Error writing VGDA...

[Tom Spence_1]

Hi all,

syslog.log:
vmunix: Error writing VGDA [1] to disk H/W path 4.6.0 (error = 5)
Sep 18 06:50:29 burns vmunix: LVM: vg[10]: pvnum=0 (dev_t=0x1c006000) is MISSING
Sep 18 06:50:29 burns vmunix: LVM: Lost quorum in vg[10], too many PVs missing!

I can't remove lv and vg on 4.6.0... And other messages that I wasn't sure what it is...

Thanks.

Tom

[Stefan Farrelly]

It sounds like youve completely lost connection to a number of disks - one being on path 4/6.0. Have you checked your hardware ? sounds like something major has failed. You shoudnt be trying to remove lvs and vgs, you should be looking into finding out what the hardware problem is and fixing it first.

[Tom Spence_1]

Right, I can't remove vg and lv... It won't bring it up after reboot. So I had to deactivated on 4.6.0. Once I did deactivated then what? I give you the info via vgdisplay, here is:

$ vgdisplay /dev/vgdata2
--- Volume groups ---
VG Name /dev/vgdata2
VG Write Access read/write
VG Status available
Max LV 255
Cur LV 1
Open LV 1
Max PV 16
Cur PV 1
Act PV 1
Max PE per PV 16384
VGDA 2
PE Size (Mbytes) 4
Total PE 323
Alloc PE 322
Free PE 1
Total PVG 1

That vgdata2 is in 4.6.0.

Tom

[James R. Ferguson]

Hi Tom:

"...lost quorum...too many PVs missing" ...suggests a major hardware failure.

What does 'ioscan -fnC disk' show?

Regards!

...JRF...

[Tom Spence_1]

JRF,

Great.. Look at file attachment after I type 'ioscan -fnC disk'.

Tom

[Ted Ellis_2]

sounds like you are having fun... when you say deactivate.... what exactly did you do? Did you vgchange that volume group to deactivate? Your ioscan did not seem to show anything as unclaimed... is the disk that threw this error still in the system?

What is the hardware that this disk is in... internal disk or some external array?

[James R. Ferguson]

Hi (again) Tom:

# diskinfo /dev/rdsk/c0t6d0

...does this report a valid size?

# dd if=/dev/rdsk/c0t6d0 of=/dev/null bs=256k

...does this report any errors?

Regards!

...JRF...

[Tom Spence_1]

Hi JRF again,

Here you go:

#diskinfo /dev/rdsk/c0t6d0
SCSI describe of /dev/rdsk/c0t6d0:
vendor: HP
product id: C2474S
type: direct access
size: 1323540 Kbytes
bytes per sector: 512

#dd if=/dev/rdsk/c0t6d0 of=/dev/null bs=256k

Is this taking a while?

Tom

[Tom Spence_1]

JRF,

Here is:

#dd if=/dev/rdsk/c0t6d0 of=/dev/null bs=256k
5170+1 records in
5170+1 records out

Is this mean 'No problem'?

Tom

[James R. Ferguson]

Hi Tom:

OK, the physical disk reports a non-zero size and there are no I/O errors if we read it.

Now, I'm confused. Do a verbose 'vgdisplay' for the volume group so we can see the physical volume information:

# vgdisplay -v /dev/vgdata2

The volume group must be active for anything to be returned.

Exactly what are you trying to do?

Regards!

...JRF...

[Anil C. Sedha]

Hi Thomas,

Can you please provide results of the following commands

vgdisplay -v vg00

vg00=volume group for which it shows disks missing

If it shows you the disks listed within the output at the end, please try a pvdisplay -v /dev/dsk/c0t0d0 (disks name)

If you don't get any message, or get an error try running fsck with "-o" option.

Regards,
Anil

[Tom Spence_1]

Hi ...JRF...

Sure, here is:

#vgdisplay -v /dev/vgdata2
--- Volume groups ---
VG Name /dev/vgdata2
VG Write Access read/write
VG Status available
Max LV 255
Cur LV 1
Open LV 1
Max PV 16
Cur PV 1
Act PV 1
Max PE per PV 16384
VGDA 2
PE Size (Mbytes) 4
Total PE 323
Alloc PE 322
Free PE 1
Total PVG 1

--- Logical volumes ---
LV Name /dev/vgdata2/lvdata2
LV Status available/syncd
LV Size (Mbytes) 1288
Current LE 322
Allocated PE 322
Used PV 1


--- Physical volumes ---
PV Name /dev/dsk/c0t6d0
PV Status unavailable
Total PE 323
Free PE 1


--- Physical volume groups ---
PVG Name vgdata2
PV Name /dev/dsk/c0t6d0

Sure I am happy to explain you... My coworker told me that she cannot open /data2. She tried to umount it which works then mount it back but can't. She decided to deactivate by go to SAM -> VG -> Decactive then she asked me to remove LV then VG. So I went to SAM -> LV -> Remove (lvdata2) but got error message, here is:

The command used to remove a logical volume, /sbin/lvremove, has failed. The stderr output fromthe command is shown below. The logical volume has not been removed.

lvremove: Couldn't delete logical volume "/dev/vgdata2/lvdata2": No such device or address

I puzzled and decided to check in /dev/vgdata2, here is:

#ls -l /dev/vgdata2
total 0
crw-r--r-- 1 root sys 64 0x0a0000 Mar 25 1999 group
brw-r----- 1 root sys 64 0x0a0001 Jul 29 17:55 lvdata2
crw-r----- 1 root sys 64 0x0a0001 Aug 27 10:56 rlvdata2

I don't get it. That is why I was soo frustrated. Thanks.

Tom

[S.K. Chan · Accepted solution]

I think you have a good disk BUT a corrupted VGRA on that disk. The "PV Status Unavailable" can only mean 2 things LVM record is corrupted or disk is bad. So far your disk seems to respond to basic checks (ioscan, diskinfo, etc). Before you do anything at all .. try to perform a full fsck on the lvol ..
# fsck -F vxfs -y -o full /dev/vgdata2/rlvdata2
and then try to mount it again. If that doesn't help and all else fail, vgexport and vgimport hopefully can solve this (if you got a good vgdata2.conf in /etc/lvmconf).
# cd /
# vgchange -a n /dev/vgdata2
==> Deactivate it.
# vgexport -m mapfile /dev/vgdata2
==> Remove VG entry fr lvmtab and all its devices files while savinf LV info in "mapfile".
# mkdir /dev/vgdata2
# mknod /dev/vgdata2/group c 64 0x0a0000
==> I'm using the same minor number (0x0a0000)
# vgimport -m mapfile /dev/vgdata2 /dev/dsk/c0t6d0
# vgchange -a y /dev/vgdata2
# vgcfgbackup /dev/vgdata2
Hope this will help ..

[Ted Ellis_2]

oops.. I see that the volume group was already deactivated in SAM by your co-worker?.. have you tried to reactivate it?

vgchange -a y /dev/vgxxxxx
then do a vgdisplay -v again