PV status unavailable

[Guettache]

The only disk of a volume group is shown as unavaialble in vgdisplay.
ioscan clains this disk. Diskinfo output is OK.
The disk is a raid 5 array. The Raid array controller does not show any bad disk.
I tried to restore the physical link using vgcfgrestore.
vgcfgrestore -n /dev/vg04 /dev/rdsk/c4t4d0
Volume Group configuration has been restored to /dev/rdsk/c4t4d0




pvdisplay /dev/dsk/c4t4d0
--- Physical volumes ---
PV Name /dev/dsk/c4t4d0
VG Name /dev/vg04
PV Status unavailable
Allocatable yes
VGDA 2
Cur LV 38
PE Size (Mbytes) 16
Total PE 60766
Free PE 45499
Allocated PE 15267
Stale PE 0
IO Timeout (Seconds) default
Autoswitch On

The PV is still unavailable, and there are no Stale PE.
I did a vgexport than vgimport. Everything is succesfully executed, but still PV is unavailable.
Can I use vgreduce and then pvcreate to reestablish the link.

Thanks,
Narimane

[Avinash20]

Hi,

Does dd work fine

Also could you check for any errors in

/var/opt/resmon/log/event.log

[Avinash20]

If everything is fine, did you try to activate the Vg

# vgchange -a y

If data is not a contraints, it better to do a pvcreate and again add it to the VG.

[Guettache]

I did vgchange:
vgchange -a y /dev/vg04
Activated volume group
Volume group "/dev/vg04" has been successfully changed.

But
syslog displays:
Feb 26 15:40:02 btcsrvr vmunix: LVM: VG 64 0x040000: PVLink 31 0x044000 Failed! The PV is not accessible.
Feb 26 15:40:02 btcsrvr vmunix: LVM: VG 64 0x040000: Lost quorum.
Feb 26 15:40:02 btcsrvr vmunix: This may block configuration changes and I/Os. In order to reestablish quorum at least 1 of the following PVs (represented by current link)

This message explains PV unavailable.

Narimane

[Guettache]

About dd , first note that the array Raid 5 is made of 6 disks of size 189456384 KB each.

dd gives an IO read error exactly at the last block of the disk
dd if=/dev/rdsk/c4t4d0 of=/dev/null bs=8192k
dd read error: I/O error
23127+0 records in
23127+0 records out
(23127 * 8192= 189456384)
Testing the sanity of the array disk on the controller is OK.

[Avinash20]

dd read error: I/O error

There seems to be an issue with the H/W .. May be connectivity.. not sure..

If this is new VG which is being created then do the vgreduce the PV, and then again recreate the LUN and assign it to HP-UX and then insf, pvcreate, vgextend.

Check and let us know

[smatador]

Hi,
There is something wrong with your array...
dd error, pv unavailable.
What array it is?

[Guettache]

Hi,

ioscan clains the disk . Diskinfo is ok

disk 4 0/8/0/0.4.0 sdisk CLAIMED DEVICE transtecT5012R24
/dev/dsk/c4t4d0 /dev/rdsk/c4t4d0

diskinfo -v /dev/rdsk/c4t4d0
SCSI describe of /dev/rdsk/c4t4d0:
vendor: transtec
product id: T5012R24
type: direct access
size: 995721216 Kbytes
bytes per sector: 512
rev level: 0001
blocks per disk: 1991442432
ISO version: 0
ECMA version: 0
ANSI version: 3
removable media: no
response format: 2

What bothers me is that a full disk size is dumped before I/O read error occurs.

[smatador]

Hi,
In the transtec T5012R array do you see green light or amber/red in the panel?
I never work with transtec, but is it possible to test the array or the connectivity. Do you have some management array?

[smatador]

Hi,
I check the hardware guide of this array
if LCD display shows OWOOOS
Possible Cause: â Wâ means â warningâ - there are many bad sectors on your HDD.
Fix: Change hard drives.

http://download.transtec.de/doit/loadva/software/w3/CD_STOR_HTML/raid/t50/transtec-Hardware-Guide-de.pdf

Hope it helps

[Guettache]

I have just resolved the problem.

I could not believe that there was an IO error though the dd output tells it, but in the other hand, it displays it after the whole disk size was scanned.

My array had two offline disks which were not used. I just removed them and insert 2 other disks, restart the array. Note that the Raid set did not have to be rebuilt as the 2 offline disks were not part of the set

I did activate the volume group,(with messages lost quorum) and vgcfgrestore again. then I got back the PV status to available.

I have no data loss.

That the sequence i did:

restart the array
vgimport,
vgcfgrestore
pvdisplay /dev/dsk/c4t4d0
--- Physical volumes ---
PV Name /dev/dsk/c4t4d0
VG Name /dev/vg04
PV Status available
Allocatable yes
VGDA 2
Cur LV 38
PE Size (Mbytes) 16
Total PE 60766
Free PE 45499
Allocated PE 15267
Stale PE 0
# Timeout (Seconds) default


Thank you for your replies . Thoug my problem is now fixed, I am stille puzzeled by the output of dd command

Narimane

[Avinash20]

dd shows IO error.
Need to check the H/W of the server.

[Guettache]

I resolved the problem . read my previous thread.

Cheers,
Narimane