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

Frank van Vugt <ftm.van.vugt@foxi.nl> writes:
> Ok, here are the results of the assertion TRAP (mind you, same error on the
> same line, but a different machine than the original)

That's another can't-happen problem.  As best I can tell, this suggests
that the tuple descriptor associated with the RELOID catalog cache has
gotten clobbered (it should have the HASOID bit set, but an Assert that
says it does is failing).

Either you've found some code path that clobbers random bits of memory,
or you've got flaky hardware.  But the same error on two different
machines makes the second theory a bit hard to swallow.

What's the hardware platform exactly?  Is it 64-bit, or anything at all
off the beaten path?

            regards, tom lane