Looking back through Borg archives


· backups ·

I've run about three different[1] sorts of snapshop backups I think, in the last 15 years or so. Firstly a dd-based script that hardlinked date directories on a series of (failure-prone) USB hard drives. That was reasonably simple, and I learnt a bit about some useful things, but in about 2015 I thought I'd better not rely on myself for something so critical and switched to SpiderOak. They were all the rage after Edward Snowden recommended (or just mentioned? I can't remember) them as a good encrypted and deduplicating backup system. They were good, for ages. But then a couple of years ago I got sick of their weird client's UI and opaque bugginess — and the fact that they no longer were actually promoting their backup product on their own homepage. So I switched to Borg.

Borg is good, and I do love the fact that the basic configuration is about setting it up how you want it to work. In my case, it's a manually-run thing because I realised that the rate of change on my local machine is not great enough to justify the continual scanning of all changes (for some parts where I do want that, I use NextCloud). But the part of Borg that I do not have working as I want it to is the spelunking side of things: when I want to retrieve a file or directory from an archive, I have to do my own bisecting of where the version I want can be found. That's partly unavoidable, because of the nature of restores like that — it's not really the business of the backup software to know what you want. But it could be quicker: mounting archives to date-named directories currently requires me to copy and paste the name of the archive. I'd rather have some sort of menu thing.

(Update, July 2024: I think Vorta is what I was looking for.)

  1. I say 'about' because I've experimented with about three dozen different systems.