RE: Long running roll back

From: Powell, Mark D <mark.powell_at_eds.com>
Date: Wed, 15 Oct 2008 09:59:33 -0400
Message-ID: <D1DC33E67722D54A93F05F702C99E2A902FFC92D@usahm208.amer.corp.eds.com>


If there is no entry in v$transaction but the session background process is still getting time kill it from the OS. If the session held any row level locks there would be an entry in v$transaction. If the session background process does not exist then Oracle has just not reused the v$session entry yet.  

  • Mark D Powell -- Phone (313) 592-5148

        From: oracle-l-bounce_at_freelists.org
[mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Ram Raman

	Sent: Tuesday, October 14, 2008 5:59 PM
	To: ORACLE-L
	Subject: Long running roll back
	
	
	Hi all, 
	 
	V: 10.203

	We had a long running process that was running for hours (20 hrs
for one qry) and the CPU usage spiked to 100%. We had to kill it. 
	The status is being shown as Killed in V$session. It has been
that way for the past several hours. Users are wondering when it will finish. Usually in this scenario I look at v$transaction to see the progress of it, but I don't see any.

        v$session:         

	       SID    SERIAL# CLIENT_INFO
	---------- ----------

----------------------------------------------------------------
PROGRAM OSUSER PROCESS ------------------------------------------------ ------------------------------ ------------ USERNAME MACHINE STATUS ------------------------------
----------------------------------------------------------------
-------- TO_CHAR(S.LOGON_T PADDR SADDR SPID PID SQL_ADDRESS ----------------- ---------------- ---------------- ------------ ---------- ---------------- 882 5587 SYSTEM 1234 SUPRKRON abc-dsonos4 KILLED 03-SEP-08 11:03am C00000009EA9C7F8 C00000009D1A8D00 19385
18 C0000000928A0DE0         

        SQL> select * from v$transaction;

        no rows selected

	SQL> 
	SQL> select * from v$sqlarea where address = 'C0000000928A0DE0';

	no rows selected

	  1  select * from v$session_wait
	  2* where sid = 882
	SQL> /

	       SID       SEQ# EVENT
	---------- ----------

----------------------------------------------------------------
P1TEXT P1 P1RAW ---------------------------------------------------------------- ---------- ---------------- P2TEXT P2 P2RAW ---------------------------------------------------------------- ---------- ---------------- P3TEXT P3 P3RAW WAIT_CLASS_ID WAIT_CLASS# ---------------------------------------------------------------- ---------- ---------------- ------------- ----------- WAIT_CLASS WAIT_TIME SECONDS_IN_WAIT STATE ---------------------------------------------------------------- ---------- --------------- ------------------- 882 20028 kksfbc child completion

0 00         

0 00         

0 00                  1893977003           0
	Other
5         2409758 WAITED KNOWN TIME

	 

	I googled on kkssfbc and I got this: 

	

http://www.freelists.org/archives/oracle-l/02-2007/msg00465.html.

        16:23:18 SQL> show parameter cursor

	NAME                                 TYPE        VALUE
	------------------------------------ -----------
------------------------------
	cursor_sharing                       string      EXACT
	cursor_space_for_time                boolean     FALSE
	open_cursors                         integer     1000
	session_cached_cursors               integer     20
	

	Can anyone shed light on what is going on? THe OS process
corresponding to the killed session (882) is using one CPU fully out of 2 CPUs.

        Thanks.

--
http://www.freelists.org/webpage/oracle-l
Received on Wed Oct 15 2008 - 08:59:33 CDT

Original text of this message