From: | jgpedro(at)gmail(dot)com |
---|---|
To: | pgsql-bugs(at)postgresql(dot)org |
Subject: | BUG #12898: Failure loading materialized view with pg_restore |
Date: | 2015-03-25 15:48:12 |
Message-ID: | 20150325154812.2490.65573@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: 12898
Logged by: José Pedro
Email address: jgpedro(at)gmail(dot)com
PostgreSQL version: 9.3.6
Operating system: Ubuntu 14.04.2 LTS 64-bit
Description:
pg_restore of a materialized view fails with "permission denied for relation
<X>".
The same problem as discussed in this thread:
Problem identified as:
- Tables ACLs restored after materialized views creation;
- Owner of materialized view only gets permission to table after the ACLs
being restored.
As Tom Lane said
(http://www.postgresql.org/message-id/11166.1424357659@sss.pgh.pa.us)
"We need to rethink the ordering rules here. I believe that not dumping
ACLs till late is an intentional choice to avoid corner cases with regular
tables (eg what if user has revoked INSERT on a table), but it doesn't
work so well for matviews.
One possible avenue to a fix is to also postpone the assignment of the
matview's owner, but I'm not sure that that's a great idea from a security
standpoint.
A possibly safer idea is just to put all REFRESHes after all ACL updates.
If things fail then, well, they'd have failed anyway."
I'm opening a bug report so this issue won't be forgotten ;)
From | Date | Subject | |
---|---|---|---|
Next Message | Pedro Gimeno | 2015-03-25 16:22:07 | Re: Re: BUG #12885: The result of casting a double to an integer depends on the database version |
Previous Message | Kevin Grittner | 2015-03-25 15:41:10 | Re: "no-slave yet" early CREATE TABLE transaction gets blocked when synchronous replication |