vxdump and LTO tape size

[Lionel MAZE]

Hi,

When I do a backup of the 2 following filesystems on an LTO :
/dev/vg01/lvol3 102400000 98470416 3899008 96% /freezedata
/dev/vg01/lvol4 20971520 20688640 278516 99% /freezeapps

The first filesystem is entirely backup on the tape but not the second.
Why does another tape require whereas a tape can contain 200 GB and that I backed up hardly 100 GB?
+ vxdump -0 -B 200g -f /dev/rmt/2mn /freezedata
vxdump: Date of this level 0 dump: Tue Nov 26 23:01:17 2002
vxdump: Date of last level 0 dump: the epoch
vxdump: Dumping /dev/vg01/rlvol3 to /dev/rmt/2mn
vxdump: mapping (Pass I) [regular files]
vxdump: mapping (Pass II) [directories]
vxdump: estimated 196933852 blocks (96159.11MB) on 0.47 tape(s).
vxdump: dumping (Pass III) [directories]
vxdump: dumping (Pass IV) [regular files]
vxdump: 4.47% done, finished in 1:46
...
vxdump: 99.10% done, finished in 0:00
vxdump: vxdump: 98657170 tape blocks on 1 volumes(s)
vxdump: Closing /dev/rmt/2mn
vxdump: vxdump is done
+ vxdump -0 -B 200g -f /dev/rmt/2mn /freezeapps
vxdump: Date of this level 0 dump: Wed Nov 27 00:52:32 2002
vxdump: Date of last level 0 dump: the epoch
vxdump: Dumping /dev/vg01/rlvol4 to /dev/rmt/2mn
vxdump: mapping (Pass I) [regular files]
vxdump: mapping (Pass II) [directories]
vxdump: estimated 41006002 blocks (20022.46MB) on 0.10 tape(s).
vxdump: dumping (Pass III) [directories]
vxdump: dumping (Pass IV) [regular files]
vxdump: 14.39% done, finished in 0:29
vxdump: End of tape detected
vxdump: Closing /dev/rmt/2mn
vxdump: Change Volumes: Mount volume #2
vxdump: fopen on /dev/tty fails: No such device or address
vxdump: The ENTIRE dump is aborted.

[Lionel MAZE]

I am on HP-UX 11.00 with OnlineJFS.

[John Palmer]

200Gb assumes a compression ratio by the drive of 2:1.

Do these filesystems consist mostly of data that's already compressed? If so, 100Gb is your limit.

Regards,
John

[Lionel MAZE]

On /freezedata, there is only Oracle datafile that aren't compressed.
I use the same technique on another machine of development with the same data and an identical volumetry and that works.
Another idea ???