2
My setup is as follows:
- Two Linux machines running Debian 9
- A "storage box" on the internet rented from a cloud hosting company. This box can only be accessed via the SFTP, Samba and WebDAV protocols for transferring data. I have no means of installing any software on it, since it is not a server, only a "dumb" storage device. I do not have any SSH access either.
Now, here's what I would like to do:
- Synchronize a directory across my two Debian machines.
- Do this incrementally, so I don't have to upload and download the entire directory structure every time.
What I've tried so far:
rsync: Basically
rsync -avz /path/to/dir/local/machine USER@remote_address.com
. The problem: The remote storage box does not haversync
installed to act as a server. Therefore, the entire directory has to be parsed (and therefore transferred) every time, which is slow.WebDAV mount: Mounted remote directory locally using
fusedav
and did anrsync
locally. This took too long since modify times are not preserved when uploading viafusedav
sorsync
has to checksum each file.Duplicity: Incremental backups work fine over SFTP. There is, however, no possibility to do an incremental restore over existing files. It downloads the entire directory each time.
Borg Backup: Same issue as with duplicity, no incremental restores (that I found).
Is there any backup solution that offers incremental restore and works over SFTP? Or can I make rsync work quickly (I thought of some kind of cache, though that doesn't seem to be an option.
What about
– Pimp Juice IT – 2018-02-27T21:17:30.977rsync -h -v -r -P -t source target
or consider testing with Rsync--ignore-existing
parameter as that post suggests. Or look for an alternative non-Rsync solution such asfind . -newer xmrkite.date /path/to/backup-dir -exec scp '{}' user@host:'{}' \; && touch xmrkite.date
.... Quick research and ideas for you!!