I don’t see anything in the script to do the pm2 save or a “pm2 start mm --udpate-env” so that it would use the new version of node. Not sure if your script is already gracefully handling those kinds of updates for the MM startup.
Also, since the script seems to assume Debian-based distro’s, dpkg doesn’t execute on my system and I’m not entirely sure if the script would effectively update my system for me. Thoughts on that?
Do any of your scripts check for pm2 and update the version during any installs or upgrades? Seems like, for the latter at least, that would be helpful.
My original mirror was obviously built quite a long time ago as the node18 variable was linked to the startup process. No idea how many MM revisions I went through between then and now and was always running on the much older version of node no matter what I did to upgrade the main system.
@sdetweil I don’t believe it’s the modules.
Leap 16 fresh install is fine with a fresh MM install as well. I upgraded the server again from Leap 15 to Leap16. Before the upgrade, I removed node and npm modules linked to version 18 and MM was running fine in all mannners.
After the upgrade, I made a couple of system-level changes (like disabling the firewall that it forces on you) and added the node/npm modules for version 24 and wasn’t able to start the mirror with pm2 like normal… it continues to throw errors about node18 being missing. However, running the mm.sh script manually or launching the mirror via either npm start server or node --run server works fine.
Did more digging around and found that the issue lies solely in the hands of pm2… the back-end environment was created back in the node/npm 18 days and that was saved in the configuration. The only way to find it was to do a “pm2 dump” and scan the file.
Once I knew it was pm2 for sure, the actual fix was pretty simple:
pm2 save
npm install pm2 -g
pm2 update
pm2 unstartup
pm2 startup
If MM wasn’t still configured as a process…
cd ~
pm2 start mm.sh
pm2 save
I’m running on Leap 16, latest patches, node/npm 18 is uninstalled, mirror is properly auto-launching. Now, I can update the modules and the mirror… But not before I do a snapshot! lol
@sdetweil The bug may have been around for a while… my mm.sh script is launching the mirror via “npm run server” and that script is what pm2 uses to launch the mirror. I’ve been doing it that way for a long time, likely because the serverOnly config setting wasn’t working.
If I actively run “npm run server” and then do “ps aux |grep node” I do not see anything related to node18 in the process list.
@sdetweil As mentioned, I reverted… back to whatever I was previously running.
Working on a fresh Leap 16 install, fresh MM install. Running into problems with inability to connect to display (running serveronly, no display connected). Seems that the serverOnly config setting may not get used any more? I have to start the server with “node --run server” to get past the display errors.
BTW… fresh install, there is no ‘default’ directory under modules… Thoughts?
@sdetweil I reverted back again (Leap 15.6)
I recall seeing a 36 or something somewhere if that helps.
I had even moved the MM folder off and tried starting again, re-installed npm 22 and 24, and the errors were exactly the same. Something is out-of-whack somewhere and I haven’t had the time to dig deep enough to understand if it’s MM, my linux machine, or a combo.
I’m considering trying a fresh Leap 16 build and seeing what happens. If that works without an issue, then at least I know “it’s possible.”
@sdetweil Yep.
The only thing is puts on the screen of any potential interest is
[ -f node_modules/.bin/husky ] && husky || echo no husky installed
@sdetweil I did an OS upgrade to move to Leap 16. I had a couple of quirky issues with that and corrected them. Post-upgrade, I added node24 packages but Leap 16 no longer supports the node18 packages (so, those are gone).
I updated the modules and then MM and now it throws an error claiming that it can’t locate node18 executable. I tried searching for some sort of symbolic link, but I’m not finding anything.
@sdetweil I’ll have to try again a little later today maybe… I had to revert the snapshot (removing the module updates) to get it running again.
Update…
Updating the modules crashes the server. So, I’m going to have to get the main server updated… I had to revert to a prior snapshot to get the MM server running again.
@sdetweil I am running (currently) on opensuse Leap 15.6. I will be updating in the not-too-distant future to Leap 16.0.
In looking at the installed packages, Node.js lists “22.22.0” as the package but “22.15.1” as the installed version. So…
zypper ref ; zypper up
Let it install all of the pertinent updates… Reboot (new kernel was part of the updates)… Everything is back online. Checking the installed software list now shows Node.js as “22.22.0”.
Leap 16.0 will allow me to go to Node.js 24, but that would be an undertaking for later today or later this week (depending on what time I might have elsewhere).
