在ORACLE數據庫中,如果不對監聽日志文件(listener.log)進行截斷,那麼監聽日志文件(listener.log)會變得越來越大,想必不少人聽說過關於“LISTENER.LOG日志大小不能超過2GB,超過會導致LISTENER監聽器無法處理新的連接”,當然這個不是真理,不會絕對出現,只是發生在老舊的32bit Linux或Unix系統下面,真實的原因是一些32bit OS自帶的文件系統不支持2GB以上的文件,導致監聽服務進程(tnslsnr)append write日志文件出錯。
那麼是否不需要對監聽日志文件進行截斷維護呢? 答案是否定的。當然要對監聽日志文件(listener.log)進行定期清理,如果不定期清理,會遇到下面一些麻煩:
1:監聽日志文件(listener.log)變得越來越大,占用額外的存儲空間。(當然現在存儲白菜價,不差那幾G的空間。但是我們還是要本著工匠情懷,精益求精)
2:監聽日志文件(listener.log)變得太大會帶來一些問題:LISTENER.LOG日志大小不能超過2GB,超過會導致LISTENER監聽器無法處理新的連接。
3:監聽日志文件(listener.log)變得太大,給寫入、查看帶來的一些性能問題、麻煩。
也有人說是監聽服務進程一般使用標准C函數Write寫出到Listener.log,listener.log文件時使用的是O_WRONLY|O_CREAT|O_APPEND,O_APPEND即追加到文件的尾端,一般來說追加寫方式不會因為文件越大寫地越慢。撇開這個不談,在一個很大的監聽日志文件(listener.log)查找某一天或某一個錯誤,這個確實會帶來一些性能問題。查找起來也相當麻煩。
所以應該定期對監聽日志文件(listener.log)進行清理,另外一種說法叫截斷日志文件。關於截斷監聽日志,要注意一些問題。初學ORACLE的時候遇到一個錯誤的截斷監聽日志的,下面演示一下
[oracle@DB-Server log]$ mv listener.log listener.log.20150114
[oracle@DB-Server log]$ cp /dev/null listener.log
[oracle@DB-Server log]$ more listener.log
如上所示,這樣截斷監聽日志(listener.log)後,監聽服務進程(tnslsnr)並不會將新的監聽信息寫入listener.log,而是繼續寫入listener.log.20150114
[oracle@DB-Server log]$ tail listener.log.20150114
15-JAN-2015 22:40:01 * service_update * EPPS * 0
15-JAN-2015 22:47:59 * service_update * EPPS * 0
15-JAN-2015 22:58:00 * service_update * EPPS * 0
15-JAN-2015 23:08:00 * service_update * EPPS * 0
15-JAN-2015 23:18:01 * service_update * EPPS * 0
15-JAN-2015 23:28:01 * service_update * EPPS * 0
15-JAN-2015 23:38:02 * service_update * EPPS * 0
15-JAN-2015 23:41:02 * service_update * EPPS * 0
15-JAN-2015 23:50:26 * ping * 0
15-JAN-2015 23:51:03 * service_update * EPPS * 0
[oracle@DB-Server log]$ tail listener.log.20150114
15-JAN-2015 23:18:01 * service_update * EPPS * 0
15-JAN-2015 23:28:01 * service_update * EPPS * 0
15-JAN-2015 23:38:02 * service_update * EPPS * 0
15-JAN-2015 23:41:02 * service_update * EPPS * 0
15-JAN-2015 23:50:26 * ping * 0
15-JAN-2015 23:51:03 * service_update * EPPS * 0
15-JAN-2015 23:57:40 * ping * 0
15-JAN-2015 23:57:41 * ping * 0
15-JAN-2015 23:57:42 * ping * 0
15-JAN-2015 23:57:43 * ping * 0
[oracle@DB-Server log]$ more listener.log
[oracle@DB-Server log]$
規范正確的流程應該這麼處理:
Step 1:首先停止監聽服務進程(tnslsnr)記錄日志。
[oracle@DB-Server log]$ lsnrctl set log_status off;
LSNRCTL for Linux: Version 10.2.0.4.0 - Production on 16-JAN-2015 00:02:59
Copyright (c) 1991, 2007, Oracle. All rights reserved.
