Comment by hosteur | Hacker News Reader

Comment by hosteur | Hacker News Reader

Documents are in Subversion hosted on a local home Debian server which is also NAS.

Photos (hundreds of GBs) are also in Subversion hosted on same server. Photos are managed/viewed either via Digikam (on both Mac, Linux, Windows) as well as through a generated HTML/js gallery served on the same server.

Home server is backed up via restic to

* local harddrives that are rotated weekly and moved offsite. * to a rented physical Debian server in a different country * to a cloud storage provider

Server storage is fully encrypted using LUKS. Backups are encrypted via restic/ssh.

Various sensitive documents are separately encrypted inside the svn repos via 'age' or similar tools.

This setup has been working for almost two decades. I have yet to lose any data.

Edit: For documents, we have a personal repo for each family member. Typically, we use a folder structure based on https://johnnydecimal.com/

We also have a repo specifically for receiped in which receipts are just stored as /<year>/<yyyy-mm-dd>-<name>.pdf (or .jpg if it is a scan/photo).

Legal documents like wills, etc. are stored in a safe as well as a physical copy stored at a relative + digital versions stored in the svn repo.

Honest question: why do you use a VCS to store the documents? The kind of documents OP is talking about don't get updates, same for the photos.

Great question. The reason is that my documents do get updated. This is not merely an archive as I also store various more live documents in it like budgets, etc. Some documents are indeed purely archived. But even for those, I find it really handy to have a complete log of when they were added. Who added them, etc. All this is stored in SVN. Also, SVN provides a really easy way to sync the folder to various devices (svn CLI tool on Linux and Mac, and TortoiseSVN on Windows clients).

I am aware that nowadays, SVN is not the modern choice compared to Git. When I designed this system, Git was not available. But I actually find that SVN's centralized architecture fits really well for this purpose.

For photos, I really like having them in SVN as I never lose any image even if I modify it. Sure, it takes up a little extra storage. But storage is cheap. Also, SVN here acts as an extra local backup.

I've used SVN in a similar capacity for personal and business use, and even at some of my Customers. It's very handy and TortoiseSVN provides a good user experience. I got into doing this back in 2004 and even though Git might be a better tool today (maybe) I've stuck with SVN.

At a previous employer, we stored 3Tb of data in SVN. The data was mostly archival, however sometimes corrections were made to historical data. Using partial checkouts it was quite easy to sync a subset locally. I don't use subversion anymore, but thinking about it I prefer the explicit and deliberate commit/checkout/merge over the automagic sync conflict generators that are onedrive, dropbox, nextcloud etc. Also all pf these start to struggle somewhere around 100k files / 1TB data mark.

Reply on news.ycombinator.com