At 08:55 AM 4/11/2004 -0500, you wrote:

>:: So, you're back to looking at conditions known to cause
>:: crashes: bad UDFs, inflating the TSB beyond the capacity of
>:: RAM, external forces (bad hardware, etc.) - or another theory. :-)
>
>Well I don't have UDFs, and hardware is fine. So I suspect the TSB... BUT if
>the TSB gets too big, it should still just kill the user - not let the WHOLE
>server die. Im sorry but its EXTREMELY poor that a single connection can
>easily kill the WHOLE server just by loading up a large transaction.

No, it's not a single user or a large transaction. It's one or many users
with long-running read-write transactions that never get committeed.

>We had
>this problem earlier when a transaction ran amuck. This is a "Security" flaw
>even. Give me any FB server and watch me crash it in just a minute...

You could look at Paul Beach's message in firebird-devel - follow up on it
and participate in finding the cause....

./helen