Connecting to (DESCRIPTION=(ADDRESS=(PROTOCOL=TCP)(HOST=192.168.9.15)(PORT=1521)))
LISTENER parameter "log_status" set to OFF
The command completed successfully
Step 2:將監聽日志文件(listener.log)復制一份,以listener.log.yyyymmdd格式命名
[oracle@DB-Server log]$ cp listener.log listener.log.20150114
Step 3:將監聽日志文件(listener.log)清空。清空文件的方法有很多
3.1 echo “” > filename
3.2 cp /dev/null 或 echo /dev/null > filename
Step 4:開啟監聽服務進程(tnslsnr)記錄日志
[oracle@DB-Server log]$ lsnrctl set log_status on;
當然也可以移走監聽日志文件(listener.log),數據庫實例會自動創建一個listener.log文件。
% lsnrctl set log_status off
% mv listener.log listener.yyyymmdd
% lsnrctl set log_status on
當然這些操作應該通過shell腳本來處理,然後結合crontab作業定期清理、截斷監聽日志文件。例如網上的一個清理、截斷監聽日志文件的shell腳本。
rq=` date +"%d" `
cp $ORACLE_HOME/network/log/listener.log $ORACLE_BACKUP/network/log/listener_$rq.log
su - oracle -c "lsnrctl set log_status off"
cp /dev/null $ORACLE_HOME/network/log/listener.log
su - oracle -c "lsnrctl set log_status on"
這樣的腳本還沒有解決一個問題,就是截斷的監聽日志文件保留多久的問題。比如我只想保留這些截斷的監聽日志一個月時間,我希望作業自動維護。不需要我去手工操作。有這樣一個腳本cls_oracle.sh可以完全做到這個,當然它還會歸檔、清理其它日志文件,例如告警文件(alert_sid.log)等等。功能非常強大。
#!/bin/bash
#
# Script used to cleanup any Oracle environment.
#
# Cleans: audit_log_dest
# background_dump_dest
# core_dump_dest
# user_dump_dest
#
# Rotates: Alert Logs
# Listener Logs
#
# Scheduling: 00 00 * * * /home/oracle/_cron/cls_oracle/cls_oracle.sh -d 31 > /home/oracle/_cron/cls_oracle/cls_oracle.log 2>
&1
#
# Created By: Tommy Wang 2012-09-10
#
# History:
#
RM="rm -f"
RMDIR="rm -rf"
LS="ls -l"
MV="mv"
TOUCH="touch"
TESTTOUCH="echo touch"
TESTMV="echo mv"
TESTRM=$LS
TESTRMDIR=$LS
SUCCESS=0
FAILURE=1
TEST=0
HOSTNAME=`hostname`
ORAENV="oraenv"
TODAY=`date +%Y%m%d`
ORIGPATH=/usr/local/bin:$PATH
ORIGLD=$LD_LIBRARY_PATH
export PATH=$ORIGPATH
# Usage function.
f_usage(){
echo "Usage: `basename $0` -d DAYS [-a DAYS] [-b DAYS] [-c DAYS] [-n DAYS] [-r DAYS] [-u DAYS] [-t] [-h]"
echo " -d = Mandatory default number of days to keep log files that are not explicitly passed as parameters."
echo " -a = Optional number of days to keep audit logs."
echo " -b = Optional number of days to keep background dumps."
echo " -c = Optional number of days to keep core dumps."
echo " -n = Optional number of days to keep network log files."
echo " -r = Optional number of days to keep clusterware log files."
echo " -u = Optional number of days to keep user dumps."
echo " -h = Optional help mode."
echo " -t = Optional test mode. Does not delete any files."
}
if [ $# -lt 1 ]; then
f_usage
exit $FAILURE
fi
# Function used to check the validity of days.
f_checkdays(){
if [ $1 -lt 1 ]; then
echo "ERROR: Number of days is invalid."
exit $FAILURE
fi
if [ $? -ne 0 ]; then
echo "ERROR: Number of days is invalid."
exit $FAILURE
fi
}
# Function used to cut log files.
f_cutlog(){
# Set name of log file.
LOG_FILE=$1
CUT_FILE=${LOG_FILE}.${TODAY}
FILESIZE=`ls -l $LOG_FILE | awk '{print $5}'`
# Cut the log file if it has not been cut today.
if [ -f $CUT_FILE ]; then
echo "Log Already Cut Today: $CUT_FILE"
elif [ ! -f $LOG_FILE ]; then
echo "Log File Does Not Exist: $LOG_FILE"
elif [ $FILESIZE -eq 0 ]; then
echo "Log File Has Zero Size: $LOG_FILE"
else
# Cut file.
echo "Cutting Log File: $LOG_FILE"
