Backup solutions: Cleaning up

[haggholm]

Having adopted SpiderOak to manage files that are too large to conveniently manage in my version controlled home directory, as well as files with data sensitive enough that I’d rather not put it on the same server as my webapps, the time has come to finally get off my arse and clean up that mess.

While there were files that were always clearly too large to commit to my repository and were only ever backed up remotely once I adopted SpiderOak as a solution, there were also files that I would now clearly backup with SpiderOak, but which I then committed to version control. There are also some files sufficiently sensitive that I should not have committed them, but I did… Time at last to delete them.

Of course, deleting files from most version control systems is non-trivial. My files sit in Subversion. To delete them, I have to dump the entire repository into a dump file, run this file through a filter to exclude specified paths, create a new repository, and import the filtered dump into this. A bit cumbersome, but ah well—at least it works. And it seems that, preliminarily, I am able to shrink the repository size by at least 54% 71% 78%.¹

While I’m messing around with this stuff, I just might switch to Mercurial. Hmm…

---

¹ From an original, hideous 2.4G to a somewhat less awful 1.1G 710M 521M. The Hg repository weighs in larger at 1.6G 965M 743M, for some weird reason (still about 33% 60% 69% smaller than the original subversion repository). —All as measured by `du -sch`.