On Thu, Mar 10, 2022 at 10:56 AM Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Each cascaded delete ought to be removing a disjoint set of rows in the
> referencing
table, so I'm not quite sure why order should matter.
>
> regards, tom lane
>
I have always thought the way to avoid deadlocks was to update rows in the
same order by the different updaters. Is there a better chain of thought
for updating and deleting rows at the same time? Do we need to put a lock
on the table to update, then have the delete queue up waiting for the lock
to be removed?
Thanks,
George