Liam Proven <[email protected]> writes: > On 01/10/2025 13:20, Stephen Borrill wrote: > >> On Wed, 1 Oct 2025, Liam Proven wrote: >>> I think I messed this up. >> >> Sorry you've had a hard time. > > Well, I can't lie, it was substantially harder than I expected. > > I'm not a rookie -- I started on SCO Xenix 286 in 1988 -- but I am a BSD > newbie. >> >> Agreed. While I've never used either method, the sequence of steps for >> upgrading a major release with sysupgrade explains what needs to be done: > > I blinked in surprise at this. Then how do you do it? This implies there > are more methods. >
[snip] Updating NetBSD in its raw form just boils down to: 1) Copy the new kernel to / or whereever it needs to be. 2) Put the needed modules for the new kernel in /stand/ 3) Boot the new kernel 4) Unpack the sets you want, or if you build everything from source, put the build artifacts in place. The sets are just compressed tar files from a distribution build. If you do a "build.sh release" you get a distribution directory with everything in it and can extract what you need from that. 5) Take care with /etc. Usually you won't want to unpack that for an update or copy it from a distribution build. The in core program postinstall takes care of getting that updated in the minimal sense. Or you can use the in core program etcupdate if you need something more. I also apply some caution to /var, /root and /dev. That was more words than actual action, I suspect. There are nearly an uncountable number of ways to perform the above actions, but it will pretty much all come down to boot new kernel and then unpack sets or install distribution artifacts, then run postinstall. I have personally never done anything except this, aside from using sysinstall (i.e. boot a ISO) once in a while to do an initial bare metal install for i386/amd64 systems. -- Brad Spencer - [email protected]