$MV $LOG_FILE $CUT_FILE
$TOUCH $LOG_FILE
fi
}
# Function used to delete log files.
f_deletelog(){
# Set name of log file.
CLEAN_LOG=$1
# Set time limit and confirm it is valid.
CLEAN_DAYS=$2
f_checkdays $CLEAN_DAYS
# Delete old log files if they exist.
find $CLEAN_LOG.[0-9][0-9][0-9][0-9][0-9][0-9][0-9][0-9] -type f -mtime +$CLEAN_DAYS -exec $RM {} \; 2>/dev/null
}
# Function used to get database parameter values.
f_getparameter(){
if [ -z "$1" ]; then
return
fi
PARAMETER=$1
sqlplus -s /nolog <<EOF | awk -F= "/^a=/ {print \$2}"
set head off pagesize 0 feedback off linesize 200
whenever sqlerror exit 1
conn / as sysdba
select 'a='||value from v\$parameter where name = '$PARAMETER';
EOF
}
# Function to get unique list of directories.
f_getuniq(){
if [ -z "$1" ]; then
return
fi
ARRCNT=0
MATCH=N
x=0
for e in `echo $1`; do
if [ ${#ARRAY[*]} -gt 0 ]; then
# See if the array element is a duplicate.
while [ $x -lt ${#ARRAY[*]} ]; do
if [ "$e" = "${ARRAY[$x]}" ]; then
MATCH=Y
fi
done
fi
if [ "$MATCH" = "N" ]; then
ARRAY[$ARRCNT]=$e
ARRCNT=`expr $ARRCNT+1`
fi
x=`expr $x + 1`
done
echo ${ARRAY[*]}
}
# Parse the command line options.
while getopts a:b:c:d:n:r:u:th OPT; do
case $OPT in
a) ADAYS=$OPTARG
;;
b) BDAYS=$OPTARG
;;
c) CDAYS=$OPTARG
;;
d) DDAYS=$OPTARG
;;
n) NDAYS=$OPTARG
;;
r) RDAYS=$OPTARG
;;
u) UDAYS=$OPTARG
;;
t) TEST=1
;;
h) f_usage
exit 0
;;
*) f_usage
exit 2
;;
esac
done
shift $(($OPTIND - 1))
# Ensure the default number of days is passed.
if [ -z "$DDAYS" ]; then
echo "ERROR: The default days parameter is mandatory."
f_usage
exit $FAILURE
fi
f_checkdays $DDAYS
echo "`basename $0` Started `date`."
# Use test mode if specified.
if [ $TEST -eq 1 ]
then
RM=$TESTRM
RMDIR=$TESTRMDIR
MV=$TESTMV
TOUCH=$TESTTOUCH
echo "Running in TEST mode."
fi
# Set the number of days to the default if not explicitly set.
ADAYS=${ADAYS:-$DDAYS}; echo "Keeping audit logs for $ADAYS days."; f_checkdays $ADAYS
BDAYS=${BDAYS:-$DDAYS}; echo "Keeping background logs for $BDAYS days."; f_checkdays $BDAYS
CDAYS=${CDAYS:-$DDAYS}; echo "Keeping core dumps for $CDAYS days."; f_checkdays $CDAYS
NDAYS=${NDAYS:-$DDAYS}; echo "Keeping network logs for $NDAYS days."; f_checkdays $NDAYS
RDAYS=${RDAYS:-$DDAYS}; echo "Keeping clusterware logs for $RDAYS days."; f_checkdays $RDAYS
UDAYS=${UDAYS:-$DDAYS}; echo "Keeping user logs for $UDAYS days."; f_checkdays $UDAYS
# Check for the oratab file.
if [ -f /var/opt/oracle/oratab ]; then
ORATAB=/var/opt/oracle/oratab
elif [ -f /etc/oratab ]; then
ORATAB=/etc/oratab
else
echo "ERROR: Could not find oratab file."
exit $FAILURE
fi
# Build list of distinct Oracle Home directories.
OH=`egrep -i ":Y|:N" $ORATAB | grep -v "^#" | grep -v "\*" | cut -d":" -f2 | sort | uniq`
# Exit if there are not Oracle Home directories.
if [ -z "$OH" ]; then
echo "No Oracle Home directories to clean."
