Re: Huge memory consumption on partitioned table with FKs

Hi Hackers,

>> I would embed all this knowledge in ri_BuildQueryKey though, without
>> adding the new function ri_GetParentConstOid.  I don't think that
>> function meaningful abstraction value, and instead it would make what I
>> suggest more difficult.

> It seems to me reasonable.

Indeed, I tried to get the conparentid with ri_BuildQueryKey.
(v2_reduce_ri_SPI-plan-hash.patch)

> Somewhat of a detour, but in reviewing the patch for
> Statement-Level RI checks, Andres and I observed that SPI
> made for a large portion of the RI overhead.

If this can be resolved by reviewing the Statement-Level RI checks,
then I think it would be better.

> BTW, one problem with Kuroda-san's patch is that it's using
> conparentid as the shared key, which can still lead to multiple
> queries being generated when using multi-level partitioning, because
> there would be many (intermediate) parent constraints in that case.

Horiguchi-san also mentiond,
In my patch, in ri_GetParentConstOid, iterate on getting
the constraint OID until comparentid == InvalidOid.
This should allow to get the "root constraint OID".

However, the names "comparentid" and "ri_GetParentConstOid"
are confusing. We should use names like "constraint_root_id",
as in Amit-san's patch.

Best Regards,
-- 
Keisuke Kuroda
NTT Software Innovation Center
keisuke.kuroda.3862@gmail.com