Re: "invalid memory alloc request size <n>" in deferred trigger causes transaction to fail, but the backend keeps running

Alvaro Herrera <alvherre(at)dcc(dot)uchile(dot)cl> writes:
> On Fri, Dec 03, 2004 at 01:03:43AM +0100, Frank van Vugt wrote:
>> As far as the other problem is concerned, I just finished the first half of
>> the conversion on a different machine that was build with assertions enabled
>> and saw a number of reports like the ones below. What's causing these?

> This happens when a transaction finishes that creates, drops, or
> modifies a user. If it happens at a different time, something is wrong.

Given that the call site is SIInsertDataEntry, it's clear that
SendPostmasterSignal is actually being told PMSIGNAL_WAKEN_CHILDREN ...
the gdb traceback is reporting the wrong value for the parameter, which
is no big surprise for values in registers.

Which makes this a perfectly normal event; it's just trying to flush the
SI messaging queue before it overflows. Nothing to see here, move
along.

regards, tom lane