Re: install corrupts existing ext3 multi-boot partitions

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

 



I had the same thing during FC2 testing and did not get any response
from the list on how to fix it.  I stumbled across a fix last week when I
ugraded the system to a new Athlon 64 system.  I created new disk
partitions an a larger 160 GB disk and used cpio to copy the files.
I could then reboot from the new disk into Red Hat 9!  The old disk
is still unbootable.  If someone can provide a fix, I could then use
the old hardware with the old disk.

Jim

----------mail to FC2 list on 5/17/04-------------------
I have a system with FC2 on /dev/hda6 (/) and /dev/hda1 (/boot).  I also have a
RedHat 9 system installed on /dev/hda8 (/) and I dual boot the system.  The disk
also has Win2K and WinXP.  The FC2 is a fresh install over a FC2 Test 3 system
as I used this system for testing of the FC2 test releases.  Somewhere along the
line during the FC2 testing, the RedHat 9 system can no longer be booted.  I get
the following messages:
...
...
XT3-fs: mounted filesystem with ordered data mode.
attempt to access beyond end of device
03:08: rw=0, want 1219858868, limit=36893241
attempt to access beyond end of device
03:08: rw=0, want 1219858868, limit=36893241
Kernel panic: No Init found.

I think this started after I did a "fixfile" on a FC2T2 system and had the
RedHat partition mounted and it was "fixed" too.  I am really not sure when the
partition became unbootable.  The limit value in the message is the size of the
partition as reported by fdisk.  I also get alignment errors reported from
sfdisk.  The same errors that are reported with other users who have dual boot
problems.  I even tried the fix suggested for the dual boot problem, but it had
no affect.  I do not have a problem with either windows systems, it is just with
RedHat 9.  The system is a Tyan 2460 with two Athlon 1200 cpus and 120 GB disk.

Any idea of now I can get this system bootable?  I can read/write it just fine
from FC2.  It just won't boot.

Jim
-------------------------end of old mail--------------------

----- Original Message ----- From: "John Reiser" <jreiser@xxxxxxxxxxxx>
To: <fedora-test-list@xxxxxxxxxx>
Sent: Sunday, October 24, 2004 8:56 PM
Subject: install corrupts existing ext3 multi-boot partitions


I believe that I may have suffered a case where a fresh install of
Fedora Core 3 [Test 3] with SELinux in targeted, permissive mode borked
existing multi-booted installations of RedHat 9, RedHat 8.0, and RedHat
7.3 [separate ext3 root filesystems for each].  The typical symptom when
booting an old system is a console message:
   Freeing unused kernel memory:  156k freed
   attempt to access beyond end of device
   03:09: rw=0, want=1219858868, limit=5863693
   Kernel panic: No init found.  Try passing init= option to kernel.
I definitely had been multibooting FC2 with selinux=disabled in
/etc/sysconfig/selinux.  I changed to targeted, permissive mode for
FC3test1, FC3test2, and FC3test3.  The last successful boot of RH9
was July 20 according to its /var/log/messages.

The cause might be that install causes all existing ext3 filesystems
get the new attribute ext_attr, and older kernels have bugs interpreting
this.  This seems to happen even for filesystems that were omitted from
the mount list when DiskDruid was run during install of FC3test3.  "Auto
management" of /media/idediskN might be a reason why.

Both RH9 e2fsck 1.32 (09-Nov-2002) and RHEL3 e2fsck 1.35 (28-Feb-2004)
say that all filesystems are OK [checked in RH9 rescue mode, and from
RHEL3: unmount, then "e2fsck -f /dev/hda9", etc.], but the old kernels
don't get it.  A multibooted RHEL3 2.4.21-20.EL does work correctly.

Which component(s) deserve(s) a bugzilla report for this?  Exactly where
in the install process did ext_attr get applied to existing filesystems?
It seems to me that the Release Notes for Fedora Core 3 should mention
this situation.

So, how do I recover?  [I need to boot the old systems to support
customers who run them.]  What about a procedure like this:
  1. Boot RH9 rescue mode.  Create a new ext3 without ext_attr.
  2. Boot FC3test3 rescue mode with selinux=0.  Mount old and new
     partitions; copy all files from old to new.
  3. Edit /etc/fstab of FC3test3 to list all ext3 partitions,
     and use 'ro' [readonly] or 'noauto' for ones that must
     not ever have ext_attr turned on.
  4. Try to boot RH9 using the new copy (after adjusting the ext3 label,
     etc.)

What am I overlooking?

--


--
fedora-test-list mailing list
fedora-test-list@xxxxxxxxxx
To unsubscribe: http://www.redhat.com/mailman/listinfo/fedora-test-list)



[Index of Archives]     [Fedora Desktop]     [Fedora SELinux]     [Photo Sharing]     [Yosemite Forum]     [KDE Users]