patched together workaround to the repo-locking nature of borg2

[SpiritInAShell]

tl;dr: not a request, not a howto, most likely not useful, just rambling about some experiments

BTW, I use proxy repositories to cirumvent the repo-locking, especially on a low-bandwidth remote storage. It looks similar to this:

• btrfs subvolumes
• rcreate --repo=local1.borg2 --other-repo=sftp-remote.borg2 for every subvolume
• simultanous pushing data to subvolume repo{1,2,3,4,...} from different disks or devices (directly or transfer, or export/import-tar)
  • doing snapshots of repoX to continue adding to repoX+1 while transfer (so some deduplication going on the local space)
• marking local repo as ready to sync
• (manually or) other task cycling through all "ready to sync" repos and doing transfer to remote storage.

Now I can prepare the data being sent to remote storage "pre-seeded", ready chunked and compressed.

Very fiddly and prone to errors. Needs much local space. Needs much time. Unneccessary high load on the local drive holding proxy repos. Limited local deduplication. Don't use this in production!

in borg2, as there is lossless export-tar and transfer in borg2, it is now possible to downstream from a central repo selected archives. With --bypass-lock on read-only snapshots this should also make parallel downstreams from remote possible. Locking nature of borg2 somewhat circumvented.

I wish borg1 could be enhanced to export-tar losslessly. Then exporting to the local repo which is associated to the remote repo would be faster.