Installed new hard drive...

[Harry Yu]

Hi guys,

One of the hard drives in our UX server die yesterday. I replaced it with a new drive today. However, I've never install a new drive to a UX server before so I don't know what do I have to do to prepare it and add it into the volume to replace the old drive's position. Please help.

- Harry

[Michael Tully · Accepted solution]

If it was part of an existing volume group, you need to use the 'vgcfgrestore' command. e.g.

# vgcfgrestore /dev/myvg /dev/rdsk/c2t6d0

The man pages have other examples. If the disk was not using any type of mirroring, then you'll have to restore your data.

[Harry Yu]

Thanks for your reply, Michael.

Do I need to format the new hard drive first? I'm more a Window admin, so I am not sure how to prepare the hard drive.

[Mel Burslan]

there is no "formatting" in hpux or in any modern flavor of unix, as opposed to the same concept in Windows.

when you run the vgcfgrestore command, it gets the header information of the old disk from the database and writes it over to the new disk.

if you can elaborate what the exact situation is, you may get more help. Like was thisa mirror disk ? Was this the bootable disk ? Was it a disk with some application data ?

[Harry Yu]

This drive is solely use for our database data. It's volume 2 out of 5 volumes. I tried Michael's suggestion and got the following...
---------------------------------------------
# vgcfgrestore -o /dev/vg03 /dev/dsk/c0t1d0
Usage: vgcfgrestore
[-R]
[-F]
{ -n vgname |
-f VGConfPath }
{ -l |
{[-o OldPVPath] PVPath}}
OldPVPath and PVPath : character special files
More arguments required.
---------------------------------------------

[Mel Burslan]

do not use old pv path directive use the commad as follows:

vgcfgrestore -n /dev/vg03 /dev/dsk/c0t1d0

[Michael Tully]

Try using the -n option.

# vgcfgrestore -n /dev/vg03 /dev/dsk/c0t1d0

(The man page has this example, the one I listed before had a typing error)

[Harry Yu]

hi guys,

When I use the command 'vgcfgrestore -n /dev/vg03 /dev/dsk/c0t01d0', I got the following error...
--------------------------------------------
# vgcfgrestore -n /dev/vg03 /dev/dsk/c0t1d0
Configuration info for Physical Volume "/dev/dsk/c0t1d0" not found
in "/etc/lvmconf/vg03.conf" :
Character-special file for PV not supplied on command line, OR
PV did not belong to Volume Group /dev/vg03 when configuration file was created.
You have mail in /var/mail/root
---------------------------------------------
help help! :( 42 users are on my back. :(

[Mel Burslan]

my bad... it is not dsk, it should be rdsk... sorry.

vgcfgrestore -n /dev/vg03 /dev/rdsk/c0t1d0

[Harry Yu]

Ok guys, I use the rdsk instead of dsk. Now, it is not doing anything. It's just kind of hang there. I can't cancel the job nor see anything happening. I don't see the disk running either. Well, it blinks but definately not running. How can I tell if it's doing something? I use TOP and I don't see the process going. Is it usually take awhile? Thanks.

[Michael Tully]

We need to ascertain whether your using the correct volume group and physical volume. Please supply:

vgcfgrestore -n /dev/vg03 -l
strings /etc/lvmconf/vg03
vgdisplay -v /dev/vg03

[Harry Yu]

Alright Mike, here are the returns...
-----------------------------------------------
# vgcfgrestore -n /dev/vg03 -l
Volume Group Configuration information in "/etc/lvmconf/vg03.conf"
VG Name /dev/vg03
---- Physical volumes : 2 ----
/dev/rdsk/c0t1d0 (Non-bootable)
/dev/rdsk/c2t1d0 (Non-bootable)
# strings /etc/lvmconf/vg03
/etc/lvmconf/vg03: No such file or directory
# vgdisplay -v /dev/vf03
vgdisplay: Volume group "/dev/vf03" does not exist in the "/etc/lvmtab" file.
vgdisplay: Cannot display volume group "/dev/vf03".
---------------------------------------------
Thanks a lot guys! :~)

[Mel Burslan]

vgcfgrestore command does not take any time at all. prolonged wait may be a sign of a dead disk drive or controller.

along with the commands Michael indicated, run this one and include the output of it as well

ioscan -fnC disk

[Mel Burslan]

another output please :

strings /etc/lvmtab

[Harry Yu]

Hi Mel,

The drive is new so I think it isn't bad. Coz I tested the bad drive and it was BAD. Ok, here is the return from ioscan...
-----------------------------------------------
# ioscan -fnC disk
Class I H/W Path Driver S/W State H/W Type Description
=====================================================================
disk 0 8/4.0.0 sdisk CLAIMED DEVICE IBM DGHS09Y
/dev/dsk/c0t0d0 /dev/rdsk/c0t0d0
disk 1 8/4.1.0 sdisk NO_HW DEVICE TANDEM 4609-A*csm#
/dev/dsk/c0t1d0 /dev/rdsk/c0t1d0
disk 2 8/4.2.0 sdisk CLAIMED DEVICE IBM DGHS09Y
/dev/dsk/c0t2d0 /dev/rdsk/c0t2d0
disk 3 8/4.5.0 sdisk CLAIMED DEVICE SEAGATE ST34573WC
/dev/dsk/c0t5d0 /dev/rdsk/c0t5d0
disk 4 8/4.8.0 sdisk CLAIMED DEVICE SEAGATE ST34573WC
/dev/dsk/c0t8d0 /dev/rdsk/c0t8d0
disk 5 8/4.9.0 sdisk CLAIMED DEVICE IBM DGHS09Y
/dev/dsk/c0t9d0 /dev/rdsk/c0t9d0
disk 6 8/4.10.0 sdisk CLAIMED DEVICE IBM DGHS09Y
/dev/dsk/c0t10d0 /dev/rdsk/c0t10d0
disk 7 8/12.0.0 sdisk CLAIMED DEVICE IBM DGHS09Y
/dev/dsk/c2t0d0 /dev/rdsk/c2t0d0
disk 8 8/12.1.0 sdisk CLAIMED DEVICE IBM DGHS09Y
---------------------------------------------
Which I believe the TANDEM is the new boy in the box. HW indicate hardware problem?

