AIX machine won't boot - LED says 0518

[Victor Lee]

Hi

I have an F50 running AIX 4.2.1 that refuses to boot up. The LED
display says 0518. From researching the LED message I found out that
it means problems mounting the /usr partition. Here is what I did:

boot from CD
start a shell without mounting filesystems
logform /dev/hd8
fsck -y all the partitions

at this point when I exit from the shell, /etc/getrootfs coredumps and
says:

mvdir: no such file or directory
mount: /dev/hd2 on /usr: no such file or directory

at which point I have no choice but to hit the reset button - and then
the machine gets stuck at 0518 again

how can I fix this?

thanks
Victor

[David]

I had exactly the same symptoms a few weeks ago. It all started with
a crash caused by autofs. The server would exhibit the same symptoms
you've described.

I could boot to the maintenance shell as long as I didn't let it mount
the filesystems. Eventually I realized that the usr directory was
missing from the / directory of the root filesystem. /var and /tmp
were missing too. The entire /dev directory and its contents were
gone as well. I think there were a few soft links missing too.
Rather strange since you can't remove those manually on a running
system. I'm thinking that fsck must have had something to do with it.

After recreating all the missing directories and populating /dev as
best I could, I still couldn't get the machine to reboot. It would
always fail trying to mount /usr. I could manually mount all the
filesystems under an alternate tree (/mnt /mnt/usr /mnt/var...) but if
I tried to overmount / as the getrootfs does it would immediately
coredump. Eventually we restored from a mksysb to an alternate disk
and manually mounted the filesystems from the bad root disk and copied
over the data we needed.

[Bill]

vi...@amdocs.com (Victor Lee) wrote in message news:<5ddb7fa1.04040...@posting.google.com>...

Perhaps run mkdir /usr from the maintenance menu ? Or 'lsvg -l
rootvg' / 'lslv hd2' from the maintenance menu ?

BV

[Victor Lee]

thanks for all the responses

Here's what I've done: I attached an extra disk to the F50, made it
the boot disk, and installed AIX 4.2.1 on it. Now the machine is up
and running. Since I am not familiar with AIX, I have further
questions:

1. how do I mount the filesystems from the original disks (that had
filesystem problems) Is it importvg?

2. someone suggested that I do a preservation install on the original
disks - is that a good idea?

thanks
Victor

[Victor Lee]

I did importvg, and this is the result:

# importvg hdisk0
0516-530 updatelv: Logical volume name hd6 changed to pagelv00.
0516-530 updatelv: Logical volume name hd5 changed to bootlv00.
0516-530 updatelv: Logical volume name hd8 changed to loglv00.
0516-530 updatelv: Logical volume name hd4 changed to fslv00.
0516-530 updatelv: Logical volume name hd2 changed to fslv01.
0516-530 updatelv: Logical volume name hd9var changed to fslv02.
0516-530 updatelv: Logical volume name hd3 changed to fslv03.
0516-530 updatelv: Logical volume name hd1 changed to fslv04.
vg00
imfs: Warning: logical volume fslv00 already exists in
/etc/filesystems.
imfs: Warning: logical volume fslv01 already exists in
/etc/filesystems.
imfs: Warning: logical volume fslv02 already exists in
/etc/filesystems.
imfs: Warning: logical volume fslv03 already exists in
/etc/filesystems.
# mount /dev/fslv01 /mnt
mount: 0506-322 Cannot determine log device to use for /dev/fslv01
(/mnt).
# fsck /dev/fslv01
Filesystem Helper: 0506-506 Invalid argument

both the mount and fsck commands have failed - I'm stuck. Any
help/input is appreciated.

thanks
Victor

[los]

vi...@amdocs.com (Victor Lee) wrote in message news:<5ddb7fa1.04042...@posting.google.com>...

You can't have 2 volume groups with the same name. The hdisk0 that
you tried to import was your old 'rootvg'. When you put in another
disk and installed 4.2.1 onto it you had a new 'rootvg'. When you ran
'importvg hdisk0' it was trying to import another vg w/ the same name
and logical volumes w/ the same name which is why it was renaming them
(interesting, I didn't know it would do that).

That's why daily backups are so important vic. Atleast the logical
volumes seem to be intact so your data should be ok. I'd be
interested to know what an lsvg or lspv command output would be after
you ran 'importvg hdisk0'.

Judging from the output you got from:

# mount /dev/fslv01 /mnt
mount: 0506-322 Cannot determine log device to use for /dev/fslv01
(/mnt).

I would try to just create a new jfs log fs for fslv01. or
Since you have an extra disk do what David did

"Eventually we restored from a mksysb to an alternate disk
and manually mounted the filesystems from the bad root disk and copied
over the data we needed."

Good luck and let us know what happens.