Upgrade disaster 11.2-RELEASE -> 12.0-RELEASE

_martin

Daemon

Reaction score: 414
Messages: 1,251

I've put that into my TODO list -- out of my curiosity I'd like to see where it does that (my hunch is that it's early in the boot, page tables are not set yet, using 32b part of regs only).

Yeah, do a backup for sure. It would be even better if you can test this in VM first. I'm using this approach when I want to have cleanly installed system ; basic installation is done from old running system. With ZFS on root you can even prep boot env without beadm.

When it comes to config I have this one liner to save etc:

find /etc /usr/local/etc/ -depth -print | cpio -oaV | gzip -c > `hostname`_etc.cpio.gz

But it's still nice to have all files at hand by some means (/old, external backup..).
 
OP
A

ahhyes

Member

Reaction score: 8
Messages: 65

I woke up this morning and had a thought.. What if I booted from the USB drive that has the 11.2-RELEASE image, and mount the hard disk filesystem, then overwrite the /boot/loader on the hard disk with the one from 11.2-RELEASE

Well well well!! That has resurrected my server!!

So the problem is in /boot/loader in 12.0-RELEASE
 
  • Thanks
Reactions: pva
OP
A

ahhyes

Member

Reaction score: 8
Messages: 65

So it boots normally now. So I am in two minds what to do right now... run freebsd-update and let it update the userland to 12.0-RELEASE or attempt a rollback. But I wont be doing either until I have made backups of the OS drive (in case things go pear shaped again).

So right now I have a 12.0-RELEASE kernel and 11.2-RELEASE userland. At least I have gotten access to my data again! :beer:
 
OP
A

ahhyes

Member

Reaction score: 8
Messages: 65

After backing things up, I decided to just run freebsd-update install and let the upgrade complete, since the likelyhood of the userland updates causing another boot issue are low. The upgrade has worked. I've bought the server back from the dead and its now running 12.0-RELEASE - but its still using the /boot/loader from 11.2-RELEASE.

So things will continue to work unless another update comes which overwrites that file again with a bad version. I hope if there is an update to /boot/loader in the near future, it is one that fixes the problem that caused this whole mess for me!
 

cbrace

Well-Known Member

Reaction score: 16
Messages: 317

My experience as well. I've reverted to v11.2 for the foreseeable future.
 

Datapanic

Well-Known Member

Reaction score: 233
Messages: 444

In conjunction with all that - I am trying to get a new install of poudriere to work with 12.0 and it's been rough, ruff, furf. SSL has been "updated" and borks a lot of ports. That, and php56 is eol soon. What a nightmare! Is there going to be an 11.3??? ??? ?/ ?
 

chrcol

Well-Known Member

Reaction score: 53
Messages: 482

Some old members on here may remember when I upgraded a machine to 9.0 and then suffered nasty filesystem corruption caused by a new ufs journaling feature, after that day I learnt my lession and now avoid x.0 releases. Best of luck to those who are getting their systems back up and running.

If you want to try out 12, I suggest STABLE, as any patches for found problems will go in there.
 
OP
A

ahhyes

Member

Reaction score: 8
Messages: 65

Some old members on here may remember when I upgraded a machine to 9.0 and then suffered nasty filesystem corruption caused by a new ufs journaling feature, after that day I learnt my lession and now avoid x.0 releases. Best of luck to those who are getting their systems back up and running.

If you want to try out 12, I suggest STABLE, as any patches for found problems will go in there.

Ouch.... well I am fortunate all my filesystems were intact and it was just a bootloader bug this time...

To be fair on the devs, I have had years of upgrades to .0 releases and never hit any severe issue, until now. First time for everything eh? :)
 

_martin

Daemon

Reaction score: 414
Messages: 1,251

After upgrading some LAB servers and few VMs I've decided today I'll upgrade my main machine. I even updated BIOS while I was at it. I'm used to upgrade my FreeBSD boxes using svn (cvs back then), I used the same approach today. And what do you know?

freebsd-12-failed.PNG


Booting 12.0 went fine, reboot after installworld was not sucessfull (as new /boot files got installed). I updated to amd64 12.0 r342177.
My board: S1200BT (07/25/2018) on Xeon E3-1240.

Luckily I was able to interrupt boot0 in very early stage and boot manually to 11.2 (I'm using ZFS on root but my BIOS (not even in EFI) didn't see virtual media from KVM so it would be painful ; here I saved a trip to DC).

EDIT: I forgot to mention I opened PR for this: PR 234110
 

Quarter Wave Vertical

Active Member

Reaction score: 16
Messages: 104

More than 2 weeks ago, I accidentally trashed my system when I attempted to install FreeBSD 12.0-RC3. I did a fresh re-installation of 11.2 with Mate, but I couldn't get Mate to work properly:

https://forums.freebsd.org/threads/...2-mate-not-working-properly.68430/post-409092

I installed 11.2 on an external HD but still no luck with the problems I had with Mate.

Earlier this evening, I attempted to upgrade that drive from 11.2 to 12. The problem is that my machine keeps crashing when I attempt this:

freebsd-update -r 12.0-RELEASE upgrade

(At least I think I typed "12.0-RELEASE" rather than "12-RELEASE".)

My machine is more than 10 years old, so I'm wondering if that might have something to do with it. I don't know off-hand what processor it uses but it's definitely Intel.

The reason I'm doing on the external drive is that I don't want to attempt the upgrade just yet on an 11.2 system I already have installed plus I'm hoping to see if I still have the same problems with Mate as I had before.

