vgdisplay -v /dev/vg00 errors

[Mike_21]

I have been getting some strange errors listed below!


# vgdisplay -v /dev/vg00|more
vgdisplay: Warning: couldn't query physical volume "/dev/dsk/c0t8d0":
The specified path does not correspond to physical volume attached to
this volume group
vgdisplay: Warning: couldn't query all of the physical volumes.
vgdisplay: Warning: couldn't query physical volume "/dev/dsk/c0t8d0":
The specified path does not correspond to physical volume attached to
this volume group
vgdisplay: Warning: couldn't query all of the physical volumes.
vgdisplay: Warning: couldn't query physical volume "/dev/dsk/c0t8d0":
The specified path does not correspond to physical volume attached to
this volume group
vgdisplay: Warning: couldn't query all of the physical volumes.
vgdisplay: Warning: couldn't query physical volume "/dev/dsk/c0t8d0":
The specified path does not correspond to physical volume attached to
this volume group

[Pedro Sousa]

Hi Mike!
try:
# mv /etc/lvmtab /etc/lvmtab.old
then
# vgscan -v
It will recreate your lvmtab.
If you see some problems on the new lvmtab, you can always move the old one to the original place.
good luck.

[Rob Smith]

Hi, looks like you may have a bad disk. Try ioscan -funC and see if the disk is claimed. Then diskinfo on the questionable disk, use the raw device, if the size is 0 bytes then you have a bad disk. Hope this helps.

Rob

[boley janowski]

what is the state of the mounts? is this disk just a mirror, or what? is there any reason the disk may not be seen? when you do an

ioscan -funC disk

this disk shows up?

if its there and this just looks like a corrupt lvmtab, do

cp /etc/lvmtab /etc/lvmtab.old
rm /etc/lvmtab
vgscan -v

compare to get an idea of what its seeing if you want:
strings /etc/lvmtab
strings /etc/lvmtab.old

now look at your vgdisplay, after you do this you should have a better idea of what you will need to do next.

good luck, and keep us updated

[Rita C Workman]

Well Hi again Mike,

My question before making suggestions is this:
Is this disk generic to one system ONLY? or is it part of a disk array and could possibly belong to another server (could be called by another device name there...).
If it's ONLY connected to ONE system, than you could do as was suggested and move the lvmtab and rebuild it. Or if you want to 'investigate' than you could check /etc/lvmconf - Is there a vg that you don't recognize? Could this be something left over from previous admins that never got used and you could free up?
If it's part of a disk array - than proceed with caution so you don't free up this disk on this box...that really belonged to a vg on a different box.

Just a couple thoughts,
Rita

[James R. Ferguson]

Hi Mike:

I think its important to ascertain *what* may have happened before attempting a repair. For instance, did you or someone else, reduce a mirror? Do you have any errors in /var/adm/syslog/syslog.log ? While a 'vgscan' may be in order to fix a corrupt /etc/lvmtab, you should read the man pages for 'vgscan' carefully before you deploy it.

...JRF...

[Pedro Sousa]

Thanks Rita and James for wisely correcting my first answer.

[Mike_21]

Ok, these systems were setup by a previous admin, so I here is the information I found out, and I find quite strange!

Machine A contains 2x9G drives Mirrored
8/12.4.0
8/12.12.0

Machine B contains 2x9G drives Mirrored
8/12.5.0
8/12.8.0

NOW! Machine A and B are connected by a SCSI cable, this explains why all these disks are showing up when doing an ioscan.

Now the interesting theory of how they are mirrored.
8/12.4.0 from Machine A is mirrored to 8/12.8.0 on Machine B.
8/12.5.0 from Machine A is mirrored to 8/12.12.0 on Machine B.

This does sound a bit strange to me until I wrote the actual setup on paper, I am still wondering why this was done this way. Is there any benefit?????

[Vincenzo Restuccia]

See http://www.dutchworks.nl/htbin/hpsysadmin?h=3&dn=20332&q=Warning:%20couldn%27t%20query%20physical%20volume%20%22%2fdev%2fdsk%2fc0&fh

[Pedro Sousa]

Mike,
can you see all these 4 disks on each system?
if so, perform:
#pvdisplay -v /dev/dsk/c0t8d0 |grep -v curr|grep -v free
and
#pvdisplay -v /dev/dsk/c0t4d0 |grep -v curr|grep -v free
if they are mirrored, they should have the same data.

[Mike_21]

Ok, I seem to have stale LV's!

0000 /dev/dsk/c0t4d0 0242 current ??? 0242 stale

c0t4d0 is suppose to be mirrored to c0t8d0

[Pedro Sousa]

Mike,
from the doc # KBRC00001719
lvdisplay -v /dev/vg01/lvol2 shows that a PE is stale. What to do ?
RESOLUTION
How to Determine which disk is bad when lvdisplay -v of an lvol with mirrored disks reports a stale PE?

A PE can be marked stale because it can't write to that PE or because it cannot be read from the mirrored disk's PE. It should not be taken for granted that the disk with the stale PE is the problem disk. A hardware call should be placed to have someone verify the hardware logs. This can normally be done with STM. If this is not an option then you can use dd command to read from each disk. The disk that returns the error would be the faulty disk.

EXAMPLE
If c0t0d0 is mirrored to c1t1d0 and 1 PE is showing stale
dd if=/dev/dsk/c0t0d0 of=/dev/null bs=128k

The above command will read all of /dev/dsk/c0t0d0 and send the output to
/dev/null. If a problem is encountered reading the disk you'll see the error.

If the above returns without error you can run ...

dd if=/dev/dsk/c1t1d0 of=/dev/null bs=128k.

If the above also returns no error a call with the response center should be placed.

good luck.