Most of you may already know how to configure the log rotation for your database alert log or the listener log etc,. As root, you can manually create a configuration file as below, and put it into /etc/logrotate.d/.
/u01/app/oracle/diag/rdbms/test1/TEST1/trace/alert_TEST1.log
{
daily
rotate 15
compress
copytruncate
delaycompress
create 0640 oracle dba
notifempty
}
It’s easy to create the log rotation configuration file for one database. But what if you have 20+ databases running on your server, are you still planning to manually locate their alert log files? Today I would like to introduce a shell script that will automatically search the running ASM instance, databases and listeners and create the configuration files for you.
Here are the steps on how to run shell scripts.
1. Run the shell script
[[email protected] ~]$ ./gen_logrotate_config.sh
2. Three configuration files will be created: oracle_rdbms_log, oracle_asm_log and oracle_listener_log. Let’s say we have 4 running databases: TEST1, TEST2, TEST3 and TEST4, one ASM instance +ASM1, three listeners: listener_scan1, listener_scan2 and LISTENER. The configuration files will be like:
[[email protected] ~]$ cat oracle_rdbms_log /u01/app/oracle/diag/rdbms/test1/TEST1/trace/alert_TEST1.log /u01/app/oracle/diag/rdbms/test2/TEST2/trace/alert_TEST2.log /u01/app/oracle/diag/rdbms/test3/TEST3/trace/alert_TEST3.log /u01/app/oracle/diag/rdbms/test3/TEST3/trace/alert_TEST4.log { daily rotate 15 compress copytruncate delaycompress create 0640 oracle dba notifempty } [[email protected] ~]$ cat oracle_asm_log /u01/app/oracle/diag/asm/+asm/+ASM1/trace/alert_+ASM1.log { daily rotate 15 compress copytruncate delaycompress create 0640 oracle dba notifempty } [[email protected] ~]$ cat oracle_listener_log /u01/app/oracle/diag/tnslsnr/TESTSERVER/listener_scan2/trace/listener_scan2.log /u01/app/oracle/diag/tnslsnr/TESTSERVER/listener_scan1/trace/listener_scan1.log /u01/app/oracle/diag/tnslsnr/TESTSERVER/listener/trace/listener.log { daily rotate 15 compress copytruncate delaycompress create 0640 oracle dba notifempty }
3. Put three configuration files to /etc/logrotate.d/ then you will see the log files will be rotated on next day.
You can find the code here and enjoy!
2 Comments. Leave new
Thanks for this script. Most useful, however, you appear to have a blank line at line 23 – between “END_SQL” and “done)…” for the “oracle_asm_log” output. This causes the output to lose the actual name of the log file for the ASM instance(s).
I’ve used this on our cloud (OCI) servers and found that Oracle have set up the servers with the ASM details hard coded into .bashrc under the grid user – not helpful! This obviously caused the script to fail when attempting to extract the alert file names. Adding an entry for +ASM1 into oratab solved that problem.
I’m not sure that the “oracle_rdbms_log” output is (always?) correct though. Our alert logs, in OCI, are located in the core_dump_dest (with cdump replaced by trace, of course) rather than in the location for background_dump_dest as per the script, but a quick edit fixed that too.
It could be that our cloud setup is “broken”, who knows! :)
Thanks again.
Cheers,
Norm.
you could have done it in one line:
/u01/app/oracle/diag/rdbms/*/*/trace/alert_*.log
and for the listener.log, this is really old fashioned!
append following two lines to listener.ora (for each listener) and restart/reload the listeners:
LOG_FILE_NUM_LISTENER=4
LOG_FILE_SIZE_LISTENER=50
This will rotate the logs when grown up to 50MB and keep the last 4 logfiles.
You don’t need any script nor cronjob.