Increaze the size of logical volume

[Fuad_1]

I am trying to increase the size of /usr, but unmounting can not proceed because many processes are running. How can I kill or end these proceses?

[Duncan Edmonstone · Accepted solution]

You can't - /usr holds files vital to operation of the OS in multi-user mode. You need to boot the system to single user mode to allow you to do this.

Once you have ll your users off:

cd / ; shutdown -r -y 0

Now from the system console wait for the prompt to give you the opportunity to interrupt the boot sequence. Once interrupted type 'boot' and answer 'y' to the question 'interact with ISL?' Once at the ISL prompt type 'hpux -is' to go to single user mode.

Now you can extend your LV/filesystem in the usual way:

lvextend -L /dev/vg00/lvolXX
extendfs -F /dev/vg00/rlvolXX

Of course if you have Online JFS installed (check using swlist) You can just extend the /usr file system on-line:

lvextend -L /dev/vg00/lvolXX
fsadm -F vxfs -b /home

HTH

Duncan

[Shahul]

Hi

Do U have Online JFS Installed in ur BOX? If Yes, Then U must be able to increase /usr from ur current runlevel. Follow this

#lvextend -L XXX /dev/vg00/lvolX

Here XXX is the final size of /usr

#fsadm -F vxfs -b XXXm /usr

Suppose, U don't have Online JFS Installed, Then Follow below procedure

Plan for an outage after getting the permission.

Shutdown the system, While booting disturb the auto boot by pressing esc key. U will come to main menu, Here U type "boot pri", Now system will ask U, Whether U want to interact with ISL or not? Select yes. In ISL prompt type like this

ISL>hpux -is

Now system will take U to single user mode. Then follow this

#lvextend -L XXX /dev/vg00/lvolX

#extendfs -F vxfs /dev/vg00/lvolX

Check up U are OK or not by bdf command.


Best of luck
Shahul

[Sanjay_6]

Hi,

Try this link,

http://support1.itrc.hp.com/service/cki/docDisplay.do?docLocale=en_US&docId=200000046785431

You can also do this by taking the system to single user mode, then unmounting and extending /usr like any other lv.

Hope this helps.

Regds

[Yogeeraj_1]

Hello,

Did the following sometimes back.
Hope this helps!
==============================================
a. Do a full backup of you server.

b. Identify the logical volume and disk information
#bdf|grep usr
/dev/vg00/lvol7 426837 283672 100481 74% /usr

c. View the disk information
lvdisplay -v /dev/vg00/lvol7 | more

Verify Free PE's

d. Reboot the server in single user mode

shutdown -r now
bo pri isl
interact with ISL? Y
hpux -is

Thus, nothing will be mounted

e. Extend /usr using the command:
/sbin/lvextend -L 2048 /dev/vg00/lvol7
/sbin/extendfs -F hfs /dev/vg00/rlvol7

f. Mount and verify the file system

g. Reboot server to normal running state
shutdown -r now

===============================================

Best Regards
Yogeeraj

[Martin Johnson]

I have found yogeeraj's suggestion has worked for me in the past. However, on 10.20 systems I have had to do a "mount /usr; umount /usr" to get the lvextend command to work correctly.


Marty

[Glen Kilpatrick]

Fuad,

I'd add the following recommendations:

[1] Do the LVM in multi-user, and boot to single-user for the extendfs. That way you won't schedule and perform a reboot only to find the dreaded "Insufficient Free PEs". And extending the partition in multi-user should present no problems even in a production environment (you just can't use the space 'til the filesystem is expanded).

[2] I believe it to be mostly harmless :) to extendfs or fsadm filesystems where bdf shows ~85% used or less. 95% or more, you may be tempting fate, particularly with fsadm, as the command(s) require some "working room" in the existing filesystem that's being extended. In that case, I'd either tempt fate if I had a fresh backup (remember, there's a /sbin/frecover, and a /sbin/tar, but that's it, no /sbin/omniback, _etc._), or "offload" some of those files elsewhere (tape, another filesystem, whatever) until the percentage used dropped to what you consider safe.

[3] For VxFS, don't play with unpatched systems. Patching is a Big Subject, many pitfalls there. But whatever, when HP discovers "undocumented festures", and fixes them, ignoring those patches may increase the possibility that you will RE-discover something already known. And most VxFS problems, the commands may be fixed by patching, but already busted filesystems aren't healed....

gk
:)