LV & PV issue !!!!

[Yogesh M Puranik]

Hello Gurus,

Actully i got below errors while excuting lvdisplay,

# lvdisplay /dev/app_msa/u11
lvdisplay: Warning: couldn't query physical volume "/dev/dsk/c24t0d1":
The specified path does not correspond to physical volume attached to
this volume group
lvdisplay: Warning: couldn't query physical volume "/dev/dsk/c25t0d1":
The specified path does not correspond to physical volume attached to
this volume group
........ so on.....

So i did,

#cp -p /etc/lvmtab /etc/lvmtab.old

#vgscan -v

It solved the problem temporarily.

but it is showing current PVs move than active PVs now.

Cur PV 16
Act PV 11


so i am in bit of confusion wht went wrong and wht to do to resolve this..pls help me.

Rgds
Yogesh

[Torsten.]

/dev/dsk/c24t0d1 is a LUN, so I assume it comes from an array.

If 4 "disks" (LUNs?) are missing, I would expect a problem with your external array.

Post

# ioscan -fnk

What array do you have? Check connectivity and health!

[Michal Kapalka (mikap)]

hi,

check if you have some NO_HW in ioscan -fn |grep NO_HW

Ask SAN admins if they made some changes.

mikap

[SoorajCleris]

Hi,

I think your LUNs are not available. Kindly post

#ioscan -fnC disk

Regards,
Sooraj

[SoorajCleris]

Hi Torsten,

May I ask why you have suggested "fnk" instead of "fn" ?

"k" will not scan the hardware again correct?
So it may give a wrong output with devices with claimed status wich actually NO_HW.

This is my doubt, please advice.

Regards,
Sooraj

[Yogesh M Puranik]

Hi all,

Thnx a lot for ur reply.

The scinario is I went thorough , the lun deallocation activity on the server ( LUN was allocated already from EVA),i had done vgreduce on primary disks, but not on alternate,bcoz of tht i got above error and i did above steps.

I knew that the devices files are not available for these pv's.

Though after settling the issue, why there is a difference between current and active PV's ??

Rgds
Yogesh

[SoorajCleris]

Hi ,

Please post the ioscan output, that will tell you everything.

you may please check vgdisplay -v output also.

cheers
Sooraj

[Torsten.]

@Sooraj Cleris: At this point it was not clear what kind of array it is. For example on a SCSI based array one bad disk can hang the complete bus, so an ioscan may hang too. Just for safety I prefer to do a "-k" first to prevent this. Normally the diags will update the status, so an "ioscan -k" information is not older than ~30 min.

[Yogesh M Puranik]

Attaching ioscan output.

[Torsten.]

Looks like you access the SAN via 1 switch only. Where is the connection to the other SAN?

[SoorajCleris]

That was a stunning explanation .. I admire your Thorsten.. Keep going.

100 points from me :)

[Yogesh M Puranik]

@Torsten:Looks like you access the SAN via 1 switch only. Where is the connection to the other SAN?

- - - - -
u mean to say...the SAN access is happening via 1 switch only and not with other switch isn't it ??

but i will be very greatful to u if can u please explain me in brief,how u can predict the same?? and what is the relation of the same causing difference between cueernt and active PV??

appritiate ur help .

[Torsten.]

I was probably wrong. You are using a dual port HBA, 2/0/8/1/0/4/0... and 2/0/8/1/0/4/1..., I overlooked this.

However,
Cur PV 16
Act PV 11
means, there are 16 configured "disks", but only 11 are available. So you have no access to 5 disks/LUNs. Compare your ioscan with the VG configurations to find out which are missing.

[Sharma Sanjeev]

Hi Yogesh,

You can use vgreduce -f option,

This option does not require a physical volume to be supplied on the command line. vgreduce obtains the name of each physical volume belonging to the volume group from the file /etc/lvmtab. It then reads the LVM structures from each PV and compares these with that held by the kernel to work out which PVs are missing. PVs which are missing will be candidates for removal. If all the physical extents on the missing PV are free then it will be removed from the volume group


Regards
Sanjeev