[PC-BSD Testing] Testing Digest, Vol 48, Issue 32

[Arthur Koziol]

On 01/25/2011 7:57 AM, Ian Robinson wrote:
>
>
> On Mon, Jan 24, 2011 at 10:41 PM, Ian Robinson <fitchkendall at gmail.com 
> <mailto:fitchkendall at gmail.com>> wrote:
>
>     >>
>     >> Tonight, I will try a "fresh install" on machine #2, which I
>     expect to
>     > >go well.  I will hold off on messing with machine #1 in case Kris
>     > wants me to test anything.
>     >>
>     > >Ian Robinson
>     > >Salem, Ohio
>
>     >Ian,
>     >
>     >Well, the firmware update was worth a try. In looking over the
>     changes
>     >that went into 8.2 over at the FreeBSD 8.2 TODO page, I don't see
>     >anything that jumps out at me related to LSI, SCSI, CAM or
>     otherwise. It
>     >would suck if some regression got introduced with SCSI support.
>     >
>     >Arthur
>
>
>     Arthur -- It was a great suggestion.  I do not think there is a
>     regression with SCSI support.  A "fresh install" worked
>     flawlessly.  I'm wondering if the problem is related specifically
>     to the upgrade script.  The kernel panic occurs as the upgrade
>     removes the "old" packages.  I wonder what is being triggered?
>
>     I'm trying to look on the DVD and in the archived packages to see
>     if I can find the upgrade installation script, but so far I have
>     not found it.
>
>     Ian Robinson
>     Salem, Ohio
>
>
>
> I did  not find the script in the DVD, but the installation script can 
> be located in a successful installation at:
>
> /PCBSD/pc-sysinstall
>
> ========== Code Excerpt: ========================
>
> # Set a variable of files we want to make backups of before doing upgrade
>   BKFILES="/etc/rc.conf /boot/loader.conf"
>
>   if [ -e "/dev/${MPART}" ] ; then
>     rc_nohalt "mount /dev/${MPART} ${FSMNT}"
>     if [ "$?" != "0" ] ; then
>       # Check if we have ZFS tank name
>       rc_halt "mount -t zfs ${MPART} ${FSMNT}"
>     fi
>   else
>     # Check if we have ZFS tank name
>     rc_halt "mount -t zfs ${MPART} ${FSMNT}"
>   fi
>
>   # Mount devfs in chroot
>   mount -t devfs devfs ${FSMNT}/dev
>
>   # Check if we have any ZFS partitions to mount
>   zfs mount -a
>
>   # Mount all the fstab goodies on disk
>   chroot ${FSMNT} /sbin/mount -a >>${LOGOUT} 2>>${LOGOUT}
>   chroot ${FSMNT} umount /proc >/dev/null 2>/dev/null
>   chroot ${FSMNT} umount /compat/linux/proc >/dev/null 2>/dev/null
>
>  # Now before we start the upgrade, make sure we set our noschg flags
>   echo_log "Cleaning up old filesystem... Please wait..."
>   rc_halt "chflags -R noschg ${FSMNT}"
>
>   # Make backup copies of some files
>   for i in ${BKFILES}
>   do
>     cp ${FSMNT}${i} ${FSMNT}${i}.preUpgrade >/dev/null 2>/dev/null
>   done
>
>   # Remove some old dirs
>   rm -rf ${FSMNT}/etc/rc.d >/dev/null 2>/dev/null
>
>   # If we are doing PC-BSD install, lets cleanup old pkgs on disk
>   if [ "$INSTALLTYPE" != "FreeBSD" ]
>   then
>     echo_log "Removing old packages, this may take a while... Please 
> wait..."
>     echo '#/bin/sh
> for i in `pkg_info -E \*`
> do
>   echo "Uninstalling package: ${i}"
>   pkg_delete -f ${i} >/dev/null 2>/dev/null
> done
> ' >${FSMNT}/.cleanPkgs.sh
>     chmod 755 ${FSMNT}/.cleanPkgs.sh
>     chroot ${FSMNT} /.cleanPkgs.sh
>     rm ${FSMNT}/.cleanPkgs.sh
>     run_chroot_cmd "pkg_delete -f \*" >/dev/null 2>/dev/null
>     run_chroot_cmd "rm -rf /usr/PCBSD" >/dev/null 2>/dev/null
>     run_chroot_cmd "rm -rf /PCBSD" >/dev/null 2>/dev/null
>     run_chroot_cmd "rm -rf /var/db/pkgs" >/dev/null 2>/dev/null
>     run_chroot_cmd "rm -rf /usr/local32" >/dev/null 2>/dev/null
>     run_chroot_cmd "rm -rf /usr/sbin" >/dev/null 2>/dev/null
>     run_chroot_cmd "rm -rf /usr/lib" >/dev/null 2>/dev/null
>     run_chroot_cmd "rm -rf /usr/bin" >/dev/null 2>/dev/null
>     run_chroot_cmd "rm -rf /boot/kernel" >/dev/null 2>/dev/null
>     run_chroot_cmd "rm -rf /sbin" >/dev/null 2>/dev/null
>     run_chroot_cmd "rm -rf /bin" >/dev/null 2>/dev/null
>     run_chroot_cmd "rm -rf /lib" >/dev/null 2>/dev/null
>     run_chroot_cmd "rm -rf /libexec" >/dev/null 2>/dev/null
>   fi

Is there some way through a VM that you could trap the installation 
progress and then drop to a terminal to grab progress from a log file? 
There's got to be some way to see the failure occur in a debug mode or 
something to know exactly where the failure happens.

Arthur
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.pcbsd.org/pipermail/testing/attachments/20110125/4d4fb45e/attachment.html>