Thanks Kevin. That provides a lot of clarity. Cheers.
On 26/06/15 20:40, Kevin Grittner wrote:
> Eoin O'Hehir <eoin.ohehir@clavisinsight.com> wrote:
>
>> I'm setting up a weekly DB maintenance job for a 9.3 server.
>>
>> Part of this will be a VACUUM FULL.
> That's something to seriously reconsider.
>
>> Is there any point in adding the ANALYZE option to the command ?
> That and a non-FULL VACUUM command after VACUUM FULL. Unlike what
> you might expect based on the FULL option, a VACUUM FULL does not
> do everything that a non-FULL VACUUM does, and actually drops any
> existing visibility map and free space map -- potentially making
> subsequent queries against the table slower until a normal VACUUM
> (or autovacuum occurs). I strongly recommend that if you really do
> need to use VACUUM FULL, you immediately follow it with VACUUM
> ANALYZE.
>
> You might also want to read this:
>
> http://rhaas.blogspot.com/2014/03/vacuum-full-doesnt-mean-vacuum-but.html
>
> --
> Kevin Grittner
> EDB: http://www.enterprisedb.com
> The Enterprise PostgreSQL Company
