AIX break rootvg mirror from system down

Hello,
aix 5.2, mirrored rootvg on hdisk0 and hdisk1. hdisk0 is dead. I can boot to cd, into sms, into maintenance mode. I can fsck all the various partitions on hdisk1 (the hd4 hd2 hd3, etc...) all is fine. But without the hdisk0 part of the mirror I cannot get the system to boot. ystem hangs on boot with a LED code of 517 (or is it 518?).

None of the lvm commands for dealing with mirrors are in the cd /usr.

I tried editing the image.data file to remove references to hdisk0, and changed all the copies from 2 to 1, fixed the PP entries to match, and hoped that would break the mirrors, but alas.

So for kicks I mounted the "real" usr and now the commands are available, but now they won't work because rootvg cannot be varyonvg'd.

# lsvg -l rootvg
0516-013 : The volume group cannot be varied on because
        there are no good copies of the descriptor area.

Kind of stuck. suggestions?

It would suggest that when the server was first installed that someone used a single disk (that would be labelled hdisk0) and the when complete they ran a mirrorvg rootvg only. There are more steps required to make it a usable mirror, namely running bosboot and setting the bootlist

Within SMS, can you choose your boot disk and enquire what is on each disk? I can't remember how right now and all my servers are running, so I can't just bounce one. I think you need to ask it to scan all devices, and it might take a while depending on your configuration.

If it sees your disk has a bootable AIX, try selecting it. That would suggest that just the bootlist is incorrect.

If SMS doesn't think that your disk has, all may not be lost, but it might be a little difficult. When in single user off media (tape / CD / DVD / or NIM) maybe you could try running the bosboot command against the correct disk? You may need to get this from the serial number.

It still may not work, but nothing else springs to mind at the moment without restoring from mksysb which I presume you don't want to do.

Robin

Thanks - I am able to see the (second) disk which is how I'm able to fsck the partitions.

SMS sees both disks:
 Select Device
 Device  Current  Device
 Number  Position  Name
 1.        1      SCSI 146814 MB Harddisk, part=2 (AIX 5.2.0)
                  ( loc=U787B.001.DNWD543-P1-T14-L8-L0 )
 2.        -      SCSI 146814 MB Harddisk, part=4 ()
                  ( loc=U787B.001.DNWD543-P1-T14-L8-L0 )

if I chose #1 and the get information on it:
It does say it is bootable:

 Device Information
    /pci@800000020000003/pci@2/pci1069,b166@1/scsi@0/sd@8,0:2
                   :  (Bootable)       
DEVICE             : SCSI 146814 MB Harddisk, part=2 (AIX 5.2.0)
    ( loc=U787B.001.DNWD543-P1-T14-L8-L0 )
NAME               : sd
DEVICE-TYPE        : block

So when I service mode boot that disk, I get:
518 Remote mount of the root (/) and /usr file systems failed during network boot.
Now the normal way to fix that, as I understand it, is to reboot to cd and follow a tech note, which basically says fsck the filesystems. Which I am able to do.

But that doesn't seem to allow the darn thing to boot. And I'm guessing that the problem is that the rootvg is mirrored and one of the disks is dead. Surely that is something that should be able to be worked around? Or am I just screwed and will need to do a re-install?

That same technote:

IBM Repairing File Systems with fsck in AIX (LED 517 or 518) - United States

Says you are basically screwed...

That's what I was just coming to grips with. Ok - first time for everything. I've got a mksysb from a couple of weeks ago. I've burned it (volumes 1 of 3) to a DVD and am trying to do the (re)install. interestingly enough I happen to have a new disk which I've swapped into as a replacement for the bad disk.

ok so far. Boot to that cd which works nicely thank you very much.

Trying to do the restore to the new disk and it is circling around a
image.data has invalid file system data.
Can not continue.

I suppose I could try to extract the image.data and look at it, but even if I could, I doubt I'd be able to "put it back" into the mksysb.

I see two options. (1) try to restore to what I think is a good disk and hope for the best, or (2) do a fresh install, then restore from backups, etc. Kind like not to do that one, but I may not have a choice.

I suspect you are in for a lot of late nights....
Restore a mksysb from a mirrored system to a single disk....hmmm
A mksysb that takes 3 DVD's...hmmm
Clean install and a preservation restore might be your best bet.
LVM redbook is you next step I think:

http://www.redbooks.ibm.com/abstracts/sg245433.html

And this is worth a read:

http://www-01.ibm.com/support/docview.wss?uid=isg3T1011782

HTH

I've never managed to span DVD media for a mksysb. How big was your rootvg? Did you have stuff other than just the base OS? (and perhaps your backup software)

I have restored a mirrored rootvg to a single disk, but that was planned in advance and we adjust the backup process.

Robin