On 2015-06-24 10:38:38 -0400, Robert Haas wrote:
> On Wed, Jun 24, 2015 at 10:29 AM, Andres Freund <andres@anarazel.de> wrote:
> > On 2015-06-24 23:05:45 +0900, Fujii Masao wrote:
> >> INSERT ON CONFLICT DO UPDATE doesn't seem to work on the current partitioning
> >> mechanism. For example, in the following SQL commands, the last UPSERT command
> >> would fail with an error. The error message is
> >
> > I think that's pretty much inevitable without baking in touple routing
> > into the core system and supporting unique-constraints that span
> > partitions. In other words, I don't think this is upsert's fault.
>
> Is the root of the problem that the trigger is called for an INSERT ..
> ON CONFLICT statement but it turns that into a plain INSERT?
If you'd not do that, you'd avoid errors when violating a unique key
inside a partition, so it'd help a bit.
But it'd not do anything sane when the conflict is actually *not*
aligned with the partitioning schema, because we'll obviously only
search for conflicts within the one table.
Greetings,
Andres Freund