file system extending issue..

[gany59]

Hello Folks,

I have gone through many forumns related to the file system extension in the cluster servers.
1.If we need to extend the file system on a cluster servers, is it need to halt the package in all servers in that cluster group, and shall we need to deactivate that specific vg before doing the file system extension.

2.concern like, after if we doing the vgimport in the second node, thats all the steps for doing the file system extension.

Thanks in advance, pls let me know the above things.

[singh sanjeev]

1 >NO cluster halt not needed for fs extention it is done online without intrupting clsuter.

2> if you extended the LV on one node..vgimport and vgimport will not effect much.[But it is good practice to do it] unless you have added new disk for expansion of file system. in that case you have to do :
#create a mapfile with vgexport
#then vgimport the mapfile on alternate node.

what is your case??

[gany59]

hello Singh,

Actually i need to add a new disk in the existing VG and thn to extend the file system size.

for the above activity, as per my knowledge i mentioned the steps below, so if the thing is wrong somewhere, pls correct that one, coz i am hardly knowing the MC/SG

1.pvcreate -f /dev/rdsk/c10t5d1

2.vgextend /dev/vg01 /dev/dsk/c0t0d1

then the details will be saved in /etc/lvmconf/vg01.conf

Sample command to grow an LVOL to 40GB size:

3.# lvextend -L 40000 /dev/vg01/lvol1

4.# fsadm -F vxfs -b 4000 /pbdata

5.# vgexport -pvsm /tmp/vg01.map /dev/vg01

copy the same /tmp/vg01.map file to the adoptive node
and check the minor number in the primary and adoptive node as identical by the command
any way if the cluster is there it will be identical only, so we need to worry abt that.


7.Recreate the dir and group file

8.Mkdir /dev/vg01

9.Mknod /dev/vg01/group c 64 0x040000

10.vgimport -vs -m /tmp/vg01.map /dev/vg01

[stephen peng]

>3.# lvextend -L 40000 /dev/vg01/lvol1

>4.# fsadm -F vxfs -b 4000 /pbdata
usually I use extendfs to extend filesystem, not fsadm, are you employing OnlineJFS?
>7.Recreate the dir and group file

>8.Mkdir /dev/vg01

>9.Mknod /dev/vg01/group c 64 0x040000
before these steps, you should vgexport vg01 at your adoptive node.

[gany59]

thanks for the valuable input,

what is the need for doing the vgexport vg01 in the adoptive node, is it necessary..

[stephen peng]

The vgimport command adds the specified volume group to the system, to my understanding, it create a new vg to the system,so you should use vgexport vg01 will clear old configurtion of vg01 at adoptive node, you know , the old vg01's configuration did not contain new extended PV on primary node.

sorry for my poor english expression.

regards
stephen

[gany59]

thanks stephen...

[Benoy Daniel]

you forgot "m" in the fsadm step.

fsadm -F vxfs -b 4000m /pbdata

[GPissutto]

If your LVOL's 40000 MB in size, you must use this other value (40000*1024), so:

# fsadm -b 40000*1024 /mount_point

If you have MC/SG, is very likely you already have OnlineJFS.

[madhuchakkaravarthy]

hi gany

1.add required size from storage and present to the particular node
2.do ioscan and insf -e and check with dmesg for any error

3. pvcreate
4.vgextend
5.lvextend
6.fsadm
7.bdf --------->check for size incresed
8.vgexport -p -s -v -m /tmp/vg.map /dev/vgname

copy the map to fail over

in failover

vgexport /dev/vgname

8.vgimport -s -v -N -m mapfile vgname

down the db or app and halt pkg in main server and run pkg in failover

9. bdf check for size increased

10. halt pkg in failover and run in mainserver.


but before starting take the vg configuration backup vgcfgbackup or map file of that particular vg.

regards

MC