Oracle 11.2: SERIAL# has changed each 5-10 seconds

Question

Hi all,
On server I see a lot of bad sessions with STATUS = 'ACTIVE' and с EVENT = 'SQL*Net message from client'.
And the most weird is following, see SERIAL# field (changed in time):
select SYSDATE, SADDR, SID, SERIAL#, AUDSID from v$session where sid = 11
     SYSDATE             SADDR                        SID                                  SERIAL#                                AUDSID                
08/10/2010 13:15:11 00000004306C7608                                     11                                   4570                                4493432
      SYSDATE             SADDR                        SID                                  SERIAL#                                AUDSID                
08/10/2010 13:15:26 00000004306C7608                                     11                                   4572                                4493432
      SYSDATE             SADDR                        SID                                  SERIAL#                                AUDSID                
08/10/2010 13:15:46 00000004306C7608                                     11                                   4574                                4493432 And also all of them have:
WAIT_TIME > 0
STATE = WAITED KNOWN TIME
SECONDS_IN_WAIT >0
Could somebody advice me... what is it?
Oracle version:
Oracle Database 11g Enterprise Edition Release 11.2.0.1.0 - 64bit Production
PL/SQL Release 11.2.0.1.0 - Production
CORE     11.2.0.1.0     Production
TNS for Solaris: Version 11.2.0.1.0 - Production
NLSRTL Version 11.2.0.1.0 - Production

Answer 1

I do not think I have ever seen the serial# value change for a session except when it was doing rollback so I cannot think of any other cause for this.
If the session user does have only select privilege and even if the session issued distributed SQL which is a transaction to Oracle which takes an undo/rollback segment entry I would expect the rollback to be nearly instant so you should never see it. Unless maybe the remote db was unavailable and reco was checking every 5 seconds for the remote site. I take it dba_2pc_pending and dba_2pc_neighbors are empty?
Since you are on 11.2 I guess attempting to seach Oracle support for know issues related to this observation would be next on my list on things to try to help me figure out what was going on.
You can also start a trace on the running session to see if the trace file reviews anything of value/use.
HTH -- Mark D Powell --