@sdetweil

Full logs…

[TAILING] Tailing last 100 lines for [all] processes (change the value with --lines option)
/home/pi/.pm2/pm2.log last 100 lines:
PM2        | 2022-05-10T15:00:25: PM2 log: PM2 successfully stopped
PM2        | 2022-05-10T15:00:44: PM2 log: ===============================================================================
PM2        | 2022-05-10T15:00:44: PM2 log: --- New PM2 Daemon started ----------------------------------------------------
PM2        | 2022-05-10T15:00:44: PM2 log: Time                 : Tue May 10 2022 15:00:44 GMT-0400 (Eastern Daylight Time)
PM2        | 2022-05-10T15:00:44: PM2 log: PM2 version          : 4.4.1
PM2        | 2022-05-10T15:00:44: PM2 log: Node.js version      : 16.15.0
PM2        | 2022-05-10T15:00:44: PM2 log: Current arch         : arm
PM2        | 2022-05-10T15:00:44: PM2 log: PM2 home             : /home/pi/.pm2
PM2        | 2022-05-10T15:00:44: PM2 log: PM2 PID file         : /home/pi/.pm2/pm2.pid
PM2        | 2022-05-10T15:00:44: PM2 log: RPC socket file      : /home/pi/.pm2/rpc.sock
PM2        | 2022-05-10T15:00:44: PM2 log: BUS socket file      : /home/pi/.pm2/pub.sock
PM2        | 2022-05-10T15:00:44: PM2 log: Application log path : /home/pi/.pm2/logs
PM2        | 2022-05-10T15:00:44: PM2 log: Worker Interval      : 30000
PM2        | 2022-05-10T15:00:44: PM2 log: Process dump file    : /home/pi/.pm2/dump.pm2
PM2        | 2022-05-10T15:00:44: PM2 log: Concurrent actions   : 2
PM2        | 2022-05-10T15:00:44: PM2 log: SIGTERM timeout      : 1600
PM2        | 2022-05-10T15:00:44: PM2 log: ===============================================================================
PM2        | 2022-05-10T15:00:45: PM2 log: App [mm:0] starting in -fork mode-
PM2        | 2022-05-10T15:00:45: PM2 log: App [mm:0] online

/home/pi/.pm2/logs/mm-error.log last 100 lines:
0|mm       | + '[' -f ./untrack-css.sh ']'
0|mm       | + grep docker /proc/1/cgroup -qa
0|mm       | + '[' -z :0 ']'
0|mm       | ++ uname -m
0|mm       | + arch=armv7l
0|mm       | + false=false
0|mm       | ++ grep -v '^[[:blank:]]*//' config/config.js
0|mm       | ++ grep -i serveronly:
0|mm       | ++ awk -F: '{print tolower($2)}'
0|mm       | ++ sed -e 's/^[[:space:]]*//'
0|mm       | ++ tr -d ',"'\''\r'
0|mm       | + serveronly=local
0|mm       | + serveronly=local
0|mm       | ++ pgrep Xorg
0|mm       | + xorg=
0|mm       | + '[' . == . ']'
0|mm       | ++ pgrep X
0|mm       | + xorg=
0|mm       | ++ uname
0|mm       | + mac=Linux
0|mm       | + '[' local. '!=' false. -o armv7l == armv6l ']'
0|mm       | + '[' local. == true. ']'
0|mm       | + '[' . == . -a Linux '!=' Darwin ']'
0|mm       | + node serveronly

