Hi there,
my issue involves these elements:
- Firebird 2.0 SS RC3
- Server single processor Pentium 4 class,
don't know if Intel or AMD, with enough RAM
- Linux RedHat 8 Kernel 2.4
- Old threading library used

The installation with rpm package goes straight without a glitch.
A client connects with the database without a problem. It can
issue any request to the database and it responds accordingly.
Doing no requests for 30 secs / 1 minute, the engine starts
loading the cpu to 100% without no apparent reason.
Can't be the sweeping thread because the transaction number
limit is set a lot higher (the default 20000).
If the client issues any simple request, the engine answers
accordingly then the cpu load goes to zero for 30 secs and then
restarts to grow to 100%. This thing can go on and on, apparently
without apparently causing any harm to the database.

Now a second client connects to the database:
- if the engine is quiet or doing something for the first client
all goes as expected
- if the engine is loading the cpu for no reason the engine crashes
and restarted by the guardian process

I have never seen such a behaviour before, even with the same
version of the engine on another Linux distribution (Fedora Core 5
if I remember correctly).

Installing the Classic variant of the engine, same 2.0RC3 version,
all goes well in any situation.

Does anyone have an explanation for this?

TIA

Mirco Malaguti