Re: 8.4 -> 9.1 : ANALYSE VERBOSE; -> out of shared memory - Mailing list pgsql-general

From David G Johnston
Subject Re: 8.4 -> 9.1 : ANALYSE VERBOSE; -> out of shared memory
Date
Msg-id 1399909847066-5803669.post@n5.nabble.com
Whole thread Raw
In response to Re: 8.4 -> 9.1 : ANALYSE VERBOSE; -> out of shared memory  (Alvaro Herrera <alvherre@2ndquadrant.com>)
List pgsql-general
Alvaro Herrera-9 wrote
> Souquieres Adam wrote:
>
>> The solution must be to define a *better strategy for tables
>> statistics*... we dont really need stats on all the tables because
>> the most part of them is static...
>
> Tune autovacuum so that it does the analyses for you?

And for those few tables that need special handling, use:

ANALYZE [table name];

and/or

VACUUM ANALYZE [table name];

The observation that you "don't need statistic [updates] on all tables
[every time]" is accurate but you decided to implement a custom solution
without taking that into account...

Maybe the manual versions of these routines could be coded to evaluate usage
statistics and skip any tables that are not in need of updating - but then
you'd need some way to control thresholds and force it to perform regardless
of what it thinks.  Considerable effort when much of that code is already
embedded into autovacuum.  Not saying the existing system is perfect but it
is what currently exists and works well in the majority of situations - the
remainder of which a logic-less VACUUM/ANALYZE handles adequately.

David J.




--
View this message in context:
http://postgresql.1045698.n5.nabble.com/8-4-9-1-ANALYSE-VERBOSE-out-of-shared-memory-tp5803630p5803669.html
Sent from the PostgreSQL - general mailing list archive at Nabble.com.


pgsql-general by date:

Previous
From: Alvaro Herrera
Date:
Subject: Re: 8.4 -> 9.1 : ANALYSE VERBOSE; -> out of shared memory
Next
From: Borodin Vladimir
Date:
Subject: Question about synchronous replication