/home/pi/.pm2/logs/mm-out.log last 100 lines:
0|mm       |
0|mm       | > magicmirror@2.13.0 start
0|mm       | > ./run-start.sh
0|mm       |
0|mm       | [2022-05-10 15:00:55.768] [LOG]    Starting MagicMirror: v2.13.0
0|mm       | [2022-05-10 15:00:55.790] [LOG]    Loading config ...
0|mm       | [2022-05-10 15:00:55.804] [LOG]    Loading module helpers ...
0|mm       | [2022-05-10 15:00:57.248] [LOG]    Initializing new module helper ...
0|mm       | [2022-05-10 15:00:57.252] [LOG]    Module helper loaded: MMM-ImagesPhotos
0|mm       | [2022-05-10 15:00:57.264] [LOG]    No helper found for module: alert.
0|mm       | [2022-05-10 15:00:57.385] [LOG]    Initializing new module helper ...
0|mm       | [2022-05-10 15:00:57.387] [LOG]    Module helper loaded: updatenotification
0|mm       | [2022-05-10 15:00:57.391] [LOG]    No helper found for module: clock.
0|mm       | [2022-05-10 15:00:57.686] [LOG]    Initializing new module helper ...
0|mm       | [2022-05-10 15:00:57.689] [LOG]    Module helper loaded: calendar
0|mm       | [2022-05-10 15:00:57.692] [LOG]    No helper found for module: currentweather.
0|mm       | [2022-05-10 15:00:57.696] [LOG]    No helper found for module: weatherforecast.
0|mm       | [2022-05-10 15:00:57.698] [LOG]    All module helpers loaded.
0|mm       | [2022-05-10 15:01:00.465] [LOG]    Starting server on port 8080 ...
0|mm       | [2022-05-10 15:01:00.504] [LOG]    Server started ...
0|mm       | [2022-05-10 15:01:00.507] [LOG]    Connecting socket for: MMM-ImagesPhotos
0|mm       | [2022-05-10 15:01:00.509] [LOG]    Starting node helper for: MMM-ImagesPhotos
0|mm       | [2022-05-10 15:01:00.511] [LOG]    Connecting socket for: updatenotification
0|mm       | [2022-05-10 15:01:00.514] [LOG]    Connecting socket for: calendar
0|mm       | [2022-05-10 15:01:00.516] [LOG]    Starting node helper for: calendar
0|mm       | [2022-05-10 15:01:00.517] [LOG]    Sockets connected & modules started ...
0|mm       | [2022-05-10 15:01:00.519] [LOG]
0|mm       | Ready to go! Please point your browser to: http://0.0.0.0:8080

@sdetweil

armv71

I’ve modified my shutdown control to stop MM, wipe the .config/chromium directory, and then power off. This allows chromium to load the correct information when it does finally launch.

There still remains an issue, however, with the run-start.sh script actually launching the chromium browser on boot. The log file basically ends with “0|mm | Ready to go! Please point your browser to: http://0.0.0.0:8080” on boot up. “pm2 restart mm” correctly brings everything up every time.

Here’s what the log looks like when it boots:

0|mm       | > magicmirror@2.13.0 start
0|mm       | > ./run-start.sh
0|mm       |
0|mm       | [2022-05-10 14:30:59.191] [LOG]    Starting MagicMirror: v2.13.0
0|mm       | [2022-05-10 14:30:59.207] [LOG]    Loading config ...
0|mm       | [2022-05-10 14:30:59.222] [LOG]    Loading module helpers ...
0|mm       | [2022-05-10 14:31:00.614] [LOG]    Initializing new module helper ...
0|mm       | [2022-05-10 14:31:00.617] [LOG]    Module helper loaded: MMM-ImagesPhotos
0|mm       | [2022-05-10 14:31:00.625] [LOG]    No helper found for module: alert.
0|mm       | [2022-05-10 14:31:00.733] [LOG]    Initializing new module helper ...
0|mm       | [2022-05-10 14:31:00.735] [LOG]    Module helper loaded: updatenotification
0|mm       | [2022-05-10 14:31:00.741] [LOG]    No helper found for module: clock.
0|mm       | [2022-05-10 14:31:01.058] [LOG]    Initializing new module helper ...
0|mm       | [2022-05-10 14:31:01.060] [LOG]    Module helper loaded: calendar
0|mm       | [2022-05-10 14:31:01.068] [LOG]    No helper found for module: currentweather.
0|mm       | [2022-05-10 14:31:01.072] [LOG]    No helper found for module: weatherforecast.
0|mm       | [2022-05-10 14:31:01.074] [LOG]    All module helpers loaded.
0|mm       | [2022-05-10 14:31:02.616] [LOG]    Starting server on port 8080 ...
0|mm       | [2022-05-10 14:31:02.665] [LOG]    Server started ...
0|mm       | [2022-05-10 14:31:02.680] [LOG]    Connecting socket for: MMM-ImagesPhotos
0|mm       | [2022-05-10 14:31:02.683] [LOG]    Starting node helper for: MMM-ImagesPhotos
0|mm       | [2022-05-10 14:31:02.685] [LOG]    Connecting socket for: updatenotification
0|mm       | [2022-05-10 14:31:02.693] [LOG]    Connecting socket for: calendar
0|mm       | [2022-05-10 14:31:02.700] [LOG]    Starting node helper for: calendar
0|mm       | [2022-05-10 14:31:02.702] [LOG]    Sockets connected & modules started ...
0|mm       | [2022-05-10 14:31:02.704] [LOG]
0|mm       | Ready to go! Please point your browser to: http://0.0.0.0:8080

