[Samba] tdb search optimization

[Bruno Gomes Pessanha]

I'm running:
# smbd -V
Version 3.0.24-2.28-1354-SUSE-CODE10

In a 600 print queues print server, but I'm experiencing a low performance while
using the first queue. I was attaching ltrace/strace to some processes with "use
mmap=no" and I found a BIG difference in information gathering in ntprinters.tdb
between the first and the last print queue (using alphabetical sorting).

This is the difference between the two strace outputs while clicking in printer
properties option:
-rw-r--r-- 1 root root 261K Nov  7 16:59 strace_first_printer.txt
-rw-r--r-- 1 root root 9.1M Nov  7 17:00 strace_last_printer.txt

Someone could explain why this behavior? 

Regards,

Bruno Gomes Pessanha