11.2-Release fail on CD and upgrade

FreeBSD-11.2 install from CD did not work. freebsd-update upgrade did not work. I use AMD-p4 since 2010. What are the developer’s using to allow such a thing to happen, or is it just me again? I would really like to know because it makes no since that no one see this but a me and you, or are we the only one who read the announcement and installed it.

FIRST RUN - BOOT:

Code:
Can't find
Can't find
Can't find … since when?
Can't find something… is it MBR, spectrum or my lunch bucket....
...
...
It roll too fast to read.
...
...
Fatal trap 12:
CPUID:3 Apic ID=03
KDB: Stack Backtrace
...
...
Trap Number: 0

Dead then a doornail. Same thing for either type install. And to think we talked about this days before and now it happen again with 11. Why release it when you know it got issues someplace somewhere known only to the developers? Where is the how-to-activate sheet?

I'm just talking to myself to keep from cracking.

Have a nice day.
 
FreeBSD-11.2 install from CD did not work. freebsd-update upgrade did not work. I use AMD-p4 since 2010.
I just upgraded a production server without a glitch. My hunch is that support for Pentium 4 was removed from the kernel. Why don't you try NetBSD i386 branch. Maybe they still have the support for such ancient hardware. What are you doing with it anyway?
 
I just upgraded a production server without a glitch. My hunch is that support for Pentium 4 was removed from the kernel. Why don't you try NetBSD i386 branch. Maybe they still have the support for such ancient hardware. What are you doing with it anyway?
I meant to say AMD Quad-4. Anyway, I installed it on another machine which has a intel processor. It installed (I should have done both machine in the first place). However, the Intel Cougar machine gives me a clue about apic (that could be what the can’t find lines had detected). So it must be something wrong with apic on my AMD Quad-4 machine. Also at the login prompt on the Cougar, I got warning about USB here and about 5 more before reaching this point:
Code:
Login: usbd_setup_device_desc::  getting device descriptor at addr 3 failed.  USB_ERR_STALLED.
…
…
There is at least two threads here with one of them stating fatal errors. I'm going to read into them now.

I don’t have the smarts or the time to debug apic. I’m just getting to know perl. Maybe it’s a USB thing or faulty hardware. For now, I’ll stick with FreeBSD-11.1 until 12.0-RELEASE but this time I’m not going to hold my breath or count my chickens before they hatch. Maybe I’ll use INTEL until I figure this thing out, but most FreeBSD major release always came through for my AMD.

That’s for the info Oko
 
Most of the other failures seem to be connected with VirtualBox and/or NVidia. I haven't tested anything with NVidia yet, and haven't upgraded anything with VirtualBox but an machine upgraded from 11.1 to 11.2 (with a source build rather than FreeBSD-update, though I don't know if that would make any difference, did fine with an install of VirtualBox. I haven't yet seen problems with fresh installs of 11.2, but of course, it's only been out for a short time.

As for testing, it's not possible to test on every kind of hardware with every variation. Even Apple, with its tremendous budget, and very limited hardware to test compared to anything running on a PC, winds up releasing broken software.

(Not to say, it's not my fault, he did it too, but pointing out that even with far easier conditions, it can happen.)
 
Support for FreeBSD 11.1 will end 3 months from now. That still quite a few months before the planned release of 12.0.
So it’s now or never?

OK … For what I just been through, I realize it don’t make no differences at which processor you are using. However, depending on what version of FreeBSD you install, it does! I kind of remember my 9x and 10x BIOS experiences. One way work for 9.0 and not for 9.2, and so on. I never wanted to go back to that again even to prove it to myself.
.........................................
FreeBSD-11.2 BIOS setting attempt:
On Chip Sata type = [Native IDE]
I change it to [AHCI]
Now 11.2 WORKS!

.........................................

How cool can that be ... but it fooled me into thinking I found the solution when it comes to a pure install of FreeBSD-11.2 on an AMD processor.