Does anyone have any suggestions? Thank you.
 

chrcol

Well-Known Member

Reaction score: 53
Messages: 482

One of the big challenges with FreeBSD 12.0-RELEASE is with OpenSSL 1.1 - several ports were not ready for it and now just don't build or work. https://bugs.freebsd.org/bugzilla/buglist.cgi?quicksearch=openssl

Additionally, even some windows apps, like LDAP Admin won't talk via ssl/tls to a FreeBSD 12.0-RELEASE running OpenLDAP.

Definitely a bold move to jump to 1.1 in base. The solution for ports is probably to install 1.0.x from ports and use that for ports to link to. When it was proposed a couple of years back to switch the default openssl port to 1.1 thankfully I managed to get them to abandon the idea as I had notice a lot of breakages when compiling against it, even in commonly used software, although I expect now alot of those are fixed, but its clear they not all fixed.

I advise using ports openssl anyway, simply because its fairly frequently updated and its much easier to keep a port updated than something in the base.
 

xtaz

Well-Known Member

Reaction score: 146
Messages: 476

Regarding OpenSSL they had no choice because older versions will not be supported upstream any longer throughout the whole time that 12.x will be supported. I quote from their website:

Code:
Version 1.1.0 will be supported until one year after the release of 1.1.1
Version 1.0.2 will be supported until 2019-12-31 (LTS).
Version 1.0.1 is no longer supported.
Version 1.0.0 is no longer supported.
Version 0.9.8 is no longer supported.

In regards to reports of peoples boot loader crashing. They changed the loader from forth to lua and I'm guessing this is causing issues with some people.
 

_martin

Daemon

Reaction score: 414
Messages: 1,251

As you can find in PR i submitted I tested the 12-stable - issue seems to be fixed there. But as you Datapanic mentioned OpenSSL is causing other issues.
I don't feel like doing any workarounds to have my needed ports up. Sidenote: I wonder who decided that uname -a should not show the built date and kernel config used.

Too many problems with 12.0 on main machine, I reverted back to 11.2 and will stay there for a while.
 

terescoj

New Member


Messages: 1

I woke up this morning and had a thought.. What if I booted from the USB drive that has the 11.2-RELEASE image, and mount the hard disk filesystem, then overwrite the /boot/loader on the hard disk with the one from 11.2-RELEASE

Well well well!! That has resurrected my server!!

So the problem is in /boot/loader in 12.0-RELEASE


Thanks for this workaround. I was stuck with this same error - the inability for the 12.0 /boot/loader to find anything on my disk. Booting from a memory stick 11.2 and copying over /boot/loader let me boot and I am also in the process of continuing the upgrade. I have everything backed up, but this (assuming it works through to the end of the upgrade) saved me a whole lot of time that I was thinking I'd need to spend on a reinstall and restore. I'd be happy to send along information about my hardware if it would be helpful for those tracking this down. I upgraded two machines: the first had no problems, the second failed to boot with the lua error described in this thread.
 

Quarter Wave Vertical

Active Member

Reaction score: 16
Messages: 104

More than 2 weeks ago, I accidentally trashed my system when I attempted to install FreeBSD 12.0-RC3. I did a fresh re-installation of 11.2 with Mate, but I couldn't get Mate to work properly:

https://forums.freebsd.org/threads/...2-mate-not-working-properly.68430/post-409092

I installed 11.2 on an external HD but still no luck with the problems I had with Mate.

Earlier this evening, I attempted to upgrade that drive from 11.2 to 12. The problem is that my machine keeps crashing when I attempt this:

freebsd-update -r 12.0-RELEASE upgrade

(At least I think I typed "12.0-RELEASE" rather than "12-RELEASE".)

My machine is more than 10 years old, so I'm wondering if that might have something to do with it. I don't know off-hand what processor it uses but it's definitely Intel.

The reason I'm doing on the external drive is that I don't want to attempt the upgrade just yet on an 11.2 system I already have installed plus I'm hoping to see if I still have the same problems with Mate as I had before.

Does anyone have any suggestions? Thank you.
An update on my previous posting.

I tried upgrading that drive to 12.0 again tonight and no joy. My machine crashed with a kernel panic.

When the actual upgrading process started, I got the following errors:
Code:
(da4:umass-sim1:1:0:0):  READ(10). CDB: 28 00 02 58 56 a8 00 00 80 00
(da4:umass-sim1:1:0:0):  CAM status: SCSI Status Error
(da4:umass-sim1:1:0:0):  SCSI status: Check Condition
(da4:umass-sim1:1:0:0):  SCSI sense: MEDIUM ERROR asc:11,0 (unrecovered read error)
(da4:umass-sim1:1:0:0):  Error 5, Unretryable error
g_vfs_done9):da4p2[READ(offset=xxxxxxxxxxx, length=yyyyy)]error =5
g_vfs_done9):da4p2[READ(offset=xxxxxxxxxxx, length=yyyyy)]error =5
where xxxxxxxxxxx is a location and yyyyy a size.

I also got this message:
Code:
sha256: ///usr/lib/debug/boot/kernel/geom_raid3.ko.debug:  Input/output error
Still, the system inspection continued and got to the point where files were supposed to be downloaded. A few moments after that, the crash occurred.

I remain baffled as I was able to install 11.2 direct onto the drive without any apparent problems.
 

SirDice

Administrator
Staff member
Administrator
Moderator

Reaction score: 13,152
Messages: 39,766

Those errors look like they're caused by a faulty memory stick. Have you tried a different stick?
 

Quarter Wave Vertical

Active Member

Reaction score: 16
Messages: 104

Those errors look like they're caused by a faulty memory stick. Have you tried a different stick?
That's a possibility. I think the HD was out of a scrapped second-hand laptop, so it might be on its last legs. I put the drive in an enclosure, which I bought brand new, but that's not necessarily a guarantee that things are working properly.

I've got some other external drives I could use to see if that error occurs again.
 
Top