problem in file system extension

[Muhammad Bilal Attique]

Dear All
i am extending the file system in mcservice guard enviornment.
i have done following steps
create the vdisk in san
present it to the servers
ioscan -fnc disk in current node
insf -e
pvcreate /dev/rdsk/cXtYdZvgextend /dev/vgarch /dev/dsk/cXtYdZ
when i given this command it showed that vg extended succsesfully
the system did not respond then i open a new session and given this command
strings /etc/lvmtab |more
it showed me the new created disk in the vg
/dev/vgdata
/dev/dsk/c13t0d4
/dev/dsk/c13t0d6
then i given the command vgdisplay
--- Physical volumes ---
PV Name /dev/dsk/c13t0d4
PV Status available
Total PE 35323
Free PE 1
Autoswitch On

PV Name /dev/dsk/c13t0d6
PV Status available
Total PE 15997
Free PE 15997
Autoswitch On

Then finally this command
lvextend -l 51320 /dev/vgdata/lvdata (total pe )

after this command system given me this message

Cannot lock "/etc/lvmconf//lvm_lock" still trying ...

and stuck and no response at all and did not show the mount point
after several time it showed the mount point but when we try to stop the cluster it did not stop the cluster and finally have to restart the machine.
now the cluster is up and database is running

please help what to do to extend the system

[Mike Shilladay]

Hi,

A process on your machine was still active. For example if you issue a mirror command, start another session and issue another mirror command on the same vg, then the message "Cannot lock "/etc/lvmconf//lvm_lock" still trying ..." will be displayed until the first mirror has completed.

So I can make the assumption that a process was still active on your machine.

Did the vgextend command complete successfully ?

Mike.

[Muhammad Bilal Attique]

when i ginven this command it showed this message only
vgextend /dev/vgdata /dev/dsk/c13t0d6
Volume group "/dev/vgdata" has been successfully extended.
***************
while when i given this command some time before on another system it also give me this message
vgextend /dev/vgarch /dev/dsk/c15t0d5
Volume group "/dev/vgarch" has been successfully extended.
Volume Group configuration for /dev/vgarch has been saved in /etc/lvmconf/vgarch.conf

[Mike Shilladay]

Hi,

Do you have by any chance a volume group that is active on two servers ?

Mike.

[Muhammad Bilal Attique]

I do not under stand that what u mean

[Mike Shilladay]

Hi,

What I was trying to say is, if you extended the volume group on one server, (e.g. vgchange -a n ), did you disable the volume group before extending it on another server ?

Mike.

[AISSAOUI Nourreddine]

Hello,

An LVM command is running or hanging and did not release the lock. You can use lsof to identify the process :

#lsof /etc/lvmconf//lvm_lock


Regards,
Nourreddine

[Muhammad Bilal Attique]

nope
the procegure is to extend the lv on main node
pvcreate /dev/rdsk/cXtYdZ
vgextend /dev/vgarch /dev/dsk/cXtYdZ
lvextend â l /dev/vgarch/ lvarch
fsadm â F vxfs â b m /mountpoint
mkdir /tmp/mapfilea
vgexport â p â v â s â m /tmp/mapfilea/vgarch.map /dev/vgarch
FTP all map in /tmp/mapfile to other node

then to perform the steps on other node

ioscan â fnC disk
insf â e
ls -l /dev/*/group
vgexport /dev/vgarch

mkdir /dev/vgarch
mknod /dev/vgarch/group c 64 0x030000
vgimport â v â s â m /tmp/mapfilea/vgarch.map /dev/vgarch

[Mike Shilladay · Accepted solution]

Hi,

I would have personally deactivated the VG on the main node before importing it. That way there is no conflict between servers.

Were you attempting to extend the vg on an active service within the cluster or did you stop the service first? (It would be recommended to stop the service first)

On your machine, can you see all the disks in your VG ? Can you do a vgdisplay on the volume group in question. If so, can you do an lvdisplay on the lv /dev/vgdata/lvdata, if that works do it report the correct values for what you expect? If not, can you make another small volume (1 extent) on vgdata, does it still report the "cannot lock /etc/lvmconf//lvm_lock" still trying message" ? If it does, then a "ps -ef | grep vg" command will help identify the problem process, or a "ps -ef | grep lv" will also help.

Just out of interest, if you have a lock disk in this cluster, can you verify it os not one of the disks you are working on ?

There are a few ideas to follow, hopefully, something may turn up in that lot. Let me know.

Mike.

[Muhammad Bilal Attique]

can any snior member help me

[Tor-Arne Nostdal]

Hi Muhammad.
The crown and leaves show activity on the forum. The answers and questions from Mike is relevant :-) - even if you don't understand them so easily.
Mike is an ITRC member since: July 25, 1997
That should perhaps indicate some "seniority"...

The more frequent visitors here would soon place a message if they think he's not giving appropriate answers.

Another way to attract their attention is to assign points - and so far you haven't done this to the answers you've got

/2r