MC/SG cannot deactivate VG

[Rainer von Bongartz]

upon shutdown a package MC/SG could not deactivate the volume group:

Error message:
vgchange: Failed to notify clvm daemon about volume group deactivation - No such
file or directory
Couldn't deactivate volume group "vgtest":

No upon startup I get:

########### Node "neptun": Starting package at Wed May 15 10:32:11 METDS
T 2002 ###########
May 15 10:32:11 - "neptun": Activating volume group vgtest with exclusive option
.
Volume group "vgtest" has been successfully changed.
/dev/vgtest/test: No such device or address
May 15 10:32:12 - Node "neptun": Checking filesystems:
/dev/vgtest/test
Can't open /dev/vgtest/rtest, errno = 6
ERROR: Function check_and_mount
ERROR: Failed to fsck one of the logical volumes.
May 15 10:32:16 - Node "neptun": Deactivating volume group vgtest
vgchange: Failed to notify clvm daemon about volume group deactivation - No such
file or directory
Couldn't deactivate volume group "vgtest":
No such file or directory
ERROR: Function deactivate_volume_group
ERROR: Failed to deactivate vgtest

########### Node "neptun": Package start failed at Wed May 15 10:32:17
METDST 2002 ###########

lvdisplay -v seems to be ok.

lvdisplay /dev/vgtest/test
--- Logical volumes ---
LV Name /dev/vgtest/test
VG Name /dev/vgtest
LV Permission read/write
LV Status available/syncd
Mirror copies 0
Consistency Recovery MWC
Schedule parallel
LV Size (Mbytes) 500
Current LE 125
Allocated PE 125
Stripes 0
Stripe Size (Kbytes) 0
Bad block on
Allocation strict
IO Timeout (Seconds) default



Any ideas

Regards
Rainer

[Frederic Sevestre]

Hi,

Did you check the LV on both nodes ?
Did you import your VG on all the nodes ?

Fr??d??ric

[melvyn burnard]

The errno 6 is no such device or address.
Have you checked the major/minor nuimbers for this vg?
do you have a conflict?

ll /dev/vg*
and chcek it through.

[Sukant Naik]

Hi Rainer,

Have you added the logical volume /dev/vgtest/test on only one node and forgot to do a vgexport -p again ?

After making changes to the Volume group on one node, like adding or deleting Logical volumes, u need to do a vgexport and vgimport on the other node.

# vgexport -p -s -m /tmp/vgtest.map /dev/vgtest
# rcp /tmp/vgtest.map othernode:/tmp

On the other node,
do a vgexport of the existing /dev/vgtest and then do the following
# vgexport /dev/vgtest
# mkdir /dev/vgtest
# mknod /dev/vgtest/group c 64 0x0n0000
where n is the minor number
# vgimport /tmp/vgtest.map /dev/vgtest

-Sukant

[Rainer von Bongartz]

Melvyn,

there seems to be no conflict:

/dev/vgtest:
total 0
cr--r--r-- 1 root sys 64 0x030000 May 2 10:08 group
cr--r----- 1 root sys 64 0x030001 May 2 10:08 rtest
:q!-r----- 1 root sys 64 0x030001 May 2 10:08 test

Major/Minor is unique across all other VG's

Regards
Rainer

[Sridhar Bhaskarla]

Hi Rainer,

Looks like you have a logical volume /dev/vgtest/test defined in your Service Guard control file but no longer exists.

Do an ll /dev/*/group and check for uniqueness of the groups.

Probably export vgtest and import it back and then try.

-Sri

[Rainer von Bongartz]

I cannot import VG because it is still active and de-activation gives again:

vgchange -a n vgtest
vgchange: Failed to notify clvm daemon about volume group deactivation - No such
file or directory
Couldn't deactivate volume group "vgtest":
No such file or directory

[Sridhar Bhaskarla · Accepted solution]

Rainer,

Sorry to say this but looks like there is some problem with the group file that is not letting you to do anything.

You may need to reboot the box if the packages are down.

-Sri

[Animesh Chakraborty]

Hi,
Try halting the cluster then startit
#cmhaltcl
#cmruncl
These process should run:
root 18 0 0 Mar 31 ? 146:49 lvmkd
root 19 0 0 Mar 31 ? 147:18 lvmkd
root 20 0 1 Mar 31 ? 146:12 lvmkd
root 21 0 0 Mar 31 ? 145:54 lvmkd
root 22 0 0 Mar 31 ? 143:52 lvmkd
root 23 0 1 Mar 31 ? 146:48 lvmkd
root 24 0 0 Mar 31 ? 0:00 lvmschedd
root 1476 0 0 Mar 31 ? 0:00 slvm_sorcvd
root 1369 1367 254 Mar 31 ? 5339:26 /usr/lbin/cmlvmd

[Rainer von Bongartz]

Animesh,


The daemon lvmschedd is not running on my box.

I don't know what this daemon is doing . Can I start it manually or should I reboot as mentioned earlier?


Regards
Rainer

[melvyn burnard]

Hmm, do not see how you can cmhaltcl if the vg will not deactivate
A reboot may be the way to go, and then retest.

[melvyn burnard]

Are you perhaps using BCV's?
If so, have you run the vgchid command on his vg/bcv?

Just a thought, as this could have resuklted in duplicate vgid's

[Rainer von Bongartz]

Reboot solved the problem; VG is behaving normal again


Regards
Rainer