card in case that’s starting to go; hit this issue on the new one. When I have a few minutes to pull the Pi from the mirror and bring it back to my desk, I’ll give npm start dev a try on the newer one.

[sdetweil]

@MikeBishop can u reflash bullseye and use my install script?
https://github.com/sdetweil/MagicMirror_scripts

[MikeBishop]

@sdetweil

That’s actually how I installed this instance to begin with. However, I’ve narrowed my instance of the problem down.

The node_helper for the module I wrote, MMM-Powerwall, has an async start() function. In previous MM versions, the startup code didn’t wait on async start functions to finish, and my init finished not long after MM started. This version includes the fix to wait for async modules to finish starting. I’ve confirmed that my start() function is running to the end, and MM does write

[21.01.2023 14:30.25.915] [LOG]   Sockets connected & modules started ...

…to the log, but apparently something about waiting on the module breaks the electron browser. If I hack the module to look synchronous, like so:

start: /*async*/ function () {
		this.asyncStart()
	},	

asyncStart: async function () {

…everything starts right up. Something about the timing or threading of waiting for modules to complete is making the electron launch fail, I think.

My next step will be to make a module that does nothing but has an async node_helper and see if I can repro the issue with that.

@chris_d, any chance that you’re also using a module with an async start function?

[MikeBishop]

@sdetweil

Okay, I think I have a clean repro of the issue in https://github.com/MikeBishop/MMM-asyncdeath – the synchronous start function produces this in the logs and loads normally:

[21.01.2023 14:55.02.692] [LOG]   Synchronous start-up -- all is well
[21.01.2023 14:55.02.693] [LOG]   Sockets connected & modules started ...
[21.01.2023 14:55.03.235] [LOG]   Launching application.

The async start function produces this in the logs and fails:

[21.01.2023 14:52.54.143] [LOG]   About to break by using await
[21.01.2023 14:52.54.884] [LOG]   Launching application.
[21.01.2023 14:52.56.512] [LOG]   Sockets connected & modules started ...

Note the reversed order of “Launching application” and “modules started”. Seems feasible this could also happen with a synchronous module startup that takes a long time, but I haven’t tested that.

[sdetweil]

@MikeBishop great work

can you post your findings to issue

https://github.com/MichMich/MagicMirror/issues/2487

I reopened it

[chris_d]

@MikeBishop I’m running MMM-Calendar3 and the default modules only.