Re: JDBC HIGH CPU and Restart Issue

["Joe Sam Shirah" <joe_sam@xxxxxxxxxxxxx>]

   Hi Paul,

   The first thing I would look at is the actual queries running when you
see the problem.  The QZDASOINT jobs don't have to do particularly with
Java; they are also used for ODBC database jobs, for example.  In the days
before Java was even around, I saw a user manage to practically bring down
the machine and come dangerously close to filling the disk with a cartesian 
join.

   The only part of Pete's response (which seemed related to Tomcat running
*on* the AS/400) in your case that might help is looking at the run priority
and timeslice of the QZDASOINT jobs, but that would only be so they play
nice with other system users/jobs.  I don't think it would address your base
problem.

   I expect the jobs get restarted and continue because the connection pool
code detects that the connection failed (when you end the QZDASOINT jobs)
and just grabs another connection (which starts another QZDASOINT job) to
retry the query.

   That's probably the best we can suggest without starting detailed
analysis, but let us know if the queries are the problem or you need to get
deper into it.

   HTH,


                                                        Joe Sam

Joe Sam Shirah -        http://www.conceptgo.com
conceptGO       -        Consulting/Development/Outsourcing
Java Filter Forum:       http://www.ibm.com/developerworks/java/
Just the JDBC FAQs: http://www.jguru.com/faq/JDBC
Going International?    http://www.jguru.com/faq/I18N
Que Java400?            http://www.jguru.com/faq/Java400

----- Original Message ----- 
From: <pholm@xxxxxxxxxxxxxxxxx>
To: <java400-l@xxxxxxxxxxxx>
Sent: Thursday, May 06, 2010 5:16 PM
Subject: JDBC HIGH CPU and Restart Issue


>   All,
>   I've found some related info on google but nothing definitive.
>   Environment:  V5R4 iSeries acting as database with Tomcat and
>   JDBC/connection pools hosted on a Windows server.
>   Problem:  Occasionally the QZDASOINT job servicing the JDBC will consume
>   most CPU.  I've lowered the QZDASOINIT priority to 30 to lessen the
> impact
>   to interactive users.  However the problem and high cpu persist
>   occasionally.  If we do a ENDJOB on these jobs they are restarted
>   automatically and resume using too much CPU.  If we end Tomcat the jobs
> go
>   away.  Restarting Tomcat restarts the QZDASOINIT jobs.  We have
> connection
>   pool code but we do not have any logic to reissue JDBC queries to the
>   system.  I suspect it to be an i5 issue.  Has anyone experienced the
> issue
>   or have insight?
>   Thanks, Paul Holm
> --