exit $SUCCESS
fi
# Get the list of running databases.
SIDS=`ps -e -o args | grep pmon | grep -v grep | awk -F_ '{print $3}' | sort`
# Gather information for each running database.
for ORACLE_SID in `echo $SIDS`
do
# Set the Oracle environment.
ORAENV_ASK=NO
export ORACLE_SID
. $ORAENV
if [ $? -ne 0 ]; then
echo "Could not set Oracle environment for $ORACLE_SID."
else
export LD_LIBRARY_PATH=$ORACLE_HOME/lib:$ORIGLD
ORAENV_ASK=YES
echo "ORACLE_SID: $ORACLE_SID"
# Get the audit_dump_dest.
ADUMPDEST=`f_getparameter audit_dump_dest`
if [ ! -z "$ADUMPDEST" ] && [ -d "$ADUMPDEST" 2>/dev/null ]; then
echo " Audit Dump Dest: $ADUMPDEST"
ADUMPDIRS="$ADUMPDIRS $ADUMPDEST"
fi
# Get the background_dump_dest.
BDUMPDEST=`f_getparameter background_dump_dest`
echo " Background Dump Dest: $BDUMPDEST"
if [ ! -z "$BDUMPDEST" ] && [ -d "$BDUMPDEST" ]; then
BDUMPDIRS="$BDUMPDIRS $BDUMPDEST"
fi
# Get the core_dump_dest.
CDUMPDEST=`f_getparameter core_dump_dest`
echo " Core Dump Dest: $CDUMPDEST"
if [ ! -z "$CDUMPDEST" ] && [ -d "$CDUMPDEST" ]; then
CDUMPDIRS="$CDUMPDIRS $CDUMPDEST"
fi
# Get the user_dump_dest.
UDUMPDEST=`f_getparameter user_dump_dest`
echo " User Dump Dest: $UDUMPDEST"
if [ ! -z "$UDUMPDEST" ] && [ -d "$UDUMPDEST" ]; then
UDUMPDIRS="$UDUMPDIRS $UDUMPDEST"
fi
fi
done
# Do cleanup for each Oracle Home.
for ORAHOME in `f_getuniq "$OH"`
do
# Get the standard audit directory if present.
if [ -d $ORAHOME/rdbms/audit ]; then
ADUMPDIRS="$ADUMPDIRS $ORAHOME/rdbms/audit"
fi
# Get the Cluster Ready Services Daemon (crsd) log directory if present.
if [ -d $ORAHOME/log/$HOSTNAME/crsd ]; then
CRSLOGDIRS="$CRSLOGDIRS $ORAHOME/log/$HOSTNAME/crsd"
fi
# Get the Oracle Cluster Registry (OCR) log directory if present.
if [ -d $ORAHOME/log/$HOSTNAME/client ]; then
OCRLOGDIRS="$OCRLOGDIRS $ORAHOME/log/$HOSTNAME/client"
fi
# Get the Cluster Synchronization Services (CSS) log directory if present.
if [ -d $ORAHOME/log/$HOSTNAME/cssd ]; then
CSSLOGDIRS="$CSSLOGDIRS $ORAHOME/log/$HOSTNAME/cssd"
fi
# Get the Event Manager (EVM) log directory if present.
if [ -d $ORAHOME/log/$HOSTNAME/evmd ]; then
EVMLOGDIRS="$EVMLOGDIRS $ORAHOME/log/$HOSTNAME/evmd"
fi
# Get the RACG log directory if present.
if [ -d $ORAHOME/log/$HOSTNAME/racg ]; then
RACGLOGDIRS="$RACGLOGDIRS $ORAHOME/log/$HOSTNAME/racg"
fi
done
# Clean the audit_dump_dest directories.
if [ ! -z "$ADUMPDIRS" ]; then
for DIR in `f_getuniq "$ADUMPDIRS"`; do
if [ -d $DIR ]; then
echo "Cleaning Audit Dump Directory: $DIR"
find $DIR -type f -name "*.aud" -mtime +$ADAYS -exec $RM {} \; 2>/dev/null
fi
done
fi
# Clean the background_dump_dest directories.
if [ ! -z "$BDUMPDIRS" ]; then
for DIR in `f_getuniq "$BDUMPDIRS"`; do
if [ -d $DIR ]; then
echo "Cleaning Background Dump Destination Directory: $DIR"
