Client -> Backup Server -> Cloud Storage

[khmarbaise]

Currently I'm in the test phase using borgbackup. The question which comes into my mind is the following:

I have create a backup from my client to a server. Now I would like to backup the server to cloud. I simply thought to do a borgbackup from the backupserver to cloud as done from client to backup server...

My current setup is doing a borg backup from client to backup server via:

borg create --verbose \
	--stats \
    	--filter AME \
	--show-rc	\
	--compression lz4....

and now I think to use compression from the backup server to cloud would not make sense from my point of view?

[justin2004]

when i have a borg repository in a borg repository i also compress the outer one.
but my outer borg repository contains more than just an already compressed borg repository.

[ThomasWaldmann]

maybe read the docs and esp. the faq, there are some infos about this.

[khmarbaise]

Maybe my question was not accurate enough.

Client (using compression lz4) -> Backup -> Clould

The question was: if it would makes sense to use another compression for the path from Backup -> Cloud...

Also I would create the backup from Backup to Cloud explicitly via borg create ...

Furthermore the FAQ only talks about :

client machine —borg create—> repo
repo —copy/sync—> copy-of-repo

But the scenario is given by copying (synching) the borg repository (in my case from the Backup server) to cloud which I already decided against this approach.

But I will have to have an encryption between my Backup Server and on the cloud.

[ThomasWaldmann]

iirc the faq also talks about error propagation, which would also apply to your scenario.

usually, already compressed and especially encrypted content (== all borg segment files, REPO/data/** in your case) compresses badly or not at all (could even add a little space overhead due to compression).

in a borg repo, there are some additional files, esp. the repo index, which are not compressed/encrypted and thus might compress ok.

Still, as the main data volume is in the segment files, I don't think using compression is worth it for backing up an already compressed (and/or encrypted) repo.

but faq recommends doing 2 backups directly from the client anyway to avoid error propagation.

[ThomasWaldmann]

e.g. one case when you need your cloud backup is when your local repo is corrupted somehow.

but then, the latest backups in the cloud will just have the same corrupted content.