Oracle database logs doesn’t rotate by it self , and as time goes by, your
server may hold logs that are too big to read and takes too much storage space .
this can get your server to a maximum capacity , and in some cases crush your server .
The best thing i found is to use logrotate to handle this rotations .
there are 2 files that needs to be rotate ( depend on your infrastructure ) this files
are alert log and listener log . both can grow to unlimited size .
Create a new logrotate rule by edit a files
/etc/logrotate.d/oracle-alert and /etc/logrotate.d/oracle-listener
the oracle-alert file should point to the alert log usualy located under
$ORACLE_HOME/diag/rdbms/<database>/<sid>/trace/alert_.log
here is an example of oracle-alert , that will rotate weekly and store for 4 files back
also it will compress that backups and create a new file with the correct permissions .
* note that Oracle will create a new alert log ,if the file is missing, upon next event
/opt/app/DB/diag/rdbms/example/example1/trace/alert_example1.log {
compress
rotate 4
weekly
create 640 oracle oinstall
}
the next thing to handle is the listener , now the listener log cannot be remove just like that
if you do so , the listener would stop logging into that file . solving it with a special
commands that restart just the loger of the listener .
the location of the listener log is under $ORACLE_HOME/diag/tnslsnr/<database>/listener/trace/listener.log
This example shows how to weekly rotate and compress
/opt/app/DB/diag/tnslsnr/example/listener/trace/listener.log {
compress
rotate 4
weekly
create 640 oracle oinstall
prerotate
su - oracle "lsnrctl set Log_status off"
endscript
postrotate
su - oracle "lsnrctl set Log_status on"
endscript
}