Updating the modules went fine (overall). This is what a git pull in the main MM directory shows me:
From https://github.com/MichMich/MagicMirror
3dbe8bfbb..fb41d24ef master -> origin/master
ec80b2508..e747db16d develop -> origin/develop
* [new tag] v2.1.3 -> v2.1.3
* [new tag] v2.14.0 -> v2.14.0
* [new tag] v2.15.0 -> v2.15.0
* [new tag] v2.16.0 -> v2.16.0
* [new tag] v2.17.0 -> v2.17.0
* [new tag] v2.17.1 -> v2.17.1
* [new tag] v2.18.0 -> v2.18.0
* [new tag] v2.19.0 -> v2.19.0
* [new tag] v2.2.1 -> v2.2.1
* [new tag] v2.20.0 -> v2.20.0
* [new tag] v2.21.0 -> v2.21.0
* [new tag] v2.22.0 -> v2.22.0
* [new tag] v2.23.0 -> v2.23.0
* [new tag] v2.24.0 -> v2.24.0
* [new tag] v2.25.0 -> v2.25.0
* [new tag] v2.26.0 -> v2.26.0
* [new tag] v2.27.0 -> v2.27.0
* [new tag] v2.28.0 -> v2.28.0
* [new tag] v2.29.0 -> v2.29.0
* [new tag] v2.30.0 -> v2.30.0
* [new tag] v2.31.0 -> v2.31.0
* [new tag] v2.32.0 -> v2.32.0
* [new tag] v2.33.0 -> v2.33.0
* [new tag] v2.34.0 -> v2.34.0
* [new tag] v2.35.0 -> v2.35.0
* [new tag] v2.36.0 -> v2.36.0
Updating 3dbe8bfbb..fb41d24ef
error: Your local changes to the following files would be overwritten by merge:
package-lock.json
package.json
Please commit your changes or stash them before you merge.
Aborting
@sdetweil Thanks for the response.
I’m familiar with the different options that come with a new install and would be ok with having to DO a new install if that’s the only way to get upgrades completed. I guess I’m wondering if there’s a simpler way to upgrade the one or two modules that I have and get the core MM upgraded as well without having to start from scratch with the configuration and such. This instance is server only (there’s nothing attached to any sort of output here, and I don’t even run MM as the client any more - it’s a linux shell with its own GUI, auto-starts firefox in kiosk mode, and the page is set to always launch the mirror address).
I’m not even concerned with how to back everything up because my installation is actually fairly simple. And I’m not actually looking to change environments - I’ve already done that from a short while back. This is all entirely for the server side which is a linux guest running on a XCP-NG virtualization host.
I’ve been using MM for a number of years in a couple of different forms. For the most part, I did what everyone else does and I ran it standalone on a RPi. After having some issues with the RPi hardware and older software causing lock-up’s, I moved to more of a “client-server” setup where the server pieces was installed and running on a linux client (headless) and my local RPi would launch a browser session to connect.
I now find myself in the situation where I would very much like to upgrade the server-side component, but I don’t see a clear directive on the steps I should consider. Attempting to use @sdetweil 's upgrade script results in errors about wrong platform and such, so that’s a no-go.
I’ve tried a couple of other npm type updates but they leave the system in a completely unstable state and I end up having to just revert back to a previous snapshot and restart the server to its former state.
Any general suggestions of where to start?
@sdetweil Yeah… I should have said I updated to the latest version of Bullseye with all patches. I haven’t made the last step to Bookworm yet…
Yeah, I actually did more troubleshooting and such over the weekend and ended up doing a full upgrade to the RPi OS and then updating MM on there as well.
My troubleshooting was showing that the server side “shouldn’t” have been the issue because launching a browser on just about any other device would result in the continual processing and rotating of photos and such. And, even though the photo processing seemed to be the only piece that would stop updating on the client, the fact that it worked everywhere else led me to take the leap and do all of the updates on the RPi.
After doing the full upgrade to the latest version of RaspiOS (Bullseye) and then updating MM to the latest (using your scripts), it seems to be working correctly again and hasn’t stopped rotating the photos since the upgrade.
@mumblebaj said in Ads are blocking site content - how to get rid of them?:
@ember1205 Easiest way for me was installing the Adguard plugin on Chrome
Thanks. I’ll give this one a try…
@thgmirror Not really an option for me.
How does the site admin not understand that this is counter-productive to the site?
This is getting frustrating as it makes it impossible to read certain posts / pages… How do I get rid of this intrusive content?
@sdetweil I’m not running on a Pi or its software… Full install of opensuse linux. And I keep it fully patched and at the current release level.
@sdetweil I keep the OS updated separately on its own. What kind of updates are you suggesting would be necessary here?