Most of the other failures seem to be connected with VirtualBox and/or NVidia. I haven't tested anything with NVidia yet, and haven't upgraded anything with VirtualBox but an machine upgraded from 11.1 to 11.2 (with a source build rather than FreeBSD-update, though I don't know if that would make any difference, did fine with an install of VirtualBox. I haven't yet seen problems with fresh installs of 11.2, but of course, it's only been out for a short time.

After wasting all day playing with the BIOS while trying to include my previous /usr/local and other saved directory files to be included, the only darn thing I ended up having to do was to set vboxnet to NO in /etc/rc.conf. .. and now I have my old mate desktop and all the rest running under FreeBSD-11.2_RELEASE as though nothing ever happen. It is so great to know that you can transfer your previous local and root to a new FreeBSD-release. I live for that but it don't help when one try to explain it when help is needed :)

Code:
vboxheadless_enable="NO"
vboxnet_enable="NO"      # for bridged networking CHANGE to NO for 11.2 for now.
vboxwatchdog_enable="NO"
vboxwebsrv_enable="NO"

and do the same type thing for /boot/loader.conf

However, now I have no Virtualbox VM’s running anymore. It's dead then a door-nail.

The good thing is FreeBSD came through and preserved my old desktop environment. Now I can move forward with confidence. I don’t like pointing fingers but I don’t know how speak perfectly. I just know I can’t let the author of the hook if I feel that something had not been said. I take it that it was all about Virtualbox and the plan to replace it with bhyve just like clang replaced gcc for the better. I guest all the clues were there; but why not provide a warning in plain English. That is all we, or at least I need to read.

Anyway, my solutions are within these two threads and I’ll get to it when I get to it. Right now, I’m all burn-out:

drhowarddrfine thread:
https://forums.freebsd.org/threads/freebsd-update-to-11-2-release-fails-to-boot-reboots.66431/

romanaOne thread:
https://forums.freebsd.org/threads/virtualbox-fatal-trap-12-on-11-2-release.66435/#post-392611

Now that what I call perfect timing to save the world.
 
I tried everything including using the 11.2 kernel in my 11.0 host and visa-versa. Stuff like that use to never fail until now. I’m not about to destroy the greatest FreeBSD setup on the planet created thru the help of this community since 2008 or 10.

I’ll use 11.2 on my VPS as should be, and inside my FreeeBSD-devel vBox guess until they repair the most useful Virtualization tool FreeBSD ever adopted. Bhyve is not for everyone. It only run FreeBSD and maybe Linux last I heard, and 32bit at that. It would be perfect in production where vBox wouldn’t stand a chance. Just an opinion of a broken hearted user.

Code:
Jun 30 06:18:30 vhost kernel: Mounting local filesystems:.
Jun 30 06:18:30 vhost kernel: ELF ldconfig path: /lib /usr/lib /usr/lib/compat /usr/local/lib /usr/local/lib/gcc48 /usr/local/lib/gcc6 /usr/local/lib/gcc7 /usr/local/lib/ipsec /usr/local/lib/nss /usr/local/lib/opencollada /usr/local/lib/qt4 /usr/local/lib/virtualbox /usr/local/llvm37/lib

Jun 30 06:29:51 vhost pkg: virtualbox-ose-kmod-5.2.8_1 installed
Jun 30 06:30:00 vhost /usr/sbin/cron[1036]: (root) CMD (/usr/libexec/atrun)
Jun 30 06:30:43 vhost kernel: pid 1044 (VirtualBox), uid 0: exited on signal 6
Jun 30 06:31:03 vhost kernel: pid 1052 (VirtualBox), uid 0: exited on signal 6
Jun 30 06:33:00 vhost /usr/sbin/cron[1063]: (operator) CMD (/usr/libexec/save-entropy)
Jun 30 06:33:42 vhost ntfs-3g[937]: Unmounting /dev/ada0s17 ()
Jun 30 06:33:43 vhost ntfs-3g[940]: Unmounting /dev/ada0s18 ()
Jun 30 06:33:49 vhost shutdown: reboot by root: 
Jun 30 06:33:49 vhost kernel: Jun 30 06:33:49 vhost shutdown: reboot by root: 
Jun 30 06:33:49 vhost kernel: Stopping cron.
Jun 30 06:33:49 vhost kernel: Waiting for PIDS: 782.
Jun 30 06:33:49 vhost sshguard[740]: Exiting on signal
Jun 30 06:33:49 vhost kernel: Stopping sshguard.
Jun 30 06:33:49 vhost kernel: Waiting for PIDS: 740
Jun 30 06:33:49 vhost kernel: .
Jun 30 06:33:49 vhost syslogd: exiting on signal 15

