"Per Jensen" <per2jensen@gmail.com> writes:
> When the same dump is restored to a PG-8.2beta1 some queries succeed
> and others fail. The server *restarts* as can be seen from the server
> logfile below.
Can you provide a stack trace from the core dump, or alternatively a
self-contained test case with which someone else can reproduce the
problem?
Does the crash still occur if you disable query logging? My guess
is that this is a bug in either logging or planning of parameterized
queries, but there's not enough info here to locate it.
regards, tom lane