Re: Proposal : REINDEX SCHEMA - Mailing list pgsql-hackers

From Michael Paquier
Subject Re: Proposal : REINDEX SCHEMA
Date
Msg-id CAB7nPqQB-QBVQ+PAKCTmPMUzOyVaM1XgF-VRjiqSGg9gcwJqzQ@mail.gmail.com
Whole thread Raw
In response to Re: Proposal : REINDEX SCHEMA  (Simon Riggs <simon@2ndQuadrant.com>)
Responses Re: Proposal : REINDEX SCHEMA  (Michael Paquier <michael.paquier@gmail.com>)
List pgsql-hackers
On Tue, Dec 9, 2014 at 6:00 PM, Simon Riggs <simon@2ndquadrant.com> wrote:
> On 9 December 2014 at 17:17, Michael Paquier <michael.paquier@gmail.com> wrote:
>
>>> While re-looking at that. I just found that when selecting the
>>> relations that are reindexed for a schema we ignore materialized view
>>> as the key scan is only done using 'r' as relkind. The patch attached
>>> fixes that.
>> Here is an updated patch doing as well that:
>> - Regression test checking if user has permissions on schema was broken
>> - Silent NOTICE messages of REINDEX by having client_min_messages set
>> to WARINING (thoughts about having a plpgsql function doing
>> consistency checks of relfilenode before and after reindex?)
>
> ISTM that REINDEX is not consistent with VACUUM, ANALYZE or CLUSTER in
> the way it issues NOTICE messages.
>
> I'm inclined to simply remove the NOTICE messages, except when a
> REINDEX ... VERBOSE is requested.
OK. Perhaps that's not worth mentioning in the release notes, but some
users may be used to the old behavior. What about the other issues
(regression test for permissions incorrect and matviews)?
-- 
Michael



pgsql-hackers by date:

Previous
From: Kouhei Kaigai
Date:
Subject: Re: [v9.5] Custom Plan API
Next
From: Thomas Reiss
Date:
Subject: Re: Casting issues with domains