8 mins of unknown activitiy each full hour

From: Hannes Erven <h.e_at_gmx.at>
Date: Mon, 14 Feb 2005 18:17:18 +0100
Message-ID: <4210dd4e@e-post.inode.at>

Hi all,

I managed to set up 10g on a Gentoo Linux box, and it runs fine -- except that at each full hour, the oracle process uses 100% CPU for about 6-8 minutes -- with NO users connected. The machine is a 3 GHz P4 exclusivly dedicated to testing oracle, so system capacity shouldn't be an issue.

The Top Sessions list displays:
SYSMAN Oracle Enterprise Manager.rollup

Top Modules:

        SYS$USERS Oracle Enterprise Manager.rollup 80.8

Module Oracle Enterprise Manager.rollup

	target 3	75.5
	target 2	17.5	
	target 1	7.1

The TOP SQL page identifies this as the current SQL:

DECLARE

     job BINARY_INTEGER := :job;
     next_date DATE := :mydate;
     broken BOOLEAN := FALSE;
BEGIN
     EMD_MAINTENANCE.EXECUTE_EM_DBMS_JOB_PROCS();


     :mydate := next_date;

     IF broken THEN :b := 1;
         ELSE


     :b := 0;

     END IF;

END; The Top Sessions page identifies one DEDICATED session from user SYSMAN as source of these commands. Every hour, this session (the *same* session, same SID) executes the same commands which lead to the unusual system load. Puzzingly the EM says "Session expired"

I already tried switching off automatic worload repository snapshots, but that didn't help -- according to the logs, the snapshots are done in a couple of seconds anyways.

There are no errors or alerts displayed in the EM console, the only weird thing I noticed is that it says "listener: not available" and "host: not available", altough the DB is up and connections from sqlplus and net8 are accepted.

I've run out of ideas where to search -- thank you in advance for your help!

best regards,

        -hannes Received on Mon Feb 14 2005 - 11:17:18 CST