Psyche -> Shrike Upgrade RAID 0 problems (trashed fs, cannot find boot block)

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



[I previously sent this to psyche-list by mistake.]

Background:
I am trying to upgrade 8.0 to 9.0.  (BTW, I have a subscription but had to
use BitTorrent as RH's servers were too slow).
I have /boot ext3 on /dev/hda1 and a matching /spare on /dev/hdc1, not as a
raid.
Everything else is RAID 0 with /dev/md0 ext3 /home etc.
RH 9 appears not to understand the existing RAID partitions: in my analysis,
it fails to start and stop them properly, and fails to recognize the disk
geometry in the presence of RAID arrays.

Detail follow:

First Upgrade Failure:
I ran the text-mode upgrader and at time of installation it said I was out
of space on /boot, so I did ALT-F1 and deleted old vmlinux files.
When I pressed OK, it ran for a few seconds and got a SIG 11 and rebooted.
When it came back up, it could not find the RAID partitions.  Looking at
/tmp/syslog showed me that it refused even to put them together.
I could not run fsck or assemble the file systems as is, so I ran fsck on
just /dev/hda* and let it remove the journal and conver them to ext2,
because I was desperate.
I was then able to use "mdadm --assemble --force /dev/md? /dev/hda?
/dev/hdc?" to force it to re-assemble the partitions.
A few hours later, I had all the lost directories put back together (lost
/dev, /etc/sysconfig, /usr/lib), with a few parts left over, but it booted
and ran RedHat 8.0 again and appeared to be mostly back to normal.

Second Upgrade Failure:
I re-ran the text-mode installer and it again got a SIG 11.  I would suspect
hardware problems except that the machine normally runs reliably.
I copied all the data off to a separate disk and commenced an install from
scratch, without repartitioning.

First installation failure:
RH 9 wanted me to increase /boot to 100MB, which I did by reconfiguring
/dev/hda1 boot and the RAID 0 next to it (/usr).
I realize that the /boot size was only a recommendation but since my first
failure was because of a small /boot, I decided to fix it, since I had to
trash the data there anyway.

Second installation failure:
Here is where things get really bad.  When I re-created the RAID 0 for /usr,
the wizard said it was "Raid Array 0" but said the other RAID devices I had
left untouched were /dev/md0, /dev/md1, etc.  When I then tried to continue
the installation, it insisted on putting the boot record in the first
/dev/md? in the list, which I think was /dev/md0, which happened to be /home
made of /dea/hd[ac]8.  Obviously, that was not going to work, but there were
no choices.

Third installation failure:
I went back and deleted all the /dev/md? RAID 0 and re-made RAID 0 with the
RH 9, and made a RAID1 out of /dev/hda1 and /dev/hdc1 (which had been kept
as /spare), since it appeared to want that for /boot.  Sadly, it again
offered only "RAID DEVICE 0" which was the first one I made (again, /usr on
/dev/hd[ac]8) as the only option for placing the boot block. 

I guess I will now go and delete the RAID devices again and make sure to
create the RAID 1 for /dev/hd[ac]1 as the first RAID device I create.
But I'm not impressed.

 





[Index of Archives]     [Fedora Users]     [Centos Users]     [Kernel Development]     [Red Hat Install]     [Red Hat Watch]     [Red Hat Development]     [Red Hat Phoebe Beta]     [Yosemite Forum]     [Fedora Discussion]     [Gimp]     [Stuff]     [Yosemite News]

  Powered by Linux