Oracle-L: RE: Hanging query puzzle

From: Thomas Jeff <ThomasJe_at_tce.com>
Date: Wed, 05 Feb 2003 12:09:38 -0800
Message-ID: <F001.00544D1B.20030205120938@fatcity.com>

Sorry, running 8.1.7.2 on AIX 4.3.3.

-----Original Message-----
Sent: Wednesday, February 05, 2003 2:09 PM To: Multiple recipients of list ORACLE-L

What version are you running - I have a similar problem on 9.2.0.2 on Solaris 9 that I've had an open tar on since November - Support has finally called up and said other people are having the same kind of problem - especially in regards to parallel processes. I am supposedly getting a test scenario from them to check out. I'll post the results if and when I get any.

>>> ThomasJe_at_tce.com 02/05/03 11:59AM >>>

We have a query from a 3rd-party tool that seems to either run very quick or crawls to a complete stop.

We can find no patterns to this behavior. The hang can be experienced even when there are no other
processes active in the database. Checking waits, we see a db file scattered read.

SID EVENT                        P1TEXT                     P1 P2TEXT
P2 P3TEXT                     P3

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

    1 pmon timer                   duration                  300
0                             0

   12 slave wait                   msg ptr            5.0440E+17
0                             0

   13 slave wait                   msg ptr            5.0440E+17
0                             0

   14 slave wait                   msg ptr            5.0440E+17
0                             0

   15 slave wait                   msg ptr            5.0440E+17
0                             0

   28 db file scattered read       file#                      12 block#
21047 blocks                      2

    5 smon timer                   sleep time                300 failed
0                             0

Then going to v$sess_io, we see the process is comletely stuck, no activity going on at all, and it's the
only active process in the database.

SID BLOCK_GETS CONSISTENT_GETS PHYSICAL_READS BLOCK_CHANGES CONSISTENT_CHANGES
----- ---------- --------------- -------------- -------------

   28    6233582        60812023       36589516       4076353
115

The query looks like this:

SELECT PRAssignment.*, SRM_RESOURCES.RESOURCE_TYPE FROM PRAssignment, SRM_RESOURCES
WHERE prModTime > TIMESTAMP '2003-02-05 09:23:56.0' AND PRAssignment.prResourceID=SRM_RESOURCES.ID

If I check the file/block values for the wait I get the prassignment table. Prassignment has 5K rows
while srm_resources has 300 rows. Prassignment also has a LONG RAW column, consequently we
see a high chain count, with the result that it's taking up 135 extents to cover those 5K rows.

I'm at a loss to explain why we see such inconsistent results with this query. Thoughts?

Thanks.

Jeffery D Thomas
DBA
Thomson Information Services
Thomson, Inc.

Email: jeff.thomas_at_thomson.net

Indy DBA Master Documentation available at: http://gkmqp.tce.com/tis_dba <http://gkmqp.tce.com/tis_dba> Select 'Indy DBA' then 'DBA Web Pages'

-- 
Please see the official ORACLE-L FAQ: http://www.orafaq.net
-- 
Author: Thomas Jeff
  INET: ThomasJe_at_tce.com

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).

Received on Wed Feb 05 2003 - 14:09:38 CST