> Exactly now, I have 120 connections to the database. NO ONE is actually using
> or using the database for nothing, just connected (0% of use of all 16
> processors, no HDD usage...)
>
> So, I get another fb_lock_print and it shows:
>
> Mutex wait: 20.3%
>
> Hash slots: 4093, Hash lengths (min/avg/max): 9/ 19/ 34
>
> Weird! No HDD usage, no process... Why it shows this? I suppose that it must
> be Mutex wait: 0%! Why it's not?

"Mutex Wait" values represent the % of locks requests/attempts which were blocked (since the lock manager started). It *does not* represent the level of current activity.

> Now, I will change page cache from 2000 down to 1000 and rise up lock
> hashes for 10009 and tomorrow I will take a look again.

The "Hash slots" values suggest that a larger number of slots is in order (avg and max values > 15 generally indicate that more slots are required)


Sean

P.S. Homage to Helen B for the excellent details in the Firebird Book!!!