Largefiles support

[Victor Lo]

I would like to enable the largefiles support on my system. There are two volumes created. The content of fstab shows,

/dev/vg00/lvol3 / vxfs
/dev/vg00/lvol1 /stand hfs

From what I understand the correct procedures are,

umount /
fsadm -F vxfs -o largefiles /dev/vg00/lrvol3

But the root is mounted to the vxfs, I may not able to umount it. Will the online JFS help in this situation or the only option is to newfs the volume?

How about if the filesystem is a hfs instead of vxfs, will it be less complicated?

Thanks,

[Patrick Wallek · Accepted solution]

Yes, online JFS would help. In that case you would just have to do:

# fsadm -F vxfs -o largefiles /

Note that it would use the mounted filesystem name instead of the logical volume name.

If you really want to do this you will probably have to reboot the machine, come up in LVM maintenance mode (hpux -lm I think -- note this is NOT single-user mode), do the fsadm, make sure you /etc/fstab is correct and then reboot.

[Pete Randall]

Victor,

I was unaware of a requirement that the file system be unmounted and I can't imagine how you could unmount /, even in single user mode. I do have to ask, though, why you need large files support in these two file systems. They should be relatively static and there shouldn't be any huge files in them unless you're storing something besides the operating system essentials in them. Any other data should be place in a separate volume group.


Pete

[Michael Steele_2]

/ (* root *) does not require largefiles. That doesn't mean you can't proceed with your plan to convert / (* root *) into online JFS.

And as for /stand, its HFS, needs to remain HFS, so there will never be an online command to issue.

Here's the HFS large file option.

/usr/sbin/mkfs -F hfs -o largefiles /dev/vg02/rlvol1

http://docs.hp.com/hpux/onlinedocs/os/lgfiles4.pdf

[Todd McDaniel_1]

Victor,

Patrick is correct. You dont have to umount any filesystem to assign largefiles option to it, it can be done online without problem.

My contribution to your question is this:

Largefiles as you know is for 2GB or larger. Typically largefiles are used primarily and most often for database files and less often for Application filesystems.

These filesystems are typically NOT OS filesystems. The nature of most UNIX Platforms is to maintain OS files below 2GB As many 3rd party tools dont support >2GB files for such things as tar/cpio/ftp..

HPUX r* commands and HP ftp do support >2GB but many other OS may not.

Victor, I am just curious what your reasoning is to implement Largefiles on your OS? especially on base OS filesystems such as / and /stand.

I doubt either of those filesystems are 2gb in total, more like 500-1g at most.

Mine are very small in comparison to that:

# bdf /
Filesystem .... kbytes used avail %used Mounted on
/dev/vg00/root 258048 146211 104902 58%

# bdf /stand
Filesystem ........ kbytes used avail %used Mounted on
/dev/vg00/stand 199381 53861 125581 30% /stand