@girish So after the upgrade to 7.2.2 the problem vanished. Closing the Ticket.
Thank you for the help!

Good point. I checked the services and some of them are yellow (without error, but stalled in loading). I've restarted one app that uses mongodb and now the backup works. But it's very close to the memory limit. If this happens again, I'll increase the memory assigned.

Thanks for pointing it out.

Found an issue: MongoDB needed up 600Mb of RAM (to be adjusted in services page).

This was sorted out by removing existing mongodb data and restarting mongodb. Not ideal solution for installation having existing data, but it worked for this instance.

@nebulon I was having a similar issue with Mongodb since a few days, I backuped the mongodb folder, created a new one, restarted the mongodb service, restarted the apps and the problem is solved 🙂

I found restarting Mongodb service manually fixes the error. Or maybe it was just a coincidence.

I am trying to only do this from the Cloudron hosted versions of Metabase/Grafana.

I followed your instructions and this worked perfectly, took the username, password, port, host, and database name from the environment variables within the Wekan web console, entered them in Metabase, and was able to connect and see the data.

Thank you!

@rossmaclean yes, once we publish the app in our app library we will move that repo. Usually the original packager will retain push rights if wanted though.

I guess if you want us to review a change, you could start an issue in that repo and tag us (@girish / @nebulon / ...) there. There is no formal process in place yet, so whatever works for you and notifies us 🙂

@nebulon I am going to pass on that one.
Any volunteers?

@JohnGuan If you want to backup NodeBB's mongodb, then go to the app's backup -> create backup. In the backup, there will be a mongodbdump file.

As for importing, please see https://docs.cloudron.io/guides/import-mongodb/

@robw @alex-adestech An update on this.

Crux of the issue is that the indices name generated are too long. @alex-adestech pointed out to me that this issue only sorted out in MongoDB 4.4 and not even MongoDB 4.2. Searching through the rocket.chat issue tracker I saw only person hit this and nobody replied. Search further, I found that Rocket.Chat doesn't even support MongoDB 4.2. The releases page mentions the versions they test with - https://github.com/RocketChat/Rocket.Chat/releases. This got me wondering why we hit this issue. Well, the answer it turns out is that Cloudron generates 32 byte mongodb database names for apps. This in turn ends up affecting the length of the index.

I have made a fix now for Cloudron 5.6 that generates shorter names. (We have had to do something similar for apps using MySQL as well in the past).

@girish Ah yes, sorry. Should've looked. At least maybe this will direct someone to the docs if they end up here again.

Admin credentials for mongo are really internal/implementation details. In general, do not make configuration changes to the mongodb instance because it may not work across all apps. This is the reason why all the databases are "locked" down and not exposed to users for easy access.

That warning said, you can do docker inspect mongodb | grep CLOUDRON_MONGODB_ROOT_PASSWORD to get the root password.