Can I write a file that is bigger than 2 gigs.

[Shar Hunter]

Wow! You guys are really smart.
How about this one:
I am trying to backup my database but it keeps failing around 2 gigs. Is there anyway that I can write a 4.2 gig file to my system?

[Jean-Luc Oudart · Accepted solution]

You need to enable large file option for your file system.

lots of threads on the subject
http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=204549
etc ...

Regards,
Jean-Luc

[Mark Grant]

Yes, you just need to have the filesystem mounted with the "largefiles" option enabled. You create the filesystem and mount it with "-o largefiles" as an option. You can change an existing filesystem to support largefiles with "fsadm -F vxfs -o largefiles /dev/vgXX/rlvolXX"

[RAC_1]

You should have large file enabled on filesystem. Check if you largefiles set.

How you enable it?

fsadm -F vxfs -o largefiles /mount_point

[Mark Grant]

Doh!

RAC's syntax is correct!

[Jeff Schussele]

Hi Shar,

The filesystem into which you're writing the files must support largefiles.

To determine whether it does run the following:

fsadm /mnt_point

To set it for largefiles run the following

fsadm -o largefiles /mnt_point

Then check the /etc/fstab to make sure it's entry does not say nolargefiles. If it does change it to largefiles.

HTH,
Jeff

[Robert Kerr]

You can set a large files option, see

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=34154

[Elena Leontieva]

Shar,

You want to enable largefiles option:

umount /fs_name
fsadm -F fs_type -o largefiles /dev/vg_name/rlvol_name
mount -a

Check that it worked:
fsadm -F fs_type /fs_name

HTH,
Elena.

[Geoff Wild]

Check the filesystem:

fsadm /filesystem

Example:

# fsadm /var/adm/crash
fsadm: /etc/default/fs is used for determining the file system type
largefiles

That means 2GB + is supported.

# fsadm /
fsadm: /etc/default/fs is used for determining the file system type
nolargefiles

That means you can NOT write files greater then 2 GB.

As other have said, use fsadm to enable:

fsadm -o largefiles /var/adm/crash

Then update the /etc/fstab:

/dev/vg01/lvol2 /var/adm/crash vxfs rw,suid,largefiles,delaylog,datainlog 0 2


Rgds...Geoff

[Patrick Wallek]

How are you trying to backup your DB? Are you just copying or are you trying to compress or gzip the files as they are being backed up?

Your problem may not be with your filesystem but rather with the backup method you are using.

[Shar Hunter]

Thank You!

[Shar Hunter]

It's not working.
Here is what I get:

# fsadm -o largefiles /dev/vg01/test
vxfs fsadm: /dev/vg01/test is not the root inode of a vxfs file sytem
#


# bdf /test
Filesystem kbytes used avail %used Mounted on
/dev/vg01/test 9216000 3232770 5800650 36% /test
#

# fstyp -v /dev/vg01/test
vxfs
version: 3
f_bsize: 8192
f_frsize: 2048
f_blocks: 4608000
f_bfree: 2991615
f_bavail: 2991615
f_files: 42240
f_ffree: 1073792296
f_favail: 1073792296
f_fsid: 1073807363
f_basetype: vxfs
f_namemax: 254
f_magic: a501fcf5
f_featurebits: 0
f_flag: 0
f_fsindex: 6
f_size: 4608000

[Patrick Wallek]

You use the actual mount point name:

# fsabm -F vxfs -o largefiles /test

[Shar Hunter]

I am getting closer....

# fsadm -F vxfs -o largefiles /test
fsadm: set feature (LARGEFILES) failed, errno 25

[Jeff Schussele]

Hi Shar,

OK - let's take a step back here.
What HP-UX OS version are you running?

uname -r

Rgds,
Jeff

[Shar Hunter]

Thanks for the command!

# uname -r
B.11.00

[Shar Hunter]

Should I unmount the file system first?

[Shar Hunter]

I have been reading some other threads.
I do not have online jfs so I think I need to umount it.

I think I need to do :

umount /test
# fsadm -F vxfs -o largefiles /test

Then I am going to modify the fstab file
from:
# System /etc/fstab
/dev/vg01/test /test vxfs delaylog 0 2

to:
/dev/vg01/test /test vxfs largefiles,delaylog 0 2

Then
mount /test

What do you think?
Would you do this while users are logged on (no one but me accesses the test directory)

[Jeff Schussele]

Yep - you're picking this up fast.
Indeed, you do need OnLine JFS to do -o largefiles while mounted. So unmount, change & remount.

If you're concerned about users being on run

fuser -cu /test

to see whether anyone else is in there. Make sure you're not when you do this or it will show you in there.

Rgds,
Jeff

[Shar Hunter]

I umounted /test and no one noticed! Yipee!

Now I am back to
# fsadm -F vxfs -o largefiles /test
vxfs fsadm: /test is not the root inode of a vxfs file sytem

I looked at the man page and it said:

The fsadm command is designed to perform selected administration tasks
on file systems. These tasks may differ between file system types.
special is a device file containing an unmounted file system. However,
if the file system is of the type that provides online administration
capabilities the special could be a directory. directory must be the
root of a mounted file system.

I wish Iunderstood what a "root of a mounted files system was".

[Shar Hunter]

GOOSSENS is KING!!!!!!!
I found a message from him back on Dec 12th.
It said

fsadm -F vxfs -o largefiles /dev/vg01/rora03
(note r in rora03 )

I did fsadm -F vxfs -o largefiles /dev/vg02/rtest.

IT WORKED!
Thanks everyone!

[Mark Grant]

Seems my syntax wasn't wrong after all :)

[Ulrich Deiters]

Don't forget to set the correct compiler options! The HP C compiler has to be invoked with "cc -Ae -D_FILE64 ..." in order to provide file pointers of more than 32 bit.

[RAC_1]

Now I am commenting on does it have to be raw vol for fsadm syntax?

I think it is this way.

You have online JFS, fsadm works on raw or mount point. If you do not have online -jfs, it works on raw vol.

Am I making sense here?