`restic copy` from disk based repo to rest-server explodes in size

I currently have a repository on disk, which is ~300GiB in size.

I prepared another partition with 350GiB in size and configured restic-rest-server to use that partition as data dir.

Then I did a restic init on the REST repo, copying the chunking data from my disk based repository.

Waiting several hours, restic copy started to fail with “400 Bad Request” and eventually cancelled the copy.

Checking the target disk revealed that it was full, after just ~10 snapshots had been copied. How can this happen? And how can I properly transfer the repository?

Adding another 50 GiB to the partition and making sure there was no other service writing ~20 GiB of metadata, while also making sure not to compare GB with GiB (hmpf), it finally settled and disk usage is actually as expected.

Though still, it seemed as if the target repository did indeed consume a bit more disk space during the process.

The copy command never ever deletes data. The prune command is effectively the only command in restic that can delete sizeable amounts of data.

That’s weird, then I have to check my monitoring