Date: Sat, 21 Sep 2024 04:28:02 -0500 From: Scott Bennett <bennett@sdf.org> To: pete@twisted.org.uk Cc: freebsd-stable@freebsd.org Subject: Re: 13.3R's installworld killed system--please help! Message-ID: <202409210928.48L9S2YO015150@sdf.org>
next in thread | raw e-mail | index | archive | help
On Mon, 16 Sep 2024 23:39:53 +0100 Pete French <pete@twisted.org.uk>
wrote:
>On 15/09/2024 06:23, Scott Bennett wrote:
>
>> Thank you very much for this suggestion! It hadn't crossed my mind,
>> likely because I thought I had thrown all those old CD-Rs and DVD+/-Rs away
>> when I moved to a different apartment in February. After seeing your message,
>> though, I went looking. In the very first place I checked, lo and behold, I
>> found a DVD-R I had labeled "PC-BSD 8.2 (64-bit) Installer, LiveCD, and Repair
>> Disk" and "PCBSD8.2-x64-DVD.iso Disk 1 of 1". 8-D So I then did as you
>> suggested. Unfortunately, it made no change in the resulting boot behavior.
>> :-( At least I now know I have such a disk at hand to try such things.
>
>Am gald you found the disc - I was also going to say that I am quite
>happy to burn one for you and post it, though thats a slow method of
>acquiring a CD! I remember posting someone in the US a copy of Minix
>on 5.25 floppies back in 1989, and it took a while to get there, but
>did indeed boot on arrival.
>
>Am dissapointed that the updated boot sectors didnt help though. I
>would have bet money on that being the issue (and lost the bet!)
>
>
>> Thank you both for that reassuring information. After all this, it is
>> now clear that the boot code was not the problem and that I *still* have no
>> idea what went wrong. I do not remember ever having an upgrade from source
>> actually fail before this experience. Even the trickiest one many years ago--
>> a merged procedure to upgrade from i386 to amd64 in place and from, I *think*,
>> 9.x to 10.x, went well. (Trust me, I was as nervous as I would be on a
>> non-precision approach in nighttime IMC in a non-radar, mountainous environment
>> with flashes of light around me (yes, that happened to me once), but I had
>> planned all the steps carefully, and my combined procedure was successful.)
>
>!!!! umm, yeah, I really would not like to try that! not that I would,
>never having got an IMC rating, but the little bits I did with foggles
>on convinced me that this was not the kind of flying I wanted to do ;)
>
I understand. Many creepy tales from issues of _NTSB_Reporter_ loomed in
the back seat and outside in the cloud in the dark. 8-|
>
>i386 -> amd64, however, I did that, and that worked fine, despite also
>being very nervous. I;ve only ever done source upgrades, going right
>back to FreeBSD 3, and the only times it failed to boot were when I
It has been a long time, but I think I upgraded from 5.2.1, which was my
starting release of FreeBSD, to 5.3 to 5.4 to 5.5 to 6.1 by binary upgrades
with sysinstall(8). I had a dual-boot setup at the time with Win/XP. Then I
upgraded from 6.1 to 6.2, also with sysinstall(8) but screwed something up and
decided to re-install from scratch for some reason. (My memory of the time is
foggy because I had 5 or 6 broken bones and was heavily doped up for several
weeks.) In the process I apparently gave an erroneous partition specification
to sysinstall and managed to wipe out the WinXP partition, whereupon I decided
that I was dangerously fed up with WinXP anyway and would live more happily
without it. Since that time I've always done upgrades from source on my main
machine. As noted earlier, I did try freebsd-update(8) on the laptop because I
run a GENERIC kernel on that machine and that machine is even slower than this
tower for running buildworld. As also noted earlier, I learned to ignore that
freebsd-update(8) even exists and that it is basically a direct route to way
too much repair work and general grief.
>forgot to upgrade the boot code for a newer ZFS pool.
>
>> rolled every file system back to that snapshot. After reinserting the drives
>> into the tower, I booted it and ... my 12.4-RELEASE-p2 system was up and
>> running again. What a relief!
>
>Aha! Fantastic!
>
>OK, so, you rolled back the filesystems .... but left the boot code
>intact ? So this is now running your old filesystems but booting
Yes.
>using the updated 14.2 code that you wrote using PC-BSD, yes ?
>
Not 14.2, but 13.3. I have yet to do anything with 14.x, except to switch
my src tree to the releng/14.1 branch.
>> So I'm back to where I was before attempting the upgrade. It's a good
>> system, but it is out of support, so thank you very much to everyone who
>> responded anyway. I am pondering what my next step should be.
>
>
>OK, I have forgotten the start of this thread, but you went from the
>last version of 12 to a build of 13.0 release, which you compiled under
13.3-RELEASE-p1, not 13.0. I don't do .0 releases. I learned at least
35 years ago not to bother with those.
>the installed 12?
>
Yes.
>How far did it get in the boot process - did it even find the pool and
>try and load the kernel, or not even that far? If it is now booting off
It asked for the GELI passphrase, I entered that, it began to "turn" the
cursor a couple of moves, then began moving the cursor with long delays between
moves to various locations up and down along the lefthand side of the screen.
After several moves the cursor stopped, and that was as far as it went. So I
*think* it successfully completed the verification of the GELI passprase, but
whether it then was able to understand the pool or find a valid kernel in that
pool I have no idea. All I know is that somewhere early in the normal train of
events it goes off the rails and stops.
>the installed latest boot code, then we know it can run code which
>should find the pool.
Yes.
>
>Do you have the 'bootfs' property set on the pool ?
>
AFAIK, yes. I didn't change it while doing any of this, so it *should*
be set properly still. Likewise for the GELI setting on the partition that
holds the "system" pool, which is why the boot code asked for the passphrase.
>
>This is a puzzle - I've done this repeatedly, going from 3 all the way
>to 14, and its always worked.
>
Yes. It's the first time I've seen an upgrade from source fail, too.
Two different fiascos with 13 are enough to scare me off and send me on to 14,
bypassing further contact with 13. My tentative plan is a source upgrade to
14.1-RELEASE or whatever is now the current patch level of 14.1-RELEASE, but
I'm waiting a bit longer to see whether anyone on the list has reasons to
offer as to why that route would be inadvisable. If nobody does, then I'll try
to get to that fairly soon.
Scott
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?202409210928.48L9S2YO015150>