cannot deactivate volume groups

[Ken Mueller]

Hello all, I am not able to deactivate 2 volume groups. vgchange -a n vgname
says device is busy. There are no filsystems mounted on either one. fuser -u for both lvols shows no processes. fuser -uk and then vgchange still fails. fuser -c says no filesystem mounted. What I did that probably has something to do with it is attempt to mount the same device to different mount points simultaneously:
mount -F vxfs /dev/vgredo1/lvol1 /circ/ordata/pddata/redo1
mount -F vxfs /dev/vgredo1/lvol1 /circ/ordata/pddata/redo3
(vgredo1 is one of the vgs that will not deactivate)
The other vg is vgredo2 - I have no idea why that one is claiming it is busy.
As another note, I tried vgscan -a and it hung!
Any help would be greatly appreciated...do I need to reboot to solve this quandry?
Thanks in advance, Ken

[Ken Mueller]

p.s. this is not a clustered/HA/service guard environment....

[S.K. Chan]

Assuming vgredo1 and vgredo2 are not root VGs. I can think of 2 possible reasons.
a) the vgs swap lvol or raw lvol.
b) you got duplicate minor# on "group".

For a), are there any other lvols in vgredo1 an vgredo2 that are not filesystem ? For b), check by running this ..

# ll /dev/vg*/group

and look for the minor# (typically 0x0XXXXX) and see if they are all unique.

As for your vgscan hung problem, you might have a disk which is offline. What does ..

# vgscan -pv

says?

[Ken Mueller]

thanks for the quick reply, here's the listing:

(dis:/etc)# ll /dev/vgredo*
/dev/vgredo1:
total 0
crw------- 1 root sys 64 0x0e0000 Feb 1 10:49 group
brw------- 1 root sys 64 0x0e0001 Feb 1 20:25 lvol1
brw------- 1 root sys 64 0x0e0002 Feb 1 20:25 lvol2
crw------- 1 root sys 64 0x0e0001 Feb 1 20:25 rlvol1
crw------- 1 root sys 64 0x0e0002 Feb 1 20:25 rlvol2
/dev/vgredo2:
total 0
crw------- 1 root sys 64 0x120000 Feb 1 11:09 group
brw------- 1 root sys 64 0x120001 Feb 1 20:27 lvol1
crw------- 1 root sys 64 0x120001 Feb 1 20:27 rlvol1

as you can see there are no duplicates between the 2 (I checked all the others and there aren't any dupes there either). Any other ideas?

vgscan -pv does show disks that it cannot query - but none that are part of the un-deactivatable vgs...and it still hangs at the end of the listing, immediately after displaying the disk associated with vgredo2:
vgscan: The physical volume "/dev/dsk/c9t1d0" is already recorded in the "/etc/lvmtab" file.

[Ken Mueller]

Also - these vgs do not contain swap area (at least as far as sam is concerned). I tried deleting the lvols using sam and it errored saying the lvols were open or there was a sparing operation in progress....

[Ken Mueller]

more info - these vgs are part of an EMC 4700 SAN....

[S.K. Chan]

Just trying to determine if you got a hot spare PV setup. If yes then that could be the reason why you can't deactivate the vg where that particular pv resides until we determine if we can "unspare" it. Could you provide this output ?

# vgdisplay -v vgredo1
# vgdisplay -v vgredo2
# lvdisplay -v /dev/vgredo1/lvol1
==> do this for all other lvols incl those in vgredo2
# pvdisplay -v
==> of all pv's in both vg's

[Ken Mueller]

Well, I finally just rebooted the machine and now all is fine. Thank you for taking the time to try and figure it out!
Thanks,
Ken

[S.K. Chan]

No problem,I'm glad it worked out well for you.