KuldeepReddy
asked on
Oracle session termination
Hello experts,
I need your help to fix one issue. I’m using SQL*plus 10.2.0.1.0 as oracle client and Oracle 10.2.0.3.0 as a database. In our organization, we have a separate LINUX box (AEDB01) for batch jobs and which will connect database using TNS name.
I have a one process which invokes one procedure. Basically procedure contains 3 MERGE statements to insert/update data in 3 target tables based on staging table. We have seen a couple of instances of the process running from AEDB01 have run for a long time due to the size of the uploads and, after a while, Oracle would report the session as TERMINATED. There was no indication in the job logs or in the database logs (alert.log) that this is a termination instigated by the database such as a transaction rollback or some error in the code. The database session was eventually cleaned up by Oracle but remained on AEDB01 as a shell script, so I had to clean that up manually.
FYI, I had invoked job using NOHUP command in AEDB01 box even though the oracle session will terminate after a while.
I’m suspecting that this is caused by some network timeout parameter which is killing the session after a period of time. I would like to know what parameter do I need to setup to in order to session alive until job completes.
Any help greatly appreciated.
I need your help to fix one issue. I’m using SQL*plus 10.2.0.1.0 as oracle client and Oracle 10.2.0.3.0 as a database. In our organization, we have a separate LINUX box (AEDB01) for batch jobs and which will connect database using TNS name.
I have a one process which invokes one procedure. Basically procedure contains 3 MERGE statements to insert/update data in 3 target tables based on staging table. We have seen a couple of instances of the process running from AEDB01 have run for a long time due to the size of the uploads and, after a while, Oracle would report the session as TERMINATED. There was no indication in the job logs or in the database logs (alert.log) that this is a termination instigated by the database such as a transaction rollback or some error in the code. The database session was eventually cleaned up by Oracle but remained on AEDB01 as a shell script, so I had to clean that up manually.
FYI, I had invoked job using NOHUP command in AEDB01 box even though the oracle session will terminate after a while.
I’m suspecting that this is caused by some network timeout parameter which is killing the session after a period of time. I would like to know what parameter do I need to setup to in order to session alive until job completes.
Any help greatly appreciated.
You are using NOHUP, so possibly the listener is asked to send probes to the client, but it doesn't answer. thisl eads to closing of the session. This is controlled via SQLNET.EXPIRED_TIME parameter in SQLNET.ORA file on the server. If this parameter is not present ignore my posting.
There is no reason Oracle will kill an active session. If the addition of the SQLNET.EXPIRED_TIME parameter (above comment) has no affect on the problem. Then enable client trace in the client side sqlnet.ora and server trace in the server sqlnet.ora file. Connect using the problem application to produce the error condition.
To enable client side trace:
Add the following to the client sqlnet.ora file
Trace_level_client=16
Trace_directory_client=<pa
Trace_unique_client=on
Trace_timestamp_client=on
Diag_adr_enabled=off
Upload the resulting trace (.trc) file or files generated in the trace directory. Remember to remove the trace parameters from the sqlnet.ora file after capturing the event.
To enable Server Trace:
add the following to the sqlnet.ora file on the server
NOTE: only enable server trace during periods of low connect activity on the server as all new connections will be traced while trace is enabled. Existing connections are not affected.
trace_level_server=16
trace_directory_server=<O_
trace_timestamp_server=on
Diag_adr_enabled=off
Upload the resulting trace (.trc) file or files generated in the trace directory. Remember to remove the trace parameters from the sqlnet.ora file after capturing the event.
To enable client side trace:
Add the following to the client sqlnet.ora file
Trace_level_client=16
Trace_directory_client=<pa
Trace_unique_client=on
Trace_timestamp_client=on
Diag_adr_enabled=off
Upload the resulting trace (.trc) file or files generated in the trace directory. Remember to remove the trace parameters from the sqlnet.ora file after capturing the event.
To enable Server Trace:
add the following to the sqlnet.ora file on the server
NOTE: only enable server trace during periods of low connect activity on the server as all new connections will be traced while trace is enabled. Existing connections are not affected.
trace_level_server=16
trace_directory_server=<O_
trace_timestamp_server=on
Diag_adr_enabled=off
Upload the resulting trace (.trc) file or files generated in the trace directory. Remember to remove the trace parameters from the sqlnet.ora file after capturing the event.
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
What profile associated with the Oracle user (SELECT profile FROM dba_users WHERE username = <<user name>>)? What resource limits are defined for that profile (SELECT resource_name, limit FROM dba_profiles WHERE profile = <<profile name from dba_users>> AND resource = 'KERNEL' AND LIMIT != 'UNLIMITED')
Oracle cannot time out a session. It is possible that Oracle has been configured to kill sessions that use too many resources (i.e. via a profile). It is also possible that some network component is causing the timeout (firewalls are a popular source of unexpected timeouts). If you do have a network component that is timing out the session, it is unlikely that any Oracle configuration change could resolve the problem.