BUG #12832: pg_dump ERROR with temporary functions

From: marko(at)joh(dot)to
To: pgsql-bugs(at)postgresql(dot)org
Subject: BUG #12832: pg_dump ERROR with temporary functions
Date: 2015-03-05 10:03:43
Message-ID: 20150305100343.2527.9521@wrigleys.postgresql.org
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-bugs

The following bug has been logged on the website:

Bug reference: 12832
Logged by: Marko Tiikkaja
Email address: marko(at)joh(dot)to
PostgreSQL version: 9.3.6
Operating system: Linux
Description:

Hi,

We've occasionally been having a problem where a script does roughly this:

psql -d dbname -c "CREATE TEMPORARY FUNCTION pg_temp.foo() AS ..; <run a
query using the temporary function a number of times>"
pg_dump -d dbname

And the pg_dump call ERRORs out like this:

pg_dump: [archiver (db)] query failed: ERROR: cache lookup failed for
function 2540046
pg_dump: [archiver (db)] query was: SELECT tableoid, oid, proname, prolang,
pronargs, proargtypes, prorettype, proacl, pronamespace,
pg_catalog.pg_get_function_identity_arguments(oid) AS proiargs,(SELECT
rolname FROM pg_catalog.pg_roles WHERE oid = proowner) AS rolname FROM
pg_proc p WHERE NOT proisagg AND (pronamespace != (SELECT oid FROM
pg_namespace WHERE nspname = 'pg_catalog')
AND NOT EXISTS (SELECT 1 FROM pg_depend WHERE classid =
'pg_proc'::regclass AND objid = p.oid AND deptype = 'i'))

The problem seems to be that the snapshot of pg_proc the pg_dump session has
sees the function, but by the time it gets to run
pg_get_function_identity_arguments() on it, it's been cleaned up by the
backend which created it. Now this is an issue with normal functions as
well if they're dropped concurrently with a pg_dump run, but running into
problems like this when doing everything sequentially seems more astonishing
to my little brain.

Any thoughts on adding NOT pg_is_other_temp_schema(pg_proc.pronamespace)
quals to the queries which look at pg_proc to counter this problem?
Obviously this can be worked around in client code with a strategically
placed DISCARD TEMP, but in general this seems like a POLA violation.

Responses

Browse pgsql-bugs by date

  From Date Subject
Next Message Elena Fernandez Carmona 2015-03-05 10:27:10 compatibilty postgres 9.2 RHEL 6.4
Previous Message pgriboval 2015-03-05 08:41:09 BUG #12830: postgres-9.4 upgrade is missing PGPORT