VG exclusive mounted in 2 nodes

[Karuppiah Ganesan]

Hello All,

I have a very stange problem here.

There is a storage migration work and I have to do vgexport and vgimport to the cluster nodes. While I am doing this is a one of the first node(node1) I selected, I do see an error.

This is a Package VG is having only one disk with an alternate path in it.

I could not see the alternate path in the node1 and when I try to deactivate the VG that is mounted in the readonly mode it is giving the following error.

(In Node1)
# vgchange -a n /dev/vg001
vgchange: Warning: couldn't query physical volume "/dev/dsk/c55t13d1":
The specified path does not correspond to physical volume attached to
this volume group
vgchange: Warning: couldn't query all of the physical volumes.
vgchange: Failed to notify clvm daemon about volume group deactivation - No such file or directory
Couldn't deactivate volume group "/dev/vg001":
No such file or directory


Now when I see vgdisplay for that VG , now it is showing as exclusive and read/write.
Does this seems to be a strange one...:-) !!!!

Any help please....

Thanks in advance.

[Aneesh Mohan]

Hi ,

> # vgchange -a n /dev/vg001


is it vg01 or vg001 ?


Aneesh

[Karuppiah Ganesan]

Hi Aneesh,

This is vg001 only.

I have already given the vgchange -a n vg001 error message.

Thanks..

[Aneesh Mohan]

Please issue vgsync /dev/vg001 before doing vgchange .


Aneesh

[J. Bravo]

Hello:

Please, attach us the output of:

cmviewcl -v
vgdisplay -v vg001, in both nodes
strings /etc/lvmtab
ll /dev/vg001

Regards;

J. Bravo.

[Karuppiah Ganesan]

Hi Aneesh,

I am afraid if I do vgsync in the node1 will it affect the same cluster VG that is active in exclusive mode in the node4 (where the package is currently running).

Node_Switching_Parameters:
NODE_TYPE STATUS SWITCHING NAME
Primary up enabled node001
Alternate up enabled node004 (current)
Alternate up enabled node002
Alternate up enabled node003

[Karuppiah Ganesan]

Hi Aneesh,

I am afraid if I do vgsync in the node1 will it affect the same cluster VG that is active in exclusive mode in the node4 (where the package is currently running).

Node_Switching_Parameters:
NODE_TYPE STATUS SWITCHING NAME
Primary up enabled node1
Alternate up enabled node4 (current)
Alternate up enabled node2
Alternate up enabled node3

[Karuppiah Ganesan]

Hi Bravo,

Pls. see the attachment..

Thanks.

[J. Bravo]

Hi!

It seems to be that your alterante link is missing. Do you have tried to execute "ioscan -fnC disk"? Please, after execute the ioscan, run dmesg to view possible errors over the disk.

Regards;

J. Bravo.

[Steven E. Protter]

Shalom,

The simple solution would be to insure only one node has access to the shared storage and do:

vgreduce -f vg001

Then follow the instructions that come along after that.

The thing I've always liked about HP-UX LVM is that when you mount in exclusive mode, the second node is denied the ability to mount, versus LVM Linux which merrily lets two nodes mount the same volume group exclusive without even a peep.

SEP

[Karuppiah Ganesan]

Hi

#vgreduce /dev/vg001 /dev/dsk/c55t13d1
vgreduce: Couldn't query physical volume "/dev/dsk/c55t13d1":
The specified path does not correspond to physical volume attached to
this volume group
--------

#vgreduce -f /dev/vg001 - does not give any errors but no improvements... :(

ioscan /diskinfo - shows claim /OK status for this disk.

dmesg - not much information about this.

But,
# pvdisplay /dev/dsk/c55t13d1
pvdisplay: Warning: couldn't query physical volume "/dev/dsk/c55t13d1":
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/vg001".
pvdisplay: Cannot display physical volume "/dev/dsk/c55t13d1".


Hope this information helps to help me.

[Karuppiah Ganesan]

Backround info.

This VG is mounted in Readonly but due to unknown reason it turned read/write now in the other node (node1).

This is the strange part of the problem.

[J. Bravo · Accepted solution]

Hi:

Try with "vgreduce -l vg001 /dev/dsk/c55t13d1" and "vgextend vg001 /dev/dsk/c55t13d1"

Regards;

J. Bravo.

[Karuppiah Ganesan]

Hi Bravo,

vgreduce - worked sucessfully removed the affected PV link from lvmtab.
Thanks

Now I want to deactivate the vg - vg001

# vgchange -a n /dev/vg001
vgchange: Failed to notify clvm daemon about volume group deactivation - No such file or directory
Couldn't deactivate volume group "/dev/vg001":
No such file or directory

As this vg is cluster aware it is not allowiing me to deactivate.
Is it recomended to run "vgchange -c n vg001" while the package is up and running in the other node.

Please suggest..

Thanks.