# Clean up old trace files.
find $DIR -type f -name "*.tr[c,m]" -mtime +$BDAYS -exec $RM {} \; 2>/dev/null
find $DIR -type d -name "cdmp*" -mtime +$BDAYS -exec $RMDIR {} \; 2>/dev/null
fi
if [ -d $DIR ]; then
# Cut the alert log and clean old ones.
for f in `find $DIR -type f -name "alert\_*.log" ! -name "alert_[0-9A-Z]*.[0-9]*.log" 2>/dev/null`; do
echo "Alert Log: $f"
f_cutlog $f
f_deletelog $f $BDAYS
done
fi
done
fi
# Clean the core_dump_dest directories.
if [ ! -z "$CDUMPDIRS" ]; then
for DIR in `f_getuniq "$CDUMPDIRS"`; do
if [ -d $DIR ]; then
echo "Cleaning Core Dump Destination: $DIR"
find $DIR -type d -name "core*" -mtime +$CDAYS -exec $RMDIR {} \; 2>/dev/null
fi
done
fi
# Clean the user_dump_dest directories.
if [ ! -z "$UDUMPDIRS" ]; then
for DIR in `f_getuniq "$UDUMPDIRS"`; do
if [ -d $DIR ]; then
echo "Cleaning User Dump Destination: $DIR"
find $DIR -type f -name "*.trc" -mtime +$UDAYS -exec $RM {} \; 2>/dev/null
fi
done
fi
# Cluster Ready Services Daemon (crsd) Log Files
for DIR in `f_getuniq "$CRSLOGDIRS $OCRLOGDIRS $CSSLOGDIRS $EVMLOGDIRS $RACGLOGDIRS"`; do
if [ -d $DIR ]; then
echo "Cleaning Clusterware Directory: $DIR"
find $DIR -type f -name "*.log" -mtime +$RDAYS -exec $RM {} \; 2>/dev/null
fi
done
# Clean Listener Log Files.
# Get the list of running listeners. It is assumed that if the listener is not running, the log file does not need to be cut.
ps -e -o args | grep tnslsnr | grep -v grep | while read LSNR; do
# Derive the lsnrctl path from the tnslsnr process path.
TNSLSNR=`echo $LSNR | awk '{print $1}'`
ORACLE_PATH=`dirname $TNSLSNR`
ORACLE_HOME=`dirname $ORACLE_PATH`
PATH=$ORACLE_PATH:$ORIGPATH
LD_LIBRARY_PATH=$ORACLE_HOME/lib:$ORIGLD
LSNRCTL=$ORACLE_PATH/lsnrctl
echo "Listener Control Command: $LSNRCTL"
# Derive the listener name from the running process.
LSNRNAME=`echo $LSNR | awk '{print $2}' | tr "[:upper:]" "[:lower:]"`
echo "Listener Name: $LSNRNAME"
# Get the listener version.
LSNRVER=`$LSNRCTL version | grep "LSNRCTL" | grep "Version" | awk '{print $5}' | awk -F. '{print $1}'`
echo "Listener Version: $LSNRVER"
# Get the TNS_ADMIN variable.
echo "Initial TNS_ADMIN: $TNS_ADMIN"
unset TNS_ADMIN
TNS_ADMIN=`$LSNRCTL status $LSNRNAME | grep "Listener Parameter File" | awk '{print $4}'`
if [ ! -z $TNS_ADMIN ]; then
export TNS_ADMIN=`dirname $TNS_ADMIN`
else
export TNS_ADMIN=$ORACLE_HOME/network/admin
fi
echo "Network Admin Directory: $TNS_ADMIN"
# If the listener is 11g, get the diagnostic dest, etc...
if [ $LSNRVER -ge 11 ]; then
# Get the listener log file directory.
LSNRDIAG=`$LSNRCTL<<EOF | grep log_directory | awk '{print $6}'
set current_listener $LSNRNAME
show log_directory
EOF`
echo "Listener Diagnostic Directory: $LSNRDIAG"
# Get the listener trace file name.
LSNRLOG=`lsnrctl<<EOF | grep trc_directory | awk '{print $6"/"$1".log"}'
set current_listener $LSNRNAME
show trc_directory
EOF`
echo "Listener Log File: $LSNRLOG"
# If 10g or lower, do not use diagnostic dest.
else
# Get the listener log file location.
LSNRLOG=`$LSNRCTL status $LSNRNAME | grep "Listener Log File" | awk '{print $4}'`
fi
# See if the listener is logging.
if [ -z "$LSNRLOG" ]; then
echo "Listener Logging is OFF. Not rotating the listener log."
# See if the listener log exists.
elif [ ! -r "$LSNRLOