Обсуждение: Query runs on 9.2, but not on 9.4

Hi all,

I've got a query I use to show the amount of shared buffers in use that is giving me some issues. It uses the pg_buffercache extension. The query is:

SELECT

c.relname,

pg_size_pretty(count(*) * 8192) as buffered,

round(100.0 * count(*) /

(SELECT setting FROM pg_settings

WHERE name='shared_buffers')::integer,1)

AS buffers_percent,

round(100.0 * count(*) * 8192 /

pg_relation_size(c.oid),1)

AS percent_of_relation

FROM pg_class c

INNER JOIN pg_buffercache b

ON b.relfilenode = c.relfilenode

INNER JOIN pg_database d

ON (b.reldatabase = d.oid AND d.datname = current_database())

GROUP BY c.oid,c.relname

ORDER BY 3 DESC

LIMIT 25;

On a 9.2 system, it runs fine and produces a list of the top 25 tables using the shared buffer, but on 9.3 and above, it throw a "division by zero" error. I've run explain for the query on both the 9.2 and the 9.4 server, and the only difference I can see is the 9.4 server shows an additional index scan that the 9.2 server does not. Here's the explain from the 9.4 server:

QUERY PLAN

-----------------------------------------------------------------------------------------------------------------------------

 Limit  (cost=59.73..59.74 rows=5 width=68)

   InitPlan 1 (returns $0)

     ->  Function Scan on pg_show_all_settings a  (cost=0.00..12.50 rows=5 width=32)

           Filter: (name = 'shared_buffers'::text)

   ->  Sort  (cost=47.22..47.24 rows=5 width=68)

         Sort Key: (round(((100.0 * (count(*))::numeric) / (($0)::integer)::numeric), 1))

         ->  HashAggregate  (cost=46.91..47.16 rows=5 width=68)

               Group Key: c.oid, c.relname

               ->  Nested Loop  (cost=1.30..46.85 rows=5 width=68)

                     ->  Hash Join  (cost=1.03..14.83 rows=5 width=4)

                           Hash Cond: (p.reldatabase = d.oid)

                           ->  Function Scan on pg_buffercache_pages p  (cost=0.00..10.00 rows=1000 width=8)

                           ->  Hash  (cost=1.01..1.01 rows=1 width=4)

                                 ->  Seq Scan on pg_database d  (cost=0.00..1.01 rows=1 width=4)

                                       Filter: (datname = current_database())

                     ->  Index Scan using pg_class_tblspc_relfilenode_index on pg_class c  (cost=0.28..6.39 rows=1 width=72)

                           Index Cond: (relfilenode = p.relfilenode)

(17 rows)

The explain for the 9.2 server is the same, except as noted, the 9.4 has that additional Index Scan at the very end. I've tried contacting the original author on evol-monkey.blogspot.com, but the authentication there is broken, and won't let me comment on his page.

Anybody have any ideas for why this breaks and what I can do to fix it? I've been looking at for a day or so with no success.

--

Jay

On Aug 5, 2015, at 7:56 AM, John Scalia <jayknowsunix@gmail.com> wrote:
>
> SELECT
> c.relname,
> pg_size_pretty(count(*) * 8192) as buffered,
> round(100.0 * count(*) /
> (SELECT setting FROM pg_settings
> WHERE name='shared_buffers')::integer,1)
> AS buffers_percent,
> round(100.0 * count(*) * 8192 /
> pg_relation_size(c.oid),1)
> AS percent_of_relation
> FROM pg_class c
> INNER JOIN pg_buffercache b
> ON b.relfilenode = c.relfilenode
> INNER JOIN pg_database d
> ON (b.reldatabase = d.oid AND d.datname = current_database())
> GROUP BY c.oid,c.relname
> ORDER BY 3 DESC
> LIMIT 25;

So, the first thing I would do is change:

> round(100.0 * count(*) /
> (SELECT setting FROM pg_settings
> WHERE name='shared_buffers')::integer,1)

to:

> (SELECT setting FROM pg_settings
> WHERE name='shared_buffers')::integer,1)

and change:

> round(100.0 * count(*) * 8192 /
> pg_relation_size(c.oid),1)

to:

> pg_relation_size(c.oid),1)

Then look for 0s in those output columns. I'll bet that the 9.4 vs 9.2 difference is simply a value of 0 for one of those, not some exotic thing about the query plan. If I'm right about that, then you can start hunting down a specific explanation.

--
Scott Ribe
scott_ribe@elevated-dev.com
http://www.elevated-dev.com/
https://www.linkedin.com/in/scottribe/
(303) 722-0567 voice