Re: Bug #769: Slow vacuuming due to error in optimization - Mailing list pgsql-bugs

From Tom Lane
Subject Re: Bug #769: Slow vacuuming due to error in optimization
Date
Msg-id 2053.1032210284@sss.pgh.pa.us
Whole thread Raw
In response to Bug #769: Slow vacuuming due to error in optimization  (pgsql-bugs@postgresql.org)
List pgsql-bugs
Stephen Marshall <smarshall@wsicorp.com> writes:
> Tom Lane wrote:
>> How large is "large", and what FSM parameters are you using?  Do you
>> know how many pages were getting passed into MultiRecordFreeSpace?

> vacuum_mem is 16384
> max_fsm_relations is 100
> max_fsm_pages is 300000

> pg_largeobject data size =  4211617.693 K
> pg_largeobject file size  =  7932208.000 K
> number of large objects =  870
> number of lo pages        =  2106287
> number of messages      =    469458

> Upon vacuuming, I found 300000 pages were passed to MultiRecordFreeSpace,
> with the minPage specified as 0 and the maxPage as -2.  The pages passed
> exactly equaled the max_fsm_pages parameter.

Yeah; if you look at vacuumlazy.c you'll see it doesn't bother keeping
track of more than max_fsm_pages, since it knows the FSM will not
remember more than that either.

> This vacuum took over 30 minutes, with more than 25 minutes spent in
> MultiRecordFreeSpace.

Yow.  Definitely need to fix that code ...

            regards, tom lane

pgsql-bugs by date:

Previous
From: Tom Lane
Date:
Subject: Re: Bug #769: Slow vacuuming due to error in optimization
Next
From: David Wheeler
Date:
Subject: 7.3b1 on Mac OS X (Darwin)