Re: CLUSTER and synchronized scans and pg_dump et al

Steve Atkins wrote:
> On Jan 28, 2008, at 8:36 AM, Tom Lane wrote:
>
>> Andrew Dunstan <andrew(at)dunslane(dot)net> writes:
>>> Kevin Grittner wrote:
>>>> It would seem reasonable to me for pg_dump to use ORDER BY to select
>>>> data from clustered tables.
>>
>>> What will be the performance hit from doing that?
>>
>> That worries me too. Also, in general pg_dump's charter is to reproduce
>> the state of the database as best it can, not to "improve" it.
>
> One common use of cluster around here is to act as a faster version
> of vacuum full when there's a lot of dead rows in a table. There's no
> intent to keep the table clustered on that index, and the cluster flag
> isn't removed with alter table (why bother, the only thing it affects is
> the cluster command).
>
> I'm guessing that's not unusual, and it'd lead to sorting tables as part
> of pg_dump.

I've done that too - and every time I typed that "CLUSTER ... " I
thought why, oh why isn't there something like REWRITE TABLE <table>",
which would work just like CLUSTER, but without the sorting ;-) Maybe
something to put on the TODO list...

We might even call it "VACCUM REWRITE" ;-)

regards, Florian Pflug