SOLVED Problem with sending and receiving message's Android/Windows
-
Matrix (Synapse server) seems to be running, but sending messages on Android and Windows not working anymore.....suddenly...
Message we get is "messages failed to send"..Also noticed that after a restart of the Cloudron instance, message's get through, but after 5 or 10 minutes the process repeats and sending message's isn't possible no more...
This happend immediately after updating Cloudron instance from 6.2.8 to v6.3.4 running on Ubuntu 18.04.4 LTS.
Before the update everything was running smoothly for over more than a year.
I don't see anything weird in the logs of the App itself....Anybody have a clue what happend here??
-
Update:
I did a sign-out (logout) on the android app and with that i got an error message saying,
"Unrecognized certificate with unknown fingerprint: CN=*.mydomain.comWhere "mydomain = my actual domain name redacted for privacy concerns"
-
@dantheman can you check if you have configured the domain via the Cloudron dashboard correctly as mentioned in https://docs.cloudron.io/domains/#matrix-server-location ?
That feature is not very old so maybe you haven't done that yet.
-
@nebulon
Check!!The Matrix server location is in place as described in the docs.
I double checked it, just in case.... -
@dantheman The following has been a common overlooked step for others before, from https://docs.cloudron.io/domains/#matrix-server-location:
Requires app on bare domain In the above example, an app must be installed on the bare domain https://cloudron.club for Cloudron to be able to respond to well known queries. If you already had an app installed on this location before Cloudron 6.1, you must go to the app's location section and click on 'Save' (without making any changes). This makes Cloudron re-generate the nginx configuration for the app which is required for this feature to work.This is setup too?
A life lived in fear is a life half-lived
-
@scooke
I have done that a while ago when federation was not working in the beginning. I have checked if it's still on the bare domain today, and it is.....I think it's something with the certificates of my domain and the fingerprint that's changed, after updating my Cloudron instance...
But my technical expertise isn't good enough to look in to that....
-
@dantheman There have been cert-related issues brought up in other posts, so this is probably related. I hope it gets figured out soon!
A life lived in fear is a life half-lived
-
@scooke
Man!! Thanks a lot for the info
I was really wondering all day long, what could have been wrong with my setup.
Hope they fix it soon and my problem is related to this.... -
@dantheman If you search the forums related to 6.3.3 update errors, there are a few posts suggesting how to check the certs, how to get rid of some. It certainly looks like this could be the issue for you.
A life lived in fear is a life half-lived
-
@scooke
Yeah i noticed the posts a couple of days ago, while i was reading the forum.
While i thought 6.3.4 was marked as a stable release, i thought i gave it a go.What should i do at this point?
Maybe downgrade my vm (where cloudron live's) on my system and revert back to 6.2.8?
And then disable automatic update's, till the next release (bugfix) come's out....Because everything was working till this morning, before the update.
-
@dantheman I don't think you can downgrade. You can restore from a backup, which will take time. And only if you have a full backup, and not one with an app here and there.
I don't want to send you down the wrong road, but I reissued my certs in the Domain & Certs panel - I only had two apps. I was trying a few other things too, so I can't be certain reissuing the certs is what fixed my borked box (a ghost app and Mastodon), and things worked after that. I also restarted a few times in the middle of all this. I should have kept better notes!
A life lived in fear is a life half-lived
-
@scooke
Actually i just did a restore from a fully created backup made last night.
Everything running smoothly now and what a relief reverting back to 6.2.8.I have around 8 apps, like Nextcloud, Matrix, Elements, Radicale, AdGuard, Bitwarden, Collabora-office and Rainloop.
All is running on a Unraid server Amd-ryzen 5 and 32GB of mem and live's in a Ubuntu 18.04 VM
server is serving around 20 a 30 active users 
-
Would indeed be good to get behind this issue, I don't see this after the update in the two matrix instances I am running. So this is hard to get further leads to investigate. Anyone else saw this happening?
-
@nebulon
At the moment i'm back on version 6.2.8. and everything is running fine.
But i am a little bit worried about updating at this point and i can't stay on 6.2.8.Maybe it's better for now to stay on 6.2.8. and wait for a new release coming up? And then try update again?
-
@dantheman to be honest, without us knowing the actual issue, we won't be able to add a fix to any new release, so just waiting will likely not really help here
I guess we have to debug this issue on your server if it happens to find out hot to fix it. -
@nebulon
I totally agree on that..If you're ready for debugging the problem on my server, just give me a sign and i will enable the support access on my instance...
-
The certificate issues were only in 6.3.1, but since you updated to 6.3.4 this should not be an issue. How about we wait a bit to see if any other user reports similar errors before you update again?
-
@girish
I'll wait till further notification and won't update in the meantime. I also noticed that my Nextcloud users are not auto syncing there mobile picture's from mobile. Manually uploading files in Nextcloud works. I don't know if this is related, but besides that everything is running ok for now. -
@dantheman I don't think this is related, but I have seen this as well personally and it was a problem in the mobile app on Android though. It was resolved for me by re-enabling it.
-
@nebulon
Thanks for the tip.
I will instruct my users to re-enable sync again on the Android app.
