Looks like the behavior has changed with 11g (both R1 and R2 - probably a bug!) - see the following:
10.2.0.4: ====== 1. After initial logon: SQL> Select timestamp, action_name, logoff_time from DBA_AUDIT_SESSION; TIMESTAMP ACTION_NAME LOGOFF_TIME ------------------ ---------------------------- ------------------ 30-DEC-10 21:26:33 LOGON 2. After log off: SQL> / TIMESTAMP ACTION_NAME LOGOFF_TIME ------------------ ---------------------------- ------------------ 30-DEC-10 21:26:33 LOGOFF *30-DEC-10 21:29:49*As you will notice, it is the desired behavior - we have the the correct/expected log-off-time:
With 11g (both R1 and R2): =================== 1. After initial logon: SQL> Select timestamp, action_name, logoff_time from DBA_AUDIT_SESSION; TIMESTAMP ACTION_NAME LOGOFF_TIME ------------------ ---------------------------- ------------------ 30-DEC-10 21:32:03 LOGON 2. After log-off: TIMESTAMP ACTION_NAME LOGOFF_TIME ------------------ ---------------------------- ------------------30-DEC-10 21:33:24 LOGOFF 30-DEC-10 21:33:24 => same timestamps
On 12/30/2009 4:13 PM, Yong Huang wrote:
Is there any consistency of the audit rows where TIMESTAMP = LOGOFF_TIME?I remember reading on MOS that said what Dave said earlier "if the session ends properly, TIMESTAMP and LOGOFF_TIME should be the same." At logoff time, the timestamp is updated to that time, unless the action is 'LOGOFF BY CLEANUP', i.e. the session terminates abnormally. I didn't find that note, but found Note:271615.1 that says "From 10g onwards the time shown in TIMESTAMP and LOGOFF_TIME columns are in different time zones". I thought that would create a difference in query result. But no. TIMESTAMP of the dba_audit_* views in 10g and up is already converted to the local timezone, even though the ntimestamp# column of aud$ is in UTC. Ric, you're talking about the columns of dba_audit_*, not aud$. What you see is normal. Yong Huang -- //www.freelists.org/webpage/oracle-l
