Solaris disaster recovery with networker

We have a windows 2003 networker cluster running networker 7.2.2.
We have two storage nodes and two L700 libaries with 4 lto2 drives each.

We are trying to restore two solaris systems that are exactly the same. We first install a new (same hardware platform) solaris system with jumpstart. Then we install the networker client. We then start "recover -s -c . On the prompt we specify "add / " More than 80000 files are selected for restore. We then delete a lot of files and start the recover. After the completion we reboot the server and everything is fine.
However on the other server we do the same. But when we enter "add / "only 4000 files are selected for restore. On this client we have to specify the filesystems one at a time.
After the restore the server also doesn't start. Does anybody have a clue why the initial "add /" doesn't select all files? Or have any ideas why the second server works different than the first.

Proof:

bash-2.03# recover -f -q -c <2nd server>
recover: Current working directory is /
recover> add /
4470 file(s) marked for recovery
recover> exit
bash-2.03# recover -f -q -c
recover: Current working directory is /
recover> add /
81970 file(s) marked for recovery

Thanks for any help.