Operating System - HP-UX
1756719 Members
2342 Online
108852 Solutions
New Discussion юеВ

Re: This one is a beauty-vgcfgrestore the header

 
Adam Noble
Super Advisor

This one is a beauty-vgcfgrestore the header

A certain individual and I promise you it's not me has overwritten one of our primary disks in a volume group. Basically they made it a BCV for another system on our XP array....not the best plan I'm sure you'll agree.

So it no longer has header info and is complaining its unavailable and more critically the oracle database sitting on part of it is in a rather bad position. The database is still running but we are getting the odd message in /var/adm/syslog and I have suggested it should be shutdown. The business has not agreed.

Anyway my course of action is to vgcfgrestore the header (I pray that works) newfs the 3 filesystems that were on that disk and restore. Does anyone have any better way of dealing with this issue. Also what damage is leaving the database running doing????

Thanks
2 REPLIES 2
Steven E. Protter
Exalted Contributor

Re: This one is a beauty-vgcfgrestore the header

Shalom,

In a disk overwritten situation, you can't trust any other method other than full data restore.

You can try vgchgid on the disk's prior to vgimport, but I do not think that will work. Good news is you have a backup.

SEP
Steven E Protter
Owner of ISN Corporation
http://isnamerica.com
http://hpuxconsulting.com
Sponsor: http://hpux.ws
Twitter: http://twitter.com/hpuxlinux
Founder http://newdatacloud.com
Rita C Workman
Honored Contributor

Re: This one is a beauty-vgcfgrestore the header

No guarantees, but...

I'd recommend a stop of work or schedule a time to do this to avoid potential data loss.

I'd stick my database in hotbackup mode and get myself a solid, current backup to avoid data loss and mgmt issues.

Then, I'm guessing you are 'taking' back your disk. I would take that disk and run vgcfgrestore on it. Try and see if you can run it on the disk with the volume group active. I did it once on an old T500 & it worked.

vgcfgrestore -n /dev/vgxx /dev/rdsk/c-t-d-

If it won't let you, then since you have a solid backup...umount f/s and take down what you have to and vgchange -a n /dev/vg and now run the restore command.

Now run your lvdisplays to see if you have any stale extents. If you need to resync things up. Then when all is good see if things will come back up.
If it does......great
If it doesn't....get the backup tape and start restoring data.

When you're back up, I'd then be checking out the procedures of who takes what disk and getting that FIXED.

Just thinking..
Rita