Thanks
I have several "daily" tables that get dropped every day..Is there a
wildcard that I can use to tell vacuumdb NOT to vacuum those
tables...i.e my goal is to ensure that vacuumdb vacuums the entire
database (minus the daily tables that get dropped) so that the XID
wraparound value gets reset correctly once vacuumdb is done. The
daily tables have suffixes like 2009_01_01, 2009_01_02 etc.
On Wed, Oct 14, 2009 at 12:34 AM, Simon Riggs <simon@2ndquadrant.com> wrote:
> On Tue, 2009-10-13 at 19:40 -0600, Scott Marlowe wrote:
>> On Tue, Oct 13, 2009 at 7:29 PM, Anj Adu <fotographs@gmail.com> wrote:
>> > I am running Postgres 8.1.9 on an 8 core Xeon 5430 box that is showing
>> > single digit CPU and IO utilization. the database size is 820G .
>> > Vacuum_cost_delay=0 and maintenance_mem = 900M
>> >
>> > Is there an option to vacuumdb or a way to make it run parallel threads.
>>
>> I think that option showed up in 8.2 or 8.3.
>
> There is no option to do that yet at any release.
>
> We have parallel pg_restore, but that doesn't apply to vacuumdb,
> reindexdb, clusterdb or any custom written jobs.
>
> You need to break out your top few tables into separate jobs, e.g.
> vacuumdb -t big1
> vacuumdb -t big2
> ...
>
> --
> Simon Riggs www.2ndQuadrant.com
>
>