@sdetweil

What’s the goal?

I have .config/chromium and .config/Electron with the latter being two years old and the former having a current date (I switched to Chromium because Electron was a ton of trouble).

This is all on Bullseye which I upgraded to (yeah, yeah, I know… don’t upgrade, fresh install, not supported, etc.). Up to that point, it was on Buster and ran fine.

Incidentally… I have an automated routine to bring the RPi down gracefully based on a virtual switch setting on a home automation controller. Here’s the gist:

• Set the virtual switch to “on”
• Every five minutes, the RPi queries the status of the switch
• If the switch is set to “1” (on), it launches a script with sudo

The script does the following:

• Using cec-client, it puts the Soundbar into Standby (off)
• Using cec-client, it puts the TV into standby (off)
• Sets the status of the virtual switch to “1” (on) to ensure that any other items that trigger of the state change of that switch will run as well
• Sets the status of the virtual switch to “0” (off)
• Sends an IFTTT message using WebHooks / curl
• Does a “shutdown now”

If I comment out the cec commands, it starts up incorrectly. If I manually just do a “shutdown now” and boot it back up, it starts up incorrectly.

Thanks - I had dug into them and found nothing of any use. Still, I went ahead and purged the logs and tried again.

I’m getting VERY weird behavior with zero consistency. I booted the mirror yesterday a half dozen different times and I was getting a split between not launching the browser and launching the browser but landing on “Web Store” and “Magic Mirror” icons in the center with no content.

First boot attempt today resulted in the browser not only launching but correctly bring up my configured mirror and content. Today’s logs show no indication of a PM2 daemon being started, but this WAS being shown in the logs yesterday (very strange).

Executed a powerdown of the Pi, powered it off, waited about ten minutes, powered it back on. Made it all the way to bringing up the desktop and then … nothing. No entry in the log that it was starting Chromium. Executing “pm2 restart mm” brought the browser up but it landed again on the “Magic Mirror” and “Web Store”.

This is frustrating.

If I boot up my RPi, it auto-launches the desktop but nothing loads for MM. If I check the status (pm2 status mm), it shows as running since boot.

If I restart MM (pm2 restart mm), all is well.

Where do I even begin to try and understand what’s going on?

This is an interesting idea, and I found this write-up that I thought was interesting.

https://labs.tadigital.com/index.php/2020/04/10/night-and-day-switch-using-css-variables/

I can’t seem to get the system to tell me that it’s out of date again (even though I know it is). That’s ok for now, though… New problem…

Chromium is crashing, and I have a guess as to what might be going on. I have been searching and can not find anything related to a log file for Chromium, though, and I would need to be able to see that to understand if my suspicion is correct. Am I just completely missing it?

@sdetweil I honestly don’t recall if I renamed the original folder MM-backup/ and then did rsync back to MagicMirror/ or just did rsync to MM-backup/. My rsync would have been with flags ‘-vaPHSx’, so either way it should have gotten everything.

