On Sun, Sep 5, 2010 at 5:09 AM, Carlos Henrique Reimer
<carlos.reimer@opendb.com.br> wrote:
> Hi Alban,
>
> The need for the vacuum full is because there were a problem with the daily
> schedulled vacuum analyze and autovacuum regarding the max_fsm_pages. As it
> was underestimated the vacuum process was not able to flag the pages to be
> reused.
>
> I've cancelled the vacuum full and will think another approach. Maybe a
> CLUSTER can do the work. Will start a CLUSTER and see if I can check the
> progress looking the size of the new table relfilenode. It will probably
> have less than 102 GB.
fastest way if you can afford downtime is something like;
select * into new_table from old_table order by pkcol;
alter old_table rename to old_table_bak;
alter new_table rename to old_table;
--
To understand recursion, one must first understand recursion.