How to extend the Cluster Filesystem --new lun

[ani007]

What is the step by step procedure, I have to follow if one of my file system has to be increased which is part of VG in a two node MC serviceguard cluster.

Note: onlineJFS installed. No space left onthe VG ,so new lun is added.Package is running on primary node. There should not be any unavailability of applications.

NODE A
#pvcrerate /dev/rdsk/cxtxdx
#vgextend vgname /dev/dsk/cxtxdx
#lvextend -L XXXXM /dev/vgname/lvname
#fsadm -F vxfs -b XXXXM /mnt_point
#vgexport -p -v -s -m vgname.map /dev/vgname
#scp -p vgname.map :/home/

NODE B
#ll /dev/vgname/group [note the minor no]
#vgexport /dev/vgname
#mkdir /dev/vgname
#mknode /dev/vgname/group c 64
#vgimport -v -s -m /tmp/vgname.map /dev/vgname

Please check & let me know if the above steps are correct or not.After that in which file i have to update the lun details or what step i have to follow after these.

[Jayakrishnan G Naik]

Yeah Ani

This looks perfect.

Regards
Jayakrishnan G Naik

[ani007]

is it require to add the new lun information to cluster control file or some thing like this ?

[Jayakrishnan G Naik]

Hi Ani

Nothing required as such. When the vg config changes in one node that should be replicated to the other, and no lun information other than (if used) appears in cluster config file.

Cluster and package config/cntl files normally have vg and lv information and your task to add a pv to a cluster vg and extending it has nothing to do with config & control files.

Thanks & Regards
Jayakrishnan G Naik

[Jayakrishnan G Naik]

In my last reply, I missed to add something
which would have confused.


"When the vg config changes in one node that should be replicated to the other, and no lun information other than "FIRST_CLUSTER_LOCK_PV" (if used) appears in cluster config file.

Regards
Jayakrishnan G Naik

[Rahul Rai_1]

hi Ani,

Is your problem got fixed? steps which is mentioned by you is perfect and few things which is perfectly said buy JGN, go ahead and do this activity..... :)


Regards,
Rahul Rai.

[ani007]

Thanks all