can largefile be enabled on default hpux 11.00?

[colin chua_2]

Hi Guys,

My DBA needs to create a an Oracle table which is larger than 2GB in file size.

I need to find out if the above is possible provided I do not have online JFS option installed.

Does it mean that I have to juz do a newfs on a new lvol than I'm gonna create and ensure that fstab for the new lvol has largefiles option turned on?

Cheers!
Colin.

[Johnson Punniyalingam · Accepted solution]

To enable largefile support on an *existing* filesystem you will only need to unmount the filesystem if you do not have Online JFS and/or the filesystem isn't VxFS. To enable the support, use the raw device file:

# fsadm -F vxfs -o largefiles /dev/vgXX/rlvolY

The conversion takes just a few seconds.

To see if largefiles support is enabled or disabled for a filesystem, do:

# fsadm -F vsfs /mountpoint

The "HP-UX Large Files" whitepaper can be found here:

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

Thanks,
Johnson

[Johnson Punniyalingam]

Even if you don't have onlineJFS you can add the largefiles support to your logical volumes. If you have already created the lv, proceed like this,

fsadm -F vxfs /dev/vg_name/lv_name

Substitute vxfs with hfs, if you have hfs filesystems.The above command should return
nomultifilesets
nolargefiles

If it returns no largefiles, to add largefiles support,

umount /dev/vg_name/lv_name
fsadm -F vxfs -o largefiles /dev/vg_name/lv_name

Next modify the /etc/fstab file and add the option largefiles to the list of vxfs options so that the next time the filesystem gets mounted, it will be mounted with the largefiles enabled.

hope this helps.

Thanks,
Johnson

[Steven E. Protter]

Shalom,

No, you have to do -o largefiles when using the newfs command.

You could make an alias that always uses this command option to replace newfs.

The problem though is you for 11.00 must add a parameter to the filesystem mount in /etc/fstab and that requires a manual edit.

SEP

[colin chua_2]

Hi Johnson,

Thanks for the quick reply.
Will try it out and let you know.

Cheers!

[colin chua_2]

Btw would like to check with you guys also.

Would there be any issues or performance degradation if my system have a mixture of mount points with nolargefiles and largefiles running concurrently?

Thanks in advance.

[T G Manikandan]

We are just enabling the feature to create files > 2Gb. if there needs to be a file created with 2 Gb it creates and also creates other files.

There is no performance issue enabling large files or having several filesystems with largefiles and others with no largefiles enabled.

[Sajjad Sahir]

dear colin

first check online jfs is there or not

swlist -l product | grep -i jfs

if online jfs is there
fsadm -F vxfs -o largefiles /dev/vgname/rlvname

-o option will enable large files

thanks and regards

Sajjad Sahir

[Steven E. Protter]

Shalom,

Any performance degradation would be minimal. Probably not noticeable.

However on a practical basis only using largefiles where largefiles are likely to be needed does prevent a single large file from filling a filesystem.

Note also that 11.00 is out of support and largefiles support is built into /etc/fstab on subsequent OS releases without having to specify.

SEP

[James R. Ferguson]

Hi:

> SEP: Any performance degradation would be minimal.

There is _no_ performance penality associated with enabling largefiles for a filesystem. Its what you do with a large file.

> SEP: Note also that 11.00 is out of support and largefiles support is built into /etc/fstab on subsequent OS releases without having to specify.

The presence or absence of the 'largefiles' or 'nolargfiles' option in '/etc/fstab' is for mount-time protection only. The state of the flag in the _filesystem_ is what matters. This has been documented since the inception of large file support (in 10.x):

http://docs.hp.com/en/940/lgfiles4.pdf

Regards!

...JRF...

[Johnson Punniyalingam]

Hi Colin,

>>Btw would like to check with you guys also.

Would there be any issues or performance degradation if my system have a mixture of mount points with nolargefiles and largefiles running concurrently?
<<

NO, It only enable to support large files,

Cheers,
Johnson

[Dennis Handly]

>I do not have online JFS option installed.

I don't know about 11.00 but you don't need it to run fsadm, even if mounted.

[Michael Steele_2]

HI

"...If even one large file is detected on the filesystem being converted, then the fsadm command will not convert the filesystem. Therefore, if it is necessary to convert a large-files filesystem that actually has large files on it to a no-large-files filesystem, the large files must be
removed before conversion..."

http://docs.hp.com/en/940/lgfiles4.pdf

See above largefiles whitepaper which OK's the conversion fron NO largefiles to YES largefiles using the 'fsadm' command but only on the RAW LVOL and only when the file system is mounted and only if there currently are no largefiles.

"...the large files must be removed before conversion..." is clearly stated in the whitepaper.

Note that I could find no publication date on this whitepaper and that it was found in this April 2005 thread:

http://forums13.itrc.hp.com/service/forums/questionanswer.do?admit=109447627+1265173882497+28353475&threadId=857114

[Michael Steele_2]

Sorry, typo

"...file system is UNmounted..."

The white paper says do not run this fsadm command on a mounted filesystem.

[Michael Steele_2]

Sorry, I do this once in awhile, post an answer to an older thread used in research. Please ignore this closed thread.