Re: Upgrading from 10.0 to 10.1

[RVP <rvp%SDF.ORG@localhost>]

On Thu, 2 Oct 2025, Martin Husemann wrote:

> On Thu, Oct 02, 2025 at 09:50:03AM -0400, Greg Troxel wrote:
> > But really the manual method is just
> >
> >   unpack new kernel and modules
> >   reboot
> >   unpack rest of sets except etc/xetc
> >   unpack etc/xetc someplace else
> >   manually merge the new etc/xetc into your etc
> >   reboot again
> >
> > where manually merge could be
> >   - "diff -ur"
> >   - think about each diff, and
> >     + decide to keep it
> >     + decide to reduce diff to new sets by applying delta to etc
>
> I would add three points:
>
> 1)	run postinstall(8) pointing it at the extracted etc/xetc sets
> 	and /, e.g. I typically do:
> 		postinstall -s /tmp/etc -x /tmp/etc -d / check
> 	when I have extracted the etc and xetc sets in /tmp/etc/.
> 	When it complains, run the command it suggest to fix things.
>
> 2)	cd /dev && sh MAKEDEV all
>
> 3)	certctl rehash
>
> ... where (3) is actually redundant nowadays as (1) is supposed to deal
> with it.

I would also add,

4) Copy the new EFI bootloader files to the NetBSD EFI ESP partition (if
   running EFI). (sysinst(8) updates the BIOS bootloader when you run it
   from an installation medium, but, doesn't for EFI boots.):

```
mount NAME=NetBSD-EFI /mnt		# <-- change this
cp /usr/mdec/*.efi /mnt/efi/boot/.
umount /mnt
```

On Wed, 1 Oct 2025, Liam Proven wrote:

> [...]
> I tried anyway. It ran for a long time -- hours -- and at the end I
> was still on 10.0.
> [...]
> Anyway, I edited the config file, attempting to specify 10.1, and ran
>
> `sysupgrade auto`
>
> It did a lot of work. I left it overnight.

I think this is where it all started to fall apart. This should _not_ take
hours. This, usually, is a sign that IPv6 isn't working correctly: either
the ISP doesn't do this right, or their dinky router is mis-configured.

The solution is to either,

a) Request only IPv4 addresses:

```
$ fgrep dhcpcp /etc/rc.conf
dhcpcd=YES
dhcpcd_flags="-4Mqw ure0 vioif0"
$
```

(edit /etc/rc.d and add `-4' to `dhcpcd_flags'); or,

b) Tell NetBSD to try connecting to IPv4 addresses first, when both IPv6 and
   IPv4 addresses are returned by DNS (with the IPv6 ones being "first" in the
   list.):

```
$ fgrep ip6addr /mnt/etc/rc.conf
ip6addrctl=YES
ip6addrctl_policy=ipv4_prefer
$
```

FWIW, I did an upgrade, just now, of 10.0-RELEASE to the latest 10.1-STABLE
using the installation media (.iso image in QEMU) and it all went smoothly. The
only manual step was running `etcupdate' after upgrade. But, this is a manual
step in any case.



I use this procedure to upgrade my NetBSD (I have a script for this, of course):

https://www.unitedbsd.com/d/820-updating-netbsd-too-embaressed-to-ask/4

Note that this is for moving along the same branch, 10.0 -> 10.1, etc. If you
go from 10.x to 11.x, then you'll have to: install the kernel, modules, gpufw,
bootloader (ideally); reboot into this; then upgrade the rest of the sets when
running the new kernel. (Of course, you don't need to worry about any of this
if you upgrade using one of the installation media.)

HTH,

-RVP