1. 49

  2. 16

    $18/year is probably cheap enough that you don’t care, but FWIW, Backblaze B2 is about 3x cheaper and works well with tools that don’t need to run on the server, like restic, but borg won’t work because it needs a server-side client. I don’t know how restic and borg compare for ease of use.

    1. 6

      I switched to restic for a few years after using borg, because of the cost savings. It’s pretty good and works with any old S3 host, I use wasabi.

      I have gone back to borg because it is so much quicker when it comes to indexing and pruning old snapshots. Last I heard restic had a fix for this in the pipeline, so perhaps this has changed.

      1. 2

        I don’t know how restic and borg compare for ease of use.

        Me neither, but one thing I really like about restic is that it lets you FUSE-mount backups as a filesystem, which is really nice for restoring specific files from a particular date. (Edit: just saw that Borg supports mounting too.)

        I am also a happy user of restic + B2 (have been using it for several years).

      2. 6

        Between this and the recent home-manager post, I think I’ve found a new blog worth subscribing.

        1. 14

          That’s very kind, but you’re in for disappointment! I hardly ever write anything.

          1. 16

            all the better, less cost to subscribe

        2. 5

          I’m in a similar boat to the author for backups so I’d like to give the setup I have:

          I have an external hard drive that sits on my desk. I use the Vorta GUI for borg which is actually made by BorgBase. I use Vorta to backup to my external drive every hour. Every four hours, I have a cron job that syncs the local repository with a Backblaze B2 bucket:

          12 */4 * * * /home/magikid/bin/b2 sync --delete --replaceNewer --threads 4 /media/magikid/Backup\ Behemoth/borg-repo b2://my-b2-bucket-name/borg-repo-name

          That gives me quick local restores if I need it but also the safety of having everything up in the cloud.

          1. 4

            I realised like 4-5 (or more?) years ago that most my work and hobby data is in some git repo or other, most of my docs and media are on a dropbox/google drive/whatever and I really don’t need to bother. And especially now that most media is also moved online.

            I was reading that “home server” post today and thought that it’d be cool, but I can never find a good reason to have a home server any more.

            1. 4

              If your data is scattered around a bunch of external services you don’t control, it could be lost if any of those services fail, decide to shut down, or decide to stop hosting your content for business or political reasons. That situation makes it more, not less, important to have backups that you control.

              1. 1

                Well, possibly. For work stuff - it’s not my job, so to say. I mean, it is, and I try to have the code on, say, github, and locally, and in the unlucky chance that both go irreversably away (I’ll play the lottery that week though), then so be it, but I still think it’s not my problem. For stuff that is not work, it’s not a big deal anyway.

                The only thing that is semi-relevant is media - family pics to be specific. There I both have a strong provider unlikely to go away soon (Google Drive/Photos) and a redundancy, as my wife has largely similar set of media in the Apple ecosystem.

                Everything else is a nice to have, but not really worth setting up and maintaining servers and services and teaching family to use them and replacing them as they go bad and all the other stuff.

            2. 3

              I can also vouch for borgbase if you need borg-supported cloud storage.

              1. 1

                I’ve been using Duplicati for a couple years, backed up to a local drive and S3. I don’t know how it compares to Borg but it seems to work well.

                1. 1

                  I wrote a tool to address some weaknesses I found in existing tools (Especially storing the decryption key offline).


                  I recently setup my home network with some secure backups using my tool and it meets my security requirements better than borg due to tighter access controls and the way keys are designed.

                  1. 1

                    Nice post! Where/how do you store the passphrase for automated backups?

                    1. 2

                      Thanks! The systemd unit that does the backups talks to pass to get the passphrase. It in turn relies on gpg-agent to not have to ask to unlock the password store. This works for me because I do backups during the day and my email client keeps the gpg-agent awake.

                      1. 1

                        Aren’t you stuck in a chicken and egg problem ? You encrypt your backups using a password, saved in a store. If you loose your whole $HOME, how do you recover ? You need the password, which itself need a gpg key, which is backed up, but encrypted right ?

                        Or maybe you backup your gpg keys and password store using other means ?

                        1. 2

                          Not certain if this is what they meant, but I assume the idea is that they both memorize the passphrase (in case recovery is needed), and also don’t want to keep typing it in for automated daily backups.

                          1. 2

                            Yes, except 1password has it memorized for me.