Mario Weilguni <mweilguni(at)sime(dot)com> writes:
> I guess two backends might have issued interleaved updates.
> I think without logging this cannot be solved here.
Yeah, it's annoying that there is no convenient way to see the contents
of pg_locks at the instant of the deadlock. Knowing which tuple locks
were held would give at least some additional information.
regards, tom lane