Borg backup stops mid-way, not resume

ā„¹ļø Support šŸ“¦ Appliances (Docker, Snappy, VM, NCP, AIO)
,
1

Hi,

I am trying to use Borg Backup for my Nextcloud AIO instance. The backup size is quite large (>2 TB), and as a result, I frequently encounter ā€œbroken pipeā€ errors when backing up to a remote server.

According to the Borg documentation, the backup process should resume from the last checkpoint after an interruption. However, this doesnā€™t seem to be the case with AIO. Is my understanding correct, or is there a way to ensure a seamless backup process?

Any advice would be greatly appreciated, as my services have been offline for about a week. Each time the backup fails, I have to restart the process from scratch.

Thanks in advance!

2025-02-27T10:10:47.340897779Z Remote: client_loop: send disconnect: Broken pipe
2025-02-27T10:10:47.360214537Z Connection to remote host is broken. Broken Pipe
2025-02-27T10:10:47.422780436Z Deleting the failed backup archive...
2

What BORG-environment does the remote server supply? How do you connect?
What sends the message

Deleting the failed backup archive...

and deletes the archive?
I think resume from checkpoint is only possible with the same archive.
Does borg list show any checkpoints in your repo?
In my experience itā€™s often better to do the inital borg-backup manually if you have large backups.
GOOD LUCK!

3

Thatā€™s exactly the point. Maybe the container is programmed to delete the archive in case of failure. If thatā€™s the case, Iā€™d like to see a fix for this in the future.

Iā€™ll try running it manually and hope the AIO reads it.

Should I initialize it from the nextcloud-aio-borgbackup container?

4

I connect via ssh and pub-key. Can you maybe provide me the instruction?

5

whatā€™s the borg repo uri? itā€™s sth that you (can) pass to borg as BORG_REPO=
eg
BORG_REPO='ssh://user@host.tld.org/path/to/repo'
in this case you use user with ssh-key-auth and borg is installed on host.tld.org (and runs as ssh ā€˜borg serveā€™ )

I think itā€™s best to init the repo from the host that uses it because when multiple hosts use one repo (which is possible) sometimes their cached configs can get in the way and require manual intervention. Be aware that init (preparing an empty repo) is not the same as creating the first archive in this repo.

Sorry, i know (next to) nothing about containers an avoid using them wherever possible.
GOOD LUCK!

6

Thatā€™s the issue. The Nextcloud AIO is container based!