Thank you for your suggestion. I know that our configuration is not optimal and I will change this. But my understanding was, that this is only a performance problem and not something that can cause Firebird to hang. Or is this a known behavior?

<SL> The problem is in the definition of “hang”.

<SL> Does Firebird hang, ie. plain stop?  No

<SL> Can the deadlock scans and lock contention (see Mutex Wait) make Firebird appear to be hung/running really slow?  Yes

<SL> FYI, we are considering changing from Classic to SuperClassic to eliminate the overhead (10-15%) that the external lock manager adds.  Like you, our databases can be very large (300GB) and have a large number of connection (sometimes 800).  This may be something that you also evaluate.