Egregoros

Signal feed

Timeline

Post

Remote status

So, I did something silly. I started backing up my primary #FreeBSD desktop after .. omg nearly 30 years. It was first installed in ~1996 and has been continuously, incrementally upgraded since then. Hardware, OS upgrades, CPU, etc.

Anyway, I was going to try and convert it to use #zfs boot environments. That's something that can't be done in-place on a legacy installation from ~1996. It's using ZFS for years but the FS layout doesn't allow a trivial rename etc in single user mode.

Then I realized.. wait, I've got important stuff on there. It's not being backed up regularly. This might not be a great idea.

So, #zrepl to the rescue! So now there is a second copy. No, a restore has not been attempted but I am reusing a process that I have tested before. Close Enough(tm), right? Nothing Could Go Wrong With This Plan(tm), right?

(Why zfs boot environments? I'm working on fixing some really annoying FreeBSD pkgbase, pkg etc limitations/problems and boot envs would be damn convenient! And of course, backups.)

Replies

2
@dvl I saw your blog posts on it. I have options. It’s just a risk vs time vs effort vs learning something trade off. I’ve done far worse things remotely. You haven’t truly lived until deploying in-place headless consoleless FreeBSD->linux and Linux->FreeBSD converters. In this case I mostly need to be careful about what data belongs in the BE and what doesn’t.