Re: FreeBSD 15.2-BETA1 Now Available

On 11/10/26 13:00, vermaden wrote:
> Hi,
> 
> 
>> "Enji Cooper (yaneurabeya)"
> 
>>
> 
>> Hi there,
> 
>> Are you aware of `pkgbasify` [1] by chance?
> 
>> Thanks!
> 
>> -Enji
> 
> 
> Yes.
> 
> https://vermaden.wordpress.com/2025/07/20/freebsd-pkgbase-pkgbasify-tool/
> 
> 
> 
> 
>> "Tod McQuillin"
> 
>>
> 
>> Thanks vermaden, this looks to me like exactly the
> 
>> right direction to be moving in.
> 
>> pkgbase will have a hard time becoming "the" way
> 
>> with such a big administrative obstacle to upgrades.
> 
> Thanks.
> 
> 
>> "Tod McQuillin"
> 
>>
> 
>> This script seems like a great place also to integrate
> 
>> ZFS boot environments into the upgrade process as
> 
>> you've previously written about.
> 
> I have that in TODO.
> 
> 
>> "Tod McQuillin"
> 
>>
> 
>> Is this still best practice on systems booting from ZFS?
> 
> I use them all the time and do not have issues with them
> 
> and I do not see a point why someone would not use them.
> 
> 
>> "Tod McQuillin"
> 
>>
> 
>> One issue I frequently find with freebsd-update here is
> 
>> that old snapshots accumulate until they use up all the
> 
>> free space in the pool, and then I have to go and prune
> 
>> old snapshots.
> 
>> This aspect would also seem to be ripe for automation.
> 
> I usually do not use 'automatic' ZFS Boot Environments at
> 
> all - I create/manage them by myself - I even always configured
> 
> freebsd-update(8) to not create them because that only
> 
> created a mess - but some people want/need automatic
> 
> creation of ZFS BE before upgrade - so its better to just
> 
> leave them there for 'manual cleanup' then to be sorry
> 
> that they were not automatically created :)
> 
> 
>> "Bakul Shah"
> 
>>
> 
>> Add a check "pkg which /bin/ls" or similar and if it
> 
>> fails invoke freebsd-update. Now you have one tool
> 
>> for updating either kind of system!
> 
> Yes its possible - but I while pkgbase-update(8) does
> 
> everything in one step + reboot 

Any update/upgrade technique that does not reboot into the new kernel
before having some new world code become active risks the new world code
used failing for lack of kernel updates/upgrades it depends on.

It the same reason that the source upgrade document has that reboot
between the 2, the instructions being intended to allow for UFS. Nothing
about pkgbase of itself removes the technical issue.

Of course, UFS contexts are more constrained by not having alternate
boot environments as a way to not run the newly installed world at
all until rebooting after the world is installed.

While I use a ZFS system, I use a lot more UFS systems. (It is a
deliberate mix.)

> the freebsd-update(8)
> 
> does not - and often requires multiple reboots - so I
> 
> will just add 'INFO:' message that this is NOT a PKGBASE
> 
> system and that freebsd-update(8) is the right tool to
> 
> be used ... or pkgbasify(8) ... good point.
> 
> 
> 
>> "Ruslan Zalata"
> 
>>
> 
>> This is really great! Thank you for writing this tool.
> 
> Thanks.
> 
> 
>> "Ruslan Zalata"
> 
>>
> 
>> I wonder, what is so fundamentally wrong in adopting
> 
>> this pkgbase-update in base, symlinking it to
> 
>> freebsd-update by default for some time? FreeBSD
> 
>> users will have smooth transfer to new upgrade/update
> 
>> system. And, what's more important, we will not be needing
> 
>> to rewrite/review/invalidate tons of existing documentation.
> 
> Good idea.
> 
> 
> 
>> "h v"
> 
>>
> 
>> on a recent 15-STABLE (identifiies itself as 15.2-PRERELEASE)
> 
>> i get just
> 
>> # pkgbase-update upgrade -r 15.2-BETA1
> 
>> NOPE: installed '15.2-PRERELEASE' is newer then '15.2-BETA1' target.
> 
> The PRERELEASE is supported now - if new such issues will arise
> 
> ... and they will as this is 0.1 version - I will just fix them in next
> 
> iteration of changes.
> 
> 
> root@pre:~ # uname -prism
> FreeBSD 15.2-PRERELEASE amd64 amd64 GENERIC
> 
> root@pre:~ # ./pkgbase-update upgrade -r 15.2-BETA1
> INFO: upgrading from '15.2-PRERELEASE' to '15.2-BETA1' version
> Updating FreeBSD-ports repository catalogue...
> (...)
> 
> 
> 
>> "Dag-Erling Smørgrav"
> 
>>
> 
>> This is already in progress. Writing a tool that works
> 
>> for the trivial day-to-day case like Vermaden did is easy,
> 
>> writing one that works correctly across a wider range
> 
>> of circumstances takes time.
> 
> Sure.
> 
> This situation will probably be similar to mine earlier beadm(8)
> 
> project that I updated to the point that it was ready to be included
> 
> into Base System (and I still maintain it to this day) but instead
> 
> it was chosen to rewrite everything from scratch as bectl(8) tool.
> 
> It was funny to watch that sometimes bectl(8) had interesting
> 
> bugs which were even fixed by running beadm(8) to switch
> 
> the 'active' ZFS BE.
> 
> It will probably be the same this time - I will develop the
> 
> pkgbase-update(8) tool that it will work for 98% cases and
> 
> FreeBSD project will provide something else anyway - I am
> 
> used to that situation.
> 
> In the mean time - since yesterday - I already added updating
> 
> 'BIOS' bootcode with gpart(8) and support for UFS based systems
> 
> and I have another several things in TODO already along with
> 
> resting for various other cases.
> 
> 
> Regards,
> 
> vermaden
> 
> 
> 
> 

-- 
===
Mark Millard
marklmi at yahoo.com