Rdiff-backup in Hetzner (via sshfs) taking too long?

Question

linux

I had rdiff-backup configured via sshfs on my server in Hetzner. He worked, everything suited me. But as the sites grew, the backup time increased from one minute to several hours. In a few hours, the disk with the backup manages to fall off and rdiff drops out with an error.
The backup disk is mounted like this:

sshfs#[email protected]:/ /media/backup fuse defaults,noatime,workaround=rename 0 0

Now we managed to make everything work by throwing out one of the sites.
The copy size is 6.4 GB, the creation time is 20 minutes, the increment time is less than a minute.
And everything would be fine, but if I add a large site that contains ~ 20 thousand pictures for ~ 20 GB, everything dies again.
How to be? Is this behavior normal or, perhaps, I have a screw covered, for example?
The screw passes the long test in smartctl. There were some errors, but the test passes with the message "Completed without error".

Answer 1

I use tartarus (with a preliminary database backup) to backup the Hetzner Dedik to the Hetzner backup server. The size of the (compressed) backup is more than 55GB, the increment is 3-5GB. Works very quickly.

Answer 2

Try making a backup with rsync. should be faster. In general, the behavior of the system causes concern, it should not slow down like that.

Answer 3

just add a 20GB file instead of this site and compare the speed. Most likely the number of files affects.

Answer 4

rsnapshot works with a bang over ssh, only before running it I take a snapshot of the file system for file consistency. I have about a million files.

Answer 5

Write to the hetzner support that the backup server is running slowly. They will change without problems. You won't need to reconfigure anything. They will overwrite the data themselves.