I’ll check for the .git directory…

@sdetweil This is all that it spits out on a test run (and MM no longer shows that it is out of date but it is):

Upgrade started - Mon Dec 28 09:07:08 EST 2020
system is Linux raspberrypi 4.19.75-v7+ #1270 SMP Tue Sep 24 18:45:11 BST 2019 armv7l GNU/Linux
the os is Distributor ID: Raspbian Description: Raspbian GNU/Linux 10 (buster) Release: 10 Codename: buster
doing test run = true, NO updates will be applied!

saving custom.css
Unable to determine upstream git repository
restoring custom.css
removing git alias
Upgrade ended - Mon Dec 28 09:07:09 EST 2020

I tried connecting to a couple of web sites via CURL and it throws errors about loading certificates and these sites should work fine. So, the combo of everything is why I’m suspecting that there is a certificate issue due to an older OS on the Pi. I’m doing an upgrade on it now…

@sdetweil I used your upgrade script on my mirror. During the upgrade, it reverted the configuration back to using Electron instead of retaining the run-start.sh directive that would launch Chromium.

I wiped out the entire MM directory and copied back the working one that wasn’t upgraded. Unfortunately, it now seems that it is refusing to run the upgrade script and I’m wondering if it’s a certificate issue and I need to update the Pi itself first…

Wanted to post what is likely a ‘final update’ to this thread. I was running for literally weeks if not months with zero issues. I even let the system run for at least a month with the “you’re out of date” banner across the top and not one crash.

I updated the mirror and it reverted itself back to using Electron (that in itself is infuriating) and it’s crashing again. So, not only is Electron the problem but it’s STILL a problem.

I’ve reverted my mirror back to the outdated code and will do another update and not allow it to change back to Electron so that it can run without crashing!

@trix2k said in adding google calendar:

well, i now have it syncing with my google calendar and with my iphone. the last issues i have is its not showing birthdays on the MM, i dont know if that its because its an all day event but i dont think it has to do with the MM i think its a google calendar thing…i hope :)

Google Calendar is actually a “collection” of calendars. The primary calendar is where all of your manually entered events are stored. This is the calendar you have linked to in the MM configuration.

Birthdays (those that are associated directly with Contacts in your account as part of their Contact Card) are stored in a different calendar. You can either add this calendar to the configuration like you did with your primary calendar, or you can manually create Birthday events in your primary calendar for the people whose birthdays you want to show up.

I have a WiFi plug that switches the device on at a specific time in the morning.

I use a cron job entry to shut the Pi down at a specific time at night.

I have the WiFi plug turn off at a specific time at night, five minutes after the time I’ve configured the cron job to shut the Pi down.

There is also an IFTTT integration where you can have the shutdown script make a specific call to create a calendar entry on a Google calendar and IFTTT will pick up that entry and use it to trigger the power-off event for the plug. I used to do this until IFTTT changed their account structure and I’ve been moving away from using any of their integrations.

Update… 30 full days since my last post about stability and have not experienced a single crash on either mirror. So, Electron is definitely the source of the problem.

Thank you to @sdetweil for the assistance in swapping over to Chromium and getting things to a stable state.

@cowboysdude I agree that most are quick to say “do this for me” and seldom say “how can I help you get this changed?” When you’re dealing with commercial products, that approach is expected because the source is closed. But, this is open-source and community-developed. We all have the ability to contribute -something-…

@raf So, figure out what’s incorrect and help get it updated. The doc you’re trying to work from was built in that same way at one point…

@raf Are you volunteering? :)

Seriously - You won’t be anywhere near as well-versed as if you can figure it out on your own. And documenting your learnings to help others is how all of this stuff grows and improves. I’ve pestered @sdetweil plenty about a lot - much of his input has been to help me understand where to look, but he has also done some coding and such on his side. I’ve offered back some things I learned along the way that he has then incorporated and I’m now going to look into helping to extend and enhance some work he has done by doing the coding myself. It take a village, as they say…

@sdetweil said in new update/upgrade script, ready for testing:

https://github.com/sdetweil/MagicMirror_scripts

Where’s a good place to discuss the scripts? I’m wondering how feasible and useful it might be to add to the upgrade-script.sh one.

A couple of quick questions…

• The “UpdateNotification” module is able to detect that there is a new version of MM available. Is this logged somewhere or easily accessible from the command line? I’m interested in writing a script that can assess the new version, possibly run a ‘test upgrade’ and then send me a IFTTT or pushover notification that a new version is available and an upgrade test succeeded (or failed).

• Is the new upgrade script bundled with MM yet? I thought that was the intent, but did not see it up through 2.10.x

@sdetweil v10 is out and listed as “stable”.