[Mel Burslan]

the device reportin NO_HW, which is the second from the top, is your problem. The replaced drive was either dead on arrival or was not properly connected to the system by the person who replaced it.

You need some h/w support before anything else.

[Harry Yu]

Hi guys,

Ok, I re-inserted the hard drive into the rack and run IOScan and everything looks good. No more "NO_HW". It shows "CLAIMED" just like other drives. I run the 'vgcfgrestore -n /dev/vg03 /dev/rdsk/c0t1d0' command again and it's hanging there now.

[Harry Yu]

ok, here is the output of "strings /etc/lvmtab"...
-----------------------------------------------
# strings /etc/lvmtab
/dev/vg00
/dev/dsk/c0t5d0
/dev/dsk/c0t8d0
/dev/vg01
/dev/dsk/c0t9d0
/dev/dsk/c0t10d0
/dev/vg02
/dev/dsk/c0t0d0
/dev/dsk/c2t0d0
/dev/vg03
/dev/dsk/c0t1d0
/dev/dsk/c2t1d0
/dev/vg04
/dev/dsk/c0t2d0
/dev/dsk/c2t2d0
---------------------------------------------

One additional thing is now the vgcfgrestore command is running (hanging) and I run ioscan again and the drive shows "NO_HW" again. Is that related?

[Harry Yu]

Some additional info about the new drive...
-----------------------------------------------
# diskinfo -v /dev/rdsk/c0t1d0
SCSI describe of /dev/rdsk/c0t1d0:
vendor: TANDEM
product id: 4609-A*csm#
type: direct access
size: 8980000 Kbytes
bytes per sector: 512
rev level: 0180
blocks per disk: 17960000
ISO version: 0
ECMA version: 0
ANSI version: 3
removable media: no
response format: 2
(Additional inquiry bytes: (32)38 (33)30 (34)33 (35)39 (36)37 (37)33 (38)46 (
39)37 (40)35 (41)4d (42)53 (43)50 (44)54 (45)38 (46)30 (47)31 (48)38 (49)20 (50)
20 (51)0 (52)0 (53)0 (54)0 (55)0 (56)0 (57)0 (58)0 (59)0 (60)0 (61)0 (62)0 (63)0
(64)0 (65)0 (66)0 (67)0 (68)0 (69)0 (70)0 (71)0 (72)0 (73)0 (74)0 (75)0 (76)0 (
77)0 (78)0 (79)0 (80)0 (81)0 (82)0 (83)0 (84)0 (85)0 (86)0 (87)0 (88)0 (89)0 (90
)0 (91)0 (92)0 (93)30 (94)39 (95)52 (96)49 (97)39 (98)39 (99)33 (100)32 (101)37
(102)0 (103)30 (104)30 (105)30 (106)31 (107)32 (108)32 (109)30 (110)39 (111)4c (
112)33 (113)39 (114)32 (115)38 (116)20 (117)20 (118)20 (119)20 (120)20 (121)46 (....
---------------------------------------------
So, the drive is good until I run the vgcfgrestore command and the drive would "NO_HW" in ioscan.

[Michael Tully]

Run diskinfo on the drive. I suspect you have a dud.

# diskinfo -v /dev/rdsk/c0t1d0

[Mel Burslan]

Harry,

I am not sure how you ended up a TANDEM branded disk drive in this bunch but this smells like a firmware issue. As soon as you try to write something to the drive, you change the drive controller status. Nothing else comes to mind now. The drive hardware is definetly not right.

On a second note, as I am done for the day on the west coast of USA, I hope Michael Tully can help you further as he is downunder and his work day must be just starting ;)

Cheers...

[Harry Yu]

Thanks Mel. I'll assign the points later. I think the drive should be okey becuase it's also made by IBM. I don't know why it wouldn't work. Maybe I need to reboot the server and then run the vgcfgrestore again? What do you think Michael?

[Michael Tully]

I kind of agree with Mel about the firmware. I've not seen a 'TANDEM' drive being used before. As soon as you attempt to write to it, the drive goes down. Where did you get the drive from ? If it is with HP, perhaps call them ask why, it doesn't work. Rebooting the server is not going to help.

[Harry Yu]

Well, our server is kinda old (240MHz w/ 512MB RAM.) Do you think that's why it takes a long time to do the vgcfgrestore? When the vgcfgrestore is running and I run ioscan, the drive appeared to be "NO_HW" but diskinfo still shows info fine. Could that because the drive is in use? The drive is labeled as Compaq drive and the seller claim it's IBM Genuine Ultra 160 Drive. The only different between this drive and the old drive was that this is a 10k rpm drive instead of 7.2k. Do you think that matters?

[Michael Tully]

Harry,

Again it will depend on whether it has the proper HP firmware loaded on it. Get the vendor to verify this. I cannot be sure about the speed rating, but we have to look at the facts as to why the disk doesn't work. We must look at the differences between it and the others. The NO_HW to me is a problem, not matter what is running.

Regards
Michael