Problem with vg and pvdisplay.

[Lai Nee Shyang_1]

Hi there,
I've a K-Class machine connected to a EMC3330. Recently, someone accidentally switch off one of the SCSI channel at the EMC end and 1/2 of the path was gone. After I switch it on again, I could do diskinfo and ioscan show the correct number of devices again. But I started getting strange message from pvdisplay and vgdisplay. One example :
++++++++++++++++++++++++++++++++++++++++++++++
<> pvdisplay /dev/dsk/c2t0d0
pvdisplay: Warning: couldn't query physical volume "/dev/dsk/c2t0d0":
The specified path does not correspond to physical volume attached to
this volume group
pvdisplay: Warning: couldn't query physical volume "/dev/dsk/c2t0d1":
The specified path does not correspond to physical volume attached to
this volume group
pvdisplay: Warning: couldn't query all of the physical volumes.
pvdisplay: Couldn't retrieve the names of the physical volumes
belonging to volume group "/dev/vgsap2".
pvdisplay: Cannot display physical volume "/dev/dsk/c2t0d0".
+++++++++++++++++++++++++++++++++++++++++++++++
Anyway I can rectify this or I have to do a reboot ?

Thanks

Lai

[harry d brown jr]

If they were screwing around in the Symmetrix, did they by chance move any cables around?

Also, you aren't using powerpath are you?

live free or die
harry

[Lai Nee Shyang_1]

Yo.

Nope, all connectivity are still intact (else I won't be able to see all devices using ioscan) and I'm not using any EMC s/w, or Powerpath.

Lai

[Animesh Chakraborty]

Hi,
Do you have alternate path configured on the box?
Try
#ioscan -fnC disk
#insf -e
#vgdisplay -v vgxx

[Lai Nee Shyang_1]

Hi.

Yes, I've alternate path and I was able to do ioscan and see all the devices, including those affected by the down path.

I've not tried insf -e, I thot it might not be useful since ioscan already produce the path.

Lai

[Animesh Chakraborty]

Hi,
No harm trying insf -e.
You can try one more thing
#cp /etc/lvmtab /etc/lvmtab.bak
#vgscan
It will recreate your lvmtab file

Best of luck
Animesh

[Lai Nee Shyang_1]

Hi Animesh,

thanks. I think I'll try that when I can find appropriate downtime.


Lai

[Scott Van Kalken]

Depending on whether or not this is an HPUX 11.00 machine or not, it could just be a patch thing. I remember that 11.00 didn't like doing a PV switch. It could take up to ten minutes. HP released a patch for this, but after a pv switch many commands did not work. I think it's been fixed now, but don't count out a patch level issue.

A simple reboot *may* fix this.

[Mary Ann Lipa]

Hi,
Maybe you it will be better if you move 'mv' /etc/lvmtab instead of copying 'cp' it. cp sometimes fails... i have always used 'mv' ad it never fails...
# mv /etc/lvmtab /etc/lvmtab.old
#vgscan -a


-Dan

[James R. Ferguson]

Hi:

If you find that you have complete connectivity (including alternate links) and the problem persists, try increasing the IO timeout as follows:

# pvchange -t 180 /dev/dsk/cXtYdZ

To see its current value, do:

# pvdisplay /dev/dsk/cXtYdZ

ALSO:

With regard to 'vgscan' and copying as opposed to moving the current file *before* running 'vgscan: You will note from the man pages that "If the /etc/lvmtab file exists, the information contained in the file is used to assist in rebuilding the file, but the existing file is updated with the new crected configuration.". Generally, one *moves* the current 'lvmtab' to 'lvmtab.old' before doing the 'vgscan'.

Regards!

...JRF...

[Sanjay_6]

Hi Lai,

Can you deactivate and activate the VG again. For all Vg's giving this problem, try,

vgchange -a n /dev/vg_name

then activate it,

vgchange -a y /dev/vg_name

Now do a vgdisplay,

vgdisplay -v /dev/vg_name

Hope this helps.

Regds

[Krishna Prasad]

Do you try a vgcfgrestore /dev/vg##

vgchange -a y /dev/vg##

If you have software mirrors in the volume group. You will want to do a vgsync /dev/vg## but since you are using EMC you shouldn't have to do this.