I have an Oracle Database installed over Unix system. On this single instance database I have a lot of sessions opened from an application pool (application is installed on another server). Every sessions are opened with the same user name. As sessions can be closed by the pool (really, it can be) I can't capture session id and - for each - enable query trace functionality. The application is a closed source code (and proprietary), so I can't (and don't want to) reverse engineering it.
Now, I need to capture every query that this user execute. Ideally I want to get one file by user session (but it will be perfect if I get one file with all queries). My real goal is to get queries but session_id and/or query's timestamp could be welcome.
Important note: as my database is a Standard Edition I have no access to any Oracle advanced pack as "Oracle Tuning Pack", "Oracle Diagnostic Pack" or so on ... I only have access to the StatPack and regular SQL query over all database dictionary objects (as a sysdba could be).
Platform information:
- Unix RedHat 6.x
- Oracle Database 11g Standard edition
- only one database instance
- number of sessions opened between 150 and 250
Thanks for your ideas
Thomas
-
Why do you need this collection of statements?miracle173– miracle1732017年03月09日 05:51:52 +00:00Commented Mar 9, 2017 at 5:51
-
@miracle173 I need this for two reasons: 1) I want to be able to analyse each query done by this application (for example for missing index analyse) 2) I want to be able to replay all query and transactions on a test databasetdaget– tdaget2017年03月22日 09:54:26 +00:00Commented Mar 22, 2017 at 9:54
1 Answer 1
As per my knowledge, there is no straightforward way to achieve this goal. However, I have tried to compose the following steps to fulfill the requirements.
You can enable trace for all sessions by using the following statement.
alter system set events '10046 TRACE NAME CONTEXT FOREVER, LEVEL 4';
To disable this tracing, use the following statement.
alter system set events '10046 trace name context off';
Steps:
SQL> conn / as sysdba
SQL> alter system set events '10046 TRACE NAME CONTEXT FOREVER, LEVEL 4';
SQL> SELECT LISTAGG(tf, ' ') WITHIN GROUP (ORDER BY tf) "Trace_List"
FROM (
SELECT sys_context('userenv','instance_name') || '_ora_'|| p.spid || '.trc' as tf FROM
v$process p join v$session s
ON (s.paddr=p.addr)
WHERE s.username='JAY'
);
Trace_List
--------------------------------------------------------------------------------
orcl_ora_3443.trc orcl_ora_3458.trc orcl_ora_3473.trc
USER_DUMP_DEST
to find trace files.
SQL> show parameter user_dump_dest
NAME TYPE VALUE
------------------------------------ ----------- ------------------------------
user_dump_dest string /u01/app/oracle/diag/rdbms/orcl/orcl/trace
Now, you have got multiple trace files scattered on your user dump dest. In order to combine and create a single trace file, you can use trcsess
utility, which allows trace information from multiple trace files to be identified and consolidated into a single trace file. You can consolidate these trace files based on following criteria. Choose the which is common for these all sessions.
In this case, I am going to use service name.
SQL> select service_name from v$session where username='JAY';
SERVICE_NAME
----------------------------------------------------------------
SYS$USERS
SYS$USERS
SYS$USERS
Let's use the trcsess
utility.
[server1@oracle ]$ trcsess output=/home/oracle/Desktop/main_trace_file.trc service='SYS$USERS' orcl_ora_3443.trc orcl_ora_3458.trc orcl_ora_3473.trc
We have created a single trace file and we can use tkprof
utility to create formatted output from this trace file.
[server1@oracle ]$ tkprof main_trace_file.trc main_formatted_trace.txt
Note: If you were using Enterprise Edition you could use fine-grained auditing which is not available in Standard Edition
References:
Explore related questions
See similar questions with these tags.