In checking things out with CLOBBER_CACHE_ALWAYS, I was getting
this problem, which seems to be unrelated to my changes:
*** /home/kgrittn/pg/master/src/test/regress/expected/alter_table.out 2013-11-01 09:07:35.418829105 -0500
--- /home/kgrittn/pg/master/src/test/regress/results/alter_table.out 2013-11-06 11:06:29.785830560 -0600
***************
*** 2320,2325 ****
) mapped;
incorrectly_mapped | have_mappings
--------------------+---------------
! 0 | t
(1 row)
--- 2320,2325 ----
) mapped;
incorrectly_mapped | have_mappings
--------------------+---------------
! 1 | t
(1 row)
======================================================================
I looked for detail with this query:
SELECT
mapped_oid, oid
FROM (
SELECT
oid, reltablespace, relfilenode, relname,
pg_filenode_relation(reltablespace, pg_relation_filenode(oid)) mapped_oid
FROM pg_class
WHERE relkind IN ('r', 'i', 'S', 't', 'm')
) mapped
WHERE (mapped_oid != oid OR mapped_oid IS NULL);
... with this result:
mapped_oid | oid
------------+------
2139062143 | 2619
(1 row)
2619 is the oid for pg_statistic, and the mapped_oid value matches
what we use to clobber the cache. Any ideas before I start
digging?
--
Kevin Grittner
EDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company
