Re: join help

From: Kashmir <kashmir_us_1999(at)yahoo(dot)com>
To: Justin <justin(at)emproshunts(dot)com>, Helio Campos Mello de Andrade <helio(dot)campos(at)gmail(dot)com>
Cc: pgsql-sql(at)postgresql(dot)org
Subject: Re: join help
Date: 2009-04-09 03:28:40
Message-ID: 35191.78889.qm@web31913.mail.mud.yahoo.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-sql

thx for the help!
this did it then for an individual collection and its fast :-):

SELECT
OneM.f_timestamp,
FiveM.f_ds,
FiveM.f_us,
OneM.f_ds,
OneM.f_us
FROM td_fetch1m_by_rrd_id OneM
left join
(select f_rrd_id, f_ds, f_us, f_timestamp from td_fetch_by_rrd_id where f_rrd_id = 444) FiveM
ON (OneM.f_timestamp = FiveM.f_timestamp)
where OneM.f_rrd_id = 444
ORDER BY OneM.f_timestamp;

________________________________
From: Justin <justin(at)emproshunts(dot)com>
To: Kashmir <kashmir_us_1999(at)yahoo(dot)com>
Cc: pgsql-sql(at)postgresql(dot)org
Sent: Wednesday, April 8, 2009 9:38:22 PM
Subject: Re: [SQL] join help

Kashmir wrote:only difference is:

first table stores data per 'f_rrd_id' evey 5min, and the second table every single minute.
I
want to run a query that would return for the same 'f_rrd_id' all
values from both tables sorted by f_timestamp, of course a set would
only have values from the 5m table if the timestamp was present there
too (every 5th set only)

being a sql-lamer, i used some query
builder help to build my query (which served me quite well in the past
for all my 'complicated' sqls), and was suggested for f_rrd_id=444 to
use something as:
SELECT
td_fetch1m_by_rrd_id.f_timestamp,
td_fetch_by_rrd_id.f_ds,
td_fetch_by_rrd_id.f_ds,
td_fetch1m_by_rrd_id.f_ds,
td_fetch1m_by_rrd_id.f_us
FROM td_fetch_by_rrd_id
RIGHT JOIN td_fetch1m_by_rrd_id ON td_fetch_by_rrd_id.f_timestamp=td_fetch1m_by_rrd_id.f_timestamp
WHERE td_fetch1m_by_rrd_id.f_rrd_id=444
ORDER BY td_fetch1m_by_rrd_id.f_timestamp;

and this works quite fine and as expected in the source env (some gui-sqler).
but when i take this into psql, i get totally messed up results, the values just dont make any sense...

The sql is joining on a time stamp?? Using the time stamp i would expect odd ball results because a several unique f_rr_id could have the same timestamp especially if its heavy write table .

every 5th set only ???? What does this mean what makes something the 5th set.

I normally avoid table aliasing but these names i'm having a real tough time reading so we are going to use 1Minute = td_fetch1m_by_rrd_id and the 5Minute = td_fetch_by_rrd_id from here on out.

You want to join whats in the 1Minute table to whats in the 5Minute only if it is in the 5Minute table and only return from 1Minute table where the timestamps is in the 5Minute table If my understanding is correct this will work minus any typos. To create a join condition we need a composite identity to join on. So what i did is cast F_rr_id and F_timestamp to text adding them together to create a unique condition to join on.

Also there is a typo above noted in bold f_ds is listed twice i believe that is a mistake.

SELECT
OneM.f_timestamp,
FiveM.f_ds,
FiveM.f_us,
OneM.f_ds,
OneM.f_us
FROM td_fetch1m_by_rrd_id OneM,
left Join (select f_rrd_id, f_ds, f_us, f_timestamp
from td_fetch_by_rrd_id ) FiveM
ON (OneM.f_rrd_id::text || OneM.f_timestamp::text) = (FiveM.f_rrd_id::text || FiveM.f_timestamp::text)
ORDER BY OneM.f_timestamp;

In response to

Browse pgsql-sql by date

  From Date Subject
Next Message Steve Midgley 2009-04-09 15:13:45 Re: Nested selects
Previous Message Justin 2009-04-09 01:38:22 Re: join help