Replacing Dead Disk in 'L' class server

[John O'Neill_6]

Hey All,

Just a quick query about replacing a dead disk in an 'L' Class server that has hot pluggable disks.

I'm running HP UX on PA-RISC. B.11.11

Here's the #ioscn -fnCdisk output:

Class I H/W Path Driver S/W State H/W Type Description
=====================================================================
disk 0 0/0/1/1.0.0 sdisk CLAIMED DEVICE SEAGATE ST39204LC
/dev/dsk/c1t0d0 /dev/rdsk/c1t0d0
disk 1 0/0/1/1.2.0 sdisk NO_HW DEVICE SEAGATE ST336704LC
/dev/dsk/c1t2d0 /dev/rdsk/c1t2d0
disk 2 0/0/2/0.0.0 sdisk CLAIMED DEVICE SEAGATE ST39204LC
/dev/dsk/c2t0d0 /dev/rdsk/c2t0d0
disk 3 0/0/2/0.2.0 sdisk CLAIMED DEVICE SEAGATE ST336704LC
/dev/dsk/c2t2d0 /dev/rdsk/c2t2d0
disk 4 0/0/2/1.2.0 sdisk CLAIMED DEVICE HP DVD-ROM 304
/dev/dsk/c3t2d0 /dev/rdsk/c3t2d0

As you can see, it's dead as a doornail.

Disks 1 and 3 are part of the /dev/vg01 volume group which consists of two disks. I'm using LVM, VG01 is not a boot volume and all Logical Volumes on /dev/vg01 are mirrored.

By following the 'When good disks go bad' text sent to me by our re-seller i need to:

A> Physically Swap the disk over
B> vgcfgrestore -n /dev/vg01 /dev/dsk/c1t2d0
C> vgchange -a y /dev/vg01

Will this suffice?

Do i need to throw in a:
#vgsync /dev/vg01

For good measure?

Now here's a possible twist.. the disk i'm looking at putting in is not empty, it's a spare disk that was previously serving in another server, so it's NOT empty.

Do i just use the '-F' option on the vgcfgrestore command, or do i lvreduce the mirrored volumes, then vgreduce, then pvcreate and then vgextend followed by lvextend ?

Which way do i go?

Just a quick double check as the person I normally deal with at my re-seller is away for a few days and i want to make sure I understand what i've been informed to do correctly by posting here.


-John

[Torsten.]

Check if your OS is patched and have the "pvchange -a ..." command available.

How to use this is also written in the manual "when good disks ..."

Even if the disk was used before, I guess "vgcfgrestore ..." should work without "-F" - try.

Normal procedure:

# pvchange -a n /dev/dsk/c1t2d0
(swap disk)
# vgcfgrestore -n /dev/vg01 /dev/rdsk/c1t2d0
(not the "*r*dsk")
# pvchange -a y /dev/dsk/c1t2d0
# vgchange -a y vg01
(maybe needed)
# vgsync vg01

[John O'Neill_6]

I have no 'pvchange -a'

Here is the version I have...

SYNOPSIS
/usr/sbin/pvchange [-A autobackup] -s pv_path

/usr/sbin/pvchange [-A autobackup] -S autoswitch pv_path

/usr/sbin/pvchange [-A autobackup] -x extensibility pv_path

/usr/sbin/pvchange [-A autobackup] -t IO_timeout pv_path

/usr/sbin/pvchange [-A autobackup] -z sparepv pv_path

it is an old server that only needs to hang on long enough for our 11.23 2660 Itanium's with hardwar RAID to take over.

[Torsten.]

Check with

# vgdisplay -v vg01

if the status is "unavailable" for this disk, it is really dead. Swap the disk, restore the config, vgchange the VG and sync.

[John O'Neill_6]

Hi,

Checked with vgdisplay

--- Physical volumes ---
PV Name /dev/dsk/c1t2d0
PV Status unavailable
Total PE 8681
Free PE 931
Autoswitch On

PV Name /dev/dsk/c2t2d0
PV Status available
Total PE 8681
Free PE 931
Autoswitch On

Must be totally dead.

Sorry to sound really pedantic, i'm not a veteran sys-admin.

So do I:

A) Remove dead Hard Disk
B) Plug in new Hard Disk
C) vgcfgrestore
D) vgchange
E) vgsync

?

Thanks for tips, i need baby steps on this hardware stuff.

-John

[Torsten. · Accepted solution]

No problem ;-)

A) Remove dead Hard Disk
B) Plug in new Hard Disk
B1) ioscan -fnCdisk
C) vgcfgrestore
D) vgchange
E) vgsync

the hardware path 0/0/1/1.2.0 is written on the chassis - easy to find.

[tkc]

hi john,

0/0/1/1.2.0 is an internal disk to the L class server. from what i know, this disk is a hot-pluggable disk and not hot-swappable. the o/s may not be able to detect the replaced disk until it's rebooted. let me know how you go.

[Torsten.]

@tkc:

Caution with the terms.

The user manual clearly says

"When proper software and hardware procedures are followed, internal disk drives can be removed and
replaced while the server is running.

The procedures in this section are in two parts, a HotSwap Software Procedure and a HotPlug Hardware Procedure, for removing the disk from the server."


there is no need to reboot.
It is NOT a hot-swap in terms of "plug'n'play" (compared to a hardware RAID controller) - you need to interact with software commands (as mentioned above).

But it is surely possible to swap the disk while the OS is running.

[John O'Neill_6]

Sorry guys.. now i'm confused, i think i'll just get an HP engineer on site and pay the fee to get a professional to do it.

Do i need to halt the processes running on VG01?? I thought that the whole purpose of this damned mirroring and LVM's and hot pluggable disks was so I could replace a faulty disk WITHOUT having to reboot the damned machine.

Can you please clarify what i need to do or should i just call HP and get them to do it (on a T&M Basis).

-John

[John O'Neill_6]

Thanks peoples.

All sorted, went really smoothly.

Happy to close this thread.

-John