topic Re: implementing largefiles on shared FS in Operating System - HP-UX

implementing largefiles on shared FS

stephen corcoran_1 — Wed, 07 Dec 2005 07:08:05 GMT

Hi ,

Can anyone tell me the recommended way of implementing largefiles on a filesystem that is shared over a cluster.

If I try to implement largefiles using SAM it gives me a warning that I must remove the FS from the cluster first

I was thinking about using the following process, if there is an easier way where I dont have to take down the cluster I would love to know

Much appreciated
Steve

Shutdown cluster

Vgchange –a n /dev/vg01 – this will take the /shared FS out of control of cluster

Manually mount /shared locally on one node.

Run fsadm to implement “largefiles”

Unmount /shared

Vgchange –a y /dev/vg01 – Put /shared FS (vg01 volume group) back into cluster control

Restart the cluster

Would it be possible to do the following:

Edit the fstab to contain largefiles and just restart the cluster ?

Re: implementing largefiles on shared FS

Muthukumar_5 — Wed, 07 Dec 2005 07:22:20 GMT

Is /shared file system having online administration capablities like Online JFS? If so you can directly change the attributes. Else you have to your steps.

hth.

Re: implementing largefiles on shared FS

stephen corcoran_1 — Wed, 07 Dec 2005 07:38:07 GMT

Thanks for the response.

How do I check whether I am using Online JFS

Steve

Re: implementing largefiles on shared FS

Devender Khatana — Thu, 08 Dec 2005 00:36:28 GMT

Hi,

#swlist |grep -i online

Will tell if you have it or not.

Also your vgchange steps are reversed.
For taking control of the VG after shutting down cluster you should use
#vgchange -a y /dev/vg01

Followed by the changes before starting the cluster you should do

#vgchange -a n /dev/vg01

Also making entried in /etc/fstab and restarting cluster will enable it provided the file system was initially created using largefiles option.

HTH,
Devender

Re: implementing largefiles on shared FS

Devender Khatana — Thu, 08 Dec 2005 05:36:58 GMT

Hi,

Just another correction.

The entries made to /etc/fstab will not take effect when the file systems are mounted using a cluster conf file. Instead you need to make entries in the conf file itself in the section where you define mounting.

LV[0]="/dev/vx/dsk/dg01/dvebmgs00"; FS[0]="/usr/sap/PRD/DVEBMGS00"; FS_MOUNT_OPT[0]="-o suid,delaylog,convosync=direct"; FS_UMOUNT_OPT[0]=""; FS_FSCK_OPT[0]=""; FS
_TYPE[0]="vxfs"

But again for the option to be enabled here the file system should be prepared to allow this option.

HTH,
Devender