topic replacing a boot disk in a addvol pair in Operating System - Tru64 Unix

replacing a boot disk in a addvol pair

Bengt Nilsson_2 — Thu, 24 Sep 2009 06:14:49 GMT

I have a v4.0F system with a small boot disk where I did addvol to increase user area by adding a new harddisk.

Now the boot disk is failing, but still running.
I have backed / and /usr to a new bootable disk and /usr/users to another disk, by piped vdump /vrestore.

My question:
Will I be able to boot on the new root disk without problems?
Or will the fact that I did addvol to /usr/users give me problems? The backed boot disk should be an exact copy of the old one, so some "memory" of this fact would emerge, I can imagine.

Or should I do some "addvol" related actions to separate this twin pair and then do a backup transfer of / and /usr to the new boot disk again before I take the system down to replace the disks?

The original disk problem is showing itself by that the boot disk is sometimes disappearing from the ">>>sh dev" list, so if I take the system down there is a definite risk I will never see it again.

Re: replacing a boot disk in a addvol pair

Rob Leadbeater — Thu, 24 Sep 2009 07:46:00 GMT

Hi,

As long as you wrote the disklabel to the new disk with the flags "-t advfs" then you should be OK.

Also bear in mind that unless you put the new disk on the same SCSI ID as the old one, you might have to play around with the symbolic links under /etc/fdmns to point to the new disks.

Cheers,

Rob

Re: replacing a boot disk in a addvol pair

Bengt Nilsson_2 — Thu, 24 Sep 2009 08:45:45 GMT

So the fact that the advfs filesystem management on the new boot disk expects a 2 volume combination with /dev/rz16h (vol1) + /dev/rz17c (vol2) for /usr/users, which will not exist when it is installed and used for boot, will not cause problems?

Re: replacing a boot disk in a addvol pair

Rob Leadbeater — Thu, 24 Sep 2009 08:57:39 GMT

Hi,

I'm getting confused...

Can you post the output of "ls -lR /etc/fdmns" and "scu show edt" and identify which is the failing disk etc.

Cheers,

Rob

Re: replacing a boot disk in a addvol pair

Bengt Nilsson_2 — Thu, 24 Sep 2009 09:37:18 GMT

This is the result.
The failing disk is rz16.
The tmp stuff is old and should be removed, no such devices are connedted today.

# ls -lR /etc/fdmns
total 64
-r-------- 1 root system 0 Aug 19 2002 .advfslock_fdmns
-r-------- 1 root system 0 Aug 19 2002 .advfslock_root_domain
-r-------- 1 root system 0 Aug 21 2002 .advfslock_rz18c
-r-------- 1 root system 0 Aug 21 2002 .advfslock_rz19c
-r-------- 1 root system 0 Jun 4 2003 .advfslock_tmp_root
-r-------- 1 root system 0 Jun 4 2003 .advfslock_tmp_users
-r-------- 1 root system 0 Jun 4 2003 .advfslock_tmp_usr
-r-------- 1 root system 0 Aug 19 2002 .advfslock_users_domain
-r-------- 1 root system 0 Aug 19 2002 .advfslock_usr_domain
drwxr-xr-x 2 root system 8192 Jun 4 2003 root_domain
drwxr-xr-x 2 root system 8192 Nov 20 2007 rz18c
drwxr-xr-x 2 root system 8192 Nov 20 2007 rz19c
drwxr-xr-x 2 root system 8192 Jun 4 2003 tmp_root
drwxr-xr-x 2 root system 8192 Jun 4 2003 tmp_users
drwxr-xr-x 2 root system 8192 Jun 4 2003 tmp_usr
drwxr-xr-x 2 root system 8192 Jan 21 2008 users_domain
drwxr-xr-x 2 root system 8192 Jun 4 2003 usr_domain

/etc/fdmns/root_domain:
total 0
lrwxr-xr-x 1 root system 10 Jun 4 2003 rz16a -> /dev/rz16a

/etc/fdmns/rz18c:
total 0
lrwxrwxrwx 1 root system 10 Nov 20 2007 rz18c -> /dev/rz18c

/etc/fdmns/rz19c:
total 0
lrwxrwxrwx 1 root system 10 Nov 20 2007 rz19c -> /dev/rz19c

/etc/fdmns/tmp_root:
total 0
lrwxr-xr-x 1 root system 10 Jun 4 2003 rz20a -> /dev/rz20a

/etc/fdmns/tmp_users:
total 0
lrwxr-xr-x 1 root system 10 Jun 4 2003 rz20h -> /dev/rz20h

/etc/fdmns/tmp_usr:
total 0
lrwxr-xr-x 1 root system 10 Jun 4 2003 rz20g -> /dev/rz20g

/etc/fdmns/users_domain:
total 0
lrwxr-xr-x 1 root system 10 Jun 4 2003 rz16h -> /dev/rz16h
lrwxrwxrwx 1 root system 10 Jan 21 2008 rz17c -> /dev/rz17c

/etc/fdmns/usr_domain:
total 0
lrwxr-xr-x 1 root system 10 Jun 4 2003 rz16g -> /dev/rz16g

# scu show edt

CAM Equipment Device Table (EDT) Information:

Device: CD-224E Bus: 1, Target: 0, Lun: 0, Type: Read-Only Direct Access
Device: BD018734A4 Bus: 2, Target: 0, Lun: 0, Type: Direct Access
Device: HUS103014FL3600 Bus: 2, Target: 1, Lun: 0, Type: Direct Access
Device: BD0726459C Bus: 2, Target: 2, Lun: 0, Type: Direct Access
Device: BD0726536C Bus: 2, Target: 3, Lun: 0, Type: Direct Access
Device: C7438A Bus: 3, Target: 5, Lun: 0, Type: Sequential Access

Re: replacing a boot disk in a addvol pair

Rob Leadbeater — Thu, 24 Sep 2009 09:46:52 GMT

OK... So users_domain still contains a partition from the failing disk.

In that case you need to use "rmvol" to remove rz16h from the domain. That will copy any data from that disk over onto rz17c.

Cheers,

Rob

Re: replacing a boot disk in a addvol pair

Bengt Nilsson_2 — Thu, 24 Sep 2009 12:26:26 GMT

Thanks, this is an excellent idea that did not occur to me until you pointed it out.
The original rz16 disk included /, /usr and /usr/users, and was extended by rz17c by addvol.
So in my narrow mind I assumed that I had to do rmvol on the rz17c, which is impossible since its contents does not fit in the remaining space of rz16h.
But if I remove rz16h instead, its contents will surely fit in the remaining space of rz17c, and the problem is solved.

Thanks!

Re: replacing a boot disk in a addvol pair

Rob Leadbeater — Fri, 25 Sep 2009 08:06:42 GMT

No problem. Remember to say thanks by assigning points.

Cheers,
Rob