I had to write this by hand. I'm not good at speed reading yet.

Then boom!
vboxdrv: SupDrvInitDevExt CIP page: rc=-26 Fail

I google that this is nothing new. It been around since vBox 4.2.4 or earlier; however, I can only figure that FreeBSD avoided it until now. Well, I’m throwing in the towel too! Thank you sincerely for 11.0 and 11.1 :) they both has everything they will ever need beyond EOL.
 
I upgraded another 6 production servers and dozen jail (iocell used for provisioning) instances after experiencing no problems for 48h in production with the first upgraded server reported above. So far things look good. Two of the most recently upgraded file servers have multiple ZFS pools with half petabyte of data served over NFSv3. Looks to me the problems are mostly related to NVidia binary blob drivers and VirtualBox (FreeBSD is not an officially supported platform). IMHO (and in my experience with VirtualBox on FreeBSD) that is to be expected. I would remove NVidia binary blob and VirtualBox from the ports tree as well as Linux emulation layer as it serves no useful purpose.
 
:oops: I forgot about the release notes. It goes deep into vBox issues:

https://www.freebsd.org/releases/11.2R/errata.html

They can remove what they want but don't abandon Virtualbox. People are happy with it. It runs perfect on FreeBSD and it's good with memory and CPU too. I'm sure 11.2 is the bomb for servers. It goes on my VPS once I'm finish. I can live with 11.0. He is a great and secure host, I use his vms all day long. One is XP and two are FreeBSD and I never had an issue what-so-ever. It enough to drive a hacker nuts if he dare and that goes for XP too. I don't think no one can secure XP like I done. He just keep coming back for more until he mark you. He got all the right tools. We got an handle on svchosts. Even Bill don't stand a chance to connect.

FreeBSD vm's takes care of its own. What you do Oko with system is beyond me. I learn a lot from reading your theads. You don't play! That's the way I am with my little boxes.
 
I don't see any plans to abandon VirtualBox, just that it, like NVidia and probably some other blobs, will have trouble using packages or working in upgrades for awhile. I haven't used bhyve, for my uses, VirtualBox is fine and requires no effort from me after all this time, but I'm pretty sure that it does do Windows now. https://wiki.freebsd.org/bhyve/Windows has a wiki writeup though last edited over a year ago.
 
thoroughly test the nvidia binary blobs before the release is announced?
FreeBSD is volunteers driven project. Why didn't you test? If most of those volunteers who care or are paid to work on FreeBSD use MACs as daily drivers it is very likely that they consider NVidia binary blobs irrelevant for the purpose of the release of the server OS. I am a big believer in US Navy's KISS principle. Removing 3rd party binary blobs is a good policy and simplifies things.
 
I don't have a spare machine setup nor space in which to set one up... spare machines are packed away. Recently moved. Not running but CURRENT also...
 
thoroughly test the nvidia binary blobs before the release is announced?
The "problem" right now is that packages are still built for 11.1 because it's still a supported version. Some packages, like the NVidia driver, need to be specifically built for 11.2 or they won't work. This "problem" will disappear as soon as 11.1 is EoL and packages are linked to 11.2.
 
Is this also the case for both quarterly, and latest pkg repos? There is not a separate 11.2 pkg repo, and will not be until 11.1 is EOL?
 
Back
Top