Cloudron Forum

No, I wanted to get confirmation first that I am supposed to do that. I will do that then and try to restart the backup.

When in Roundcube, I get this error when I open a folder:

Server Error: UID THREAD: Internal error occurred. Refer to server log for more information. [2025-04-01 05:11:13] (0.194 + 0.000 + 0.193 secs).

There is nothing in the logs, though, neither for the Roundcube app, nor for the server.
The email logs are a bit verbose, what element are we looking for? Then I can check.

I believe this issue might be related to the full text search of e-mails, which I have activated.

By checking the email logs even further, I got similar error messages, although not related to the action I was performing:

Error: Mailbox INBOX: Transaction commit failed: FTS transaction commit failed: backend deinit
Error: fts_solr: Indexing failed: connect(127.0.0.1:8983) failed: Connection refused
Error: Mailbox INBOX: Precache for UID=1001 failed: Internal error occurred. Refer to server log for more information. [2025-04-01 05:22:17]

My Cloudron server backup times out regularly at this point:

Removing directory /mnt/cloudronbackup/snapshot/mail (mail)

The logs say this:

Apr 01 10:30:00 box:tasks update 10166: {"message":"Removing mail backup 2025-03-25-010000-657/mail_v8.2.4"}
Apr 01 10:30:00 box:tasks update 10166: {"message":"2025-03-25-010000-657/mail_v8.2.4: Removing directory /mnt/cloudronbackup/2025-03-25-010000-657/mail_v8.2.4"}
Apr 01 10:30:00 box:tasks update 10166: {"message":"Removing directory /mnt/cloudronbackup/2025-03-25-010000-657/mail_v8.2.4"}
Apr 01 10:30:00 box:shell filesystem: rm -rf /mnt/cloudronbackup/2025-03-25-010000-657/mail_v8.2.4

I have upgraded already to version 8.3.1.

How can I handle this?

None of the Kanban apps Cloudron offers has proper mobile support. I'd say any new suggestion should have that to be enough of a significant improvement over the existing ones.

Planka looks nice, sure, but it doesn't seem to offer a proper mobile layout or an app.

Maybe. But I think I'll just take the extra GB into account and see dynamically if and when issues arise. So far, nothing has been missed.

Old server before moving, vs. new server after moving, and I updated the graph after the process was done

I am using a volume for Cloudron and it is mounted. However, when I go into the file manager under Volumes, it shows only two folders I manually created there, which are empty. However, when I enter the folders from the console, all files are there. System Info shows the correct amount of data used.

This might be a permission issue with the mountpoint, so let's see, which one do I need?

Could be! But how should I resolve that? delete2duplicates has actually found a few thousand e-mails, and everything seems to be there, but I still have about 20GB more than before.

Unfortunately, I still have an issue left after restoring my server today.

When I restored my cloudron, e-mails were only partially restored. I then used imapsync to finish the job, which it did - however, I am relatively certain it didn't recognise the emails that were already in the account and copied them again. Now I have a mix of e-mails, they are apparently all there, but I have 20GB more of those that will in the future slow down any backup if I don't do anything about it.

What I have already tried:

• using delete2duplicates --useheader "Message-Id" in imapsync
• imapdedump (a python tool I found for that purpose)

They deleted a few duplicates, but nothing significant. I still wonder, one user account has 80% more data, I don't really understand what else could be the reason.

I would ideally want to avoid restoring from backup now, because over the course of the day, there were already e-mails coming in that would be lost then.

Would it be possible to update the docs for using it with a volume?

I find two points helpful to add:

1. Inside Syncthing, the path is /media, although the mountpoint is /mnt (I mixed that up previously and it took me time to figure this out)
2. The mountpoint needs to have 777 rights

I'd appreciate that!

To reproduce:

• Go to E-Mails
• Set one or many aliases for an email account
• (save)
• change the account owner
• save

Expected Behaviour: Aliases stay in place
Actual Behaviour: Aliases disppear

I think this should be fixed.

... one short question remaining, if I enable backups for my private docker registry, will the images be saved then? For another restoring process, then my custom apps could be restored together with the backups, which would be amazing and reduce the custom steps involved.

Thank you, this has worked fine!

@girish said in Restoring custom apps:

@ekevu123 said in Restoring custom apps:

My expectation sincerely was that a custom app would be restored similarly to all other apps by restoring from the backup source.

tangent... but as @joseph explained, this expectation is not correct. I wonder .. did we put something in our docs to mislead? Or maybe we can put something in our docs to clarify. In Cloudron, app backup === data generated by app, nothing else. This is why the packaging format and deployment is based on readonly filesystem. Otherwise, we have to back up the whole container for each app all the time! This would be infeasible. AFAIK, no PaaS backs up source code. If they roll back it's because the image is available via their registry (and not part of backup).

Okay, I understand that - then it was just bad luck that I used a private registry on the same server.

@joseph said in Restoring custom apps:

@ekevu123 said in Restoring custom apps:

My expectation sincerely was that a custom app would be restored similarly to all other apps by restoring from the backup source.

Images are not part of the app's backup. Images are essentially source code. They are immutable after release/packaging. The assumption is that it's available always through some other means. Cloudron apps only get restored because the images are always available on docker hub.

Back to the current problem, the issue is the custom registry was not backed up? I guess you have many options:

• If you have the image locally, just push it to the registry. After that restore will work.
• Otherwise build the image again from source code and tag it with same tag and push it to the registry.
• Or take the image from the old VM and push it to new VM. This might be a bit complicated since you have to get the old VM / registry running

I would assume that if I use the tag the app is looking for, it should work? I can't use another tag, because then I would need to re-install the app and at least put the database dump back.

So, I will try that first, using the same tag as the app which errors out now is looking for and I hope this will re-install the app including the data I previously had from the database. Right?

As I said, in the private registry on the same server.

My expectation sincerely was that a custom app would be restored similarly to all other apps by restoring from the backup source.

Here the issue was that the e-mails have been moved to the new cloudron server only partially, and the new folder wasn't created in this case, so when I tried to move this in Thunderbird, it failed.

So, this is solved now

And following up on that - how do I move a custom app to a new server properly then?

Currently, I am trying to point the app back to the older server and hope that it still works as cloudron is still installed there, although I have moved the instance to a new server.

I tried to move cloudron to a new server, however, not all emails have been moved to the new server. I think I'd use imapsync now to move all, but how do I set the access to the previous account to fetch the emails from?

EDIT: What is also not working: For the emails I have, when I try to move them between folders, they just move back. So, it seems like I can send and receive emails, but somehow not move them, which is odd.

Apparently, this has resolved itself, except for other, more specific issues, but I will open new threads for these to resolve them separately.

I have just realised that when I am trying to restore a custom app which I previously installed manually on cloudron, it will try to pull the image from docker instead of the backup. But why? If I initiate a restore process, I would assume that the app should be present in the backup and could be restored from there?