Trying Restic

· 604 words · 3 minute read

I’ve been managing servers for probably two decades now. The first server I remember working on was a surplus server my dad brought home from work. It was our first domain controller at home and also operated as a file server. I’ve been working on servers ever since.

My current server is a surplus server from a previous job running an IT company with my dad. It used to be our file server and is a full size Intel tower server with enterprise hardware like redundant power supplies and a hardware RAID module. It’s pretty cool and I run a couple dozen docker-based servers on it.

Here’s my problems: that server is very large and power-hungry. I don’t like the large power consumption, and trying to move that beast across an ocean would be a hassle. It would have to go with my household goods which means I’d be without servers for months. My solution is to move to a smaller server, probably a raspberry pi form factor.

The problem there is my backup solution. Until now, I’ve been using CrashPlan. We got started with CrashPlan back in its heyday and still have access to their small business reseller platform. However, several years ago they removed their server backup product (which makes sense when you offer unlimited data), and enforced it by requiring a GUI on the computer being backed up. When I moved my server over to linux, I reluctantly installed Ubuntu desktop so that I could maintain my backups.

Well, I don’t want to waste any resources from a tiny new server on a GUI, which means I need a new option. I was reading a post by Jake Howard about backing up his servers and learned about Restic. I decided to try it out.

I don’t have a lot to say about the results, which is a good thing in my mind. It was easy to start with (especially stealing Jake’s Ansible Role for deploying it) and ran without pretty much any problems. The initial backup of my non-media backup set (about 290GB) took 48 hours, with the bottleneck being my internet upload speeds (about 20Mbps).

There were a couple of starts and stops as I figured out the backup set, and maybe that contributed to the overly large backed-up size. I backed up to Wasabi using a sub account of my old IT company (because they get a discounted reseller rate), and noticed unusually high storage usage. When I ran restic stats I found that the backup size was about twice that of my original data. I ran restic prune, which removes unused blobs and repacks the rest. So after about 14 hours of magic, restic had reduced the storage size by about 50%. I say about because the numbers don’t exactly add up so I can’t know for sure. The only downside here is that it took upload/download bandwidth to run the prune and Wasabi will charge for a minimum of two months for any file, so I’ll end up paying a couple dollars for the overage.

Now I need to back up my 2TB+ of movies. Maybe I’ll look into upping my upload bandwidth so it doesn’t take a month.

Some stats for your perusing.

Pre-prune usage:

to repack:      1777345 blobs / 544.407 GiB
this removes    1338612 blobs / 352.599 GiB
to delete:            0 blobs / 25.703 MiB
total prune:    1338612 blobs / 352.624 GiB
remaining:       687474 blobs / 288.519 GiB
unused size after prune: 0 B (0.00% of remaining size)

Post-prune usage:

Stats in restore-size mode:
Snapshots processed:   26
   Total File Count:   9565067
         Total Size:   296.204 GiB
Got thoughts? Share them by email.