Re: [HACKERS] Proposal: Local indexes for partitioned table

Hi Alvaro,

On 01/18/2018 04:43 PM, Alvaro Herrera wrote:
> Here's a patch to add pg_dump tests.  This verifies that the
> CREATE INDEX on parent and partition appear, as well as ALTER INDEX ..
> ATTACH PARTITION.
> 
> While doing this, I noticed a small bug: the ALTER INDEX would not be
> dumped when only one of the schemas are specified to pg_dump (schema of
> parent and schema of partition), but it would be if both were specified.
> This patch fixes that problem too.  AFAICS the test is correct after
> that fix (ie. the "like" and "unlike" sets are correct now.)
> 

I get

pg_restore: creating INDEX "pg_catalog.pg_authid_oid_index"
pg_restore: [archiver (db)] Error while PROCESSING TOC:
pg_restore: [archiver (db)] Error from TOC entry 5493; 1259 2677 INDEX 
pg_authid_oid_index jpedersen
pg_restore: [archiver (db)] could not execute query: ERROR:  permission 
denied: "pg_authid" is a system catalog
     Command was:
-- For binary upgrade, must preserve pg_class oids
SELECT 
pg_catalog.binary_upgrade_set_next_index_pg_class_oid('2677'::pg_catalog.oid);

CREATE UNIQUE INDEX "pg_authid_oid_index" ON "pg_authid" USING "btree" 
("oid");

during check-world.

Best regards,
  Jesper