I have no idea what is going on my push notifications on my server and why they are so inconsistent. Sometimes they will make it through and sometimes they won’t. Messages seem to work fine however anything to do with calling and video calling notifications just does not want to work.
Sometimes the notification for calls will come through but most of the time it just doesn’t want to function.
I have looked in my Docker container logs and there is nothing indicating an error. It all seems to be working, even troubleshooting notifications in the Element client and Element X client. The tests pass with flying colors.
It feels as if my Element clients are not running in the background and completely shutdown. Anytime I re-open the app it looks as if it is opening for the first time.
The only thing I could possibly think is causing the issue is I may not have setup cloudflare tunnels properly. I don’t know what would be causing ntfy or matrix to not play nice.
I am running my server on Ubuntu 20.04 with Docker and CasaOS. I am using Cloudflare Tunnels for my Matrix and ntfy containers. The logs for all the containers show nothing abnormal. The issue is appearing on 2 Google Pixel 8 devices that are running GrapheneOS. All the proper settings for the ntfy app and element have been configured (unrestricted battery, notification vibration and sound, etc.) It seems that calls go through fine if the app remains open. I also have a digital ocean vps for my coturn server.
This is what my element app looks like whenever I open it again. Maybe I interpret the loading screen wrong but to me it looks like it’s doing “first time startup” type actions. This doesn’t make sense though because both client devices are recieving text notifications.
I also found this github issue that seems to be exactly the issue I am facing (except they are having issues with messages as well it seems, when messages is the only notification that works consistently for me). However I don’t know if some of the proposed solutions are only applicable to Synapse and not Dendrite.
https://github.com/element-hq/element-android/issues/7069
Thank you all for your time and help!
Edit: some more github issues referencing my issue. These seem more specific as well:
https://github.com/element-hq/element-android/issues/8761
https://github.com/element-hq/element-x-android/issues/3031
I do not know if this value will work in my Dendrite config:
ip_range_whitelist:
Edit 2: I created a throwaway matrix account and nltifications through my ntfy server worked flawlessly. Which makes me think my issue must be with my co turn server on my vps. Forgot to mention my vps, I have updated the post with this info. Going to try and see if this is the issue.
Edit 3: I think my coturn is fine nothing in its logs really stands put to me as broken. It has to be dendrite or something else. Possibly cloudflare, but I don’t know what would be the issue.
You must log in or register to comment.
saw your other post as well, there’s so much stuff in your stack that you have to narrow your issues down to the malfunctioning item, there’s no way you’re gonna stumble upon someone with your exact setup.
start with existing element x apps communicating over matrix.org (use throwaway accounts if you don’t trust it) and unified push using ntfy.sh. if everything works there, then you can start replacing one by one of these until you achieve full functionality or something breaks.
My bad sorry should have thought about making an official matrix account and testing there. Based off of what I can tell my ntfy container is working because it works flawlessly with an official matrix account.
That leaves me with two ideas so far, there is something wrong with matrix dendrite container, or my vps coturn server (which I forgot to mention). It looks like traffic is coming through just fine on my co turn server though. I am curious if this is a firewall issue with my co turn server. That would make the most sense given that element call is also not working on element x.
It’s weird though because calling works on seperate networks just fine so I had assumed that my co turn server just worked. Odd.
Edit: I think it is the turn server even though my calls are going through. I went to my maytrix url with “/_matrix/client/r0/voip/turnServer” to diagnose webrtc and it says “errcode: ‘M_MISSING_TOKEN’” "error: ‘missing access token’ "