Re: Additional stats for Relations - Mailing list pgsql-hackers

From Jim C. Nasby
Subject Re: Additional stats for Relations
Date
Msg-id 20061013163629.GV28647@nasby.net
Whole thread Raw
In response to Additional stats for Relations  (NikhilS <nikkhils@gmail.com>)
Responses Re: Additional stats for Relations  (NikhilS <nikkhils@gmail.com>)
List pgsql-hackers
On Fri, Oct 13, 2006 at 06:17:47PM +0530, NikhilS wrote:
> Currently a "select * from pg_statio_user_tables;" displays only
> heap_blks_read, heap_blks_hit stats amongst others for the main relation. It
> would be good to have the following stats collected too. I think these stats
> can be used to better statistically analyze/understand the block I/O
> activity on the relation:
> 
> heap_blks_reused: The number of buffers returned by the FSM for use to store
> a new tuple in
The description on this is misleading... FSM doesn't return buffers, it
returns pages that have free space on them.

I'm also not sure if this metric is what you actually want, since a
single page can be returned many times from the FSM even between
vacuums. Tracking how many pages for a relation have been put into the
FSM might be more useful...

> heap_blks_extend: The number of times file extend was invoked on the
> relation
> 
> heap_blks_truncate: The total number of blocks that have been truncated due
> to vacuum activity e.g.
> 
> As an addendum to the truncate stats above, we can also have the additional
> following stats:
> 
> heap_blks_maxtruncate: The max block of buffers truncated in one go
> 
> heap_blks_ntruncate: The number of times truncate was called on this
> relation
> 
> I can come up with a patch (already have one) for the above. Any
> thought/comments?

Do you have a use-case for this info? I can see where it might be neat
to know, but I'm not sure how you'd actually use it in the real world.
-- 
Jim Nasby                                            jim@nasby.net
EnterpriseDB      http://enterprisedb.com      512.569.9461 (cell)


pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: [PERFORM] Hints proposal
Next
From: Stephen Frost
Date:
Subject: Re: [PATCHES] array_accum aggregate