binary upgrade issues

[John Rogers]

Hi, you are right - there was indeed the following messages:

kernel: ad0: FAILURE - READ_DMA status=51<READY,DSC,ERROR>
error=40<UNCORRECTABLE>

And this message appeared starting from the date I ran CFS, an
encrypted filesystem.  It's a half year old WesternDigital250GB disk.
I now disabled dma (was UDMA100 actually) by adding "hw.ata.ata_dma=0"
to /boot/loader.conf .

Before I saw your reply, I just manually created those old-index etc
by following upgrade.sh, and ran the rest of the upgrade.sh from the
"Removing schg flag from existing files..." part.  After that I have
ran portupgrade, portsnap etc, and so far don't see problem.  Do I
still need to go back to 6.0 and run upgrade.sh?

Thanks.

On 8/4/06, Colin Percival <cperciva at freebsd.org> wrote:
> John Rogers wrote:
> > Installing new kernel into /boot/GENERIC... done.
> > Moving /boot/kernel to /boot/kernel.old... done.
> > Moving /boot/GENERIC to /boot/kernel... done.
> > Removing schg flag from existing files...
> >
> > Then my connection to the server froze and I found the server rebooted
> > itself.  After login I found it was 6.1-RELEASE FreeBSD 6.1-RELEASE
> > #0: Sun May  7 04:32:43 UTC 2006.
> >
> > Don't know why it rebooted, and my concern it: had it finished
> > upgrading?
>
> Probably not.
>
> > I looked into the upgrade.sh and found it should continue
> > working on files referred in old-index, new-index-nonkern, new-index.
> > However none of these files were found in the directory.  Also I am
> > worried whether the schg flags were recovered.  How can I check these?
>
> Sounds like a generic case of 'system crashed and recently created files
> weren't written to disk yet'.  I'm really suspicious of the hardware here,
> but I'd suggest
> 1. mv /boot/kernel /boot/kernel.new
> 2. mv /boot/kernel.old /boot/kernel
> 3. reboot (back into 6.0-RELEASE)
> 4. Run the script again and hope that it manages to finish installing everything
> this time.
>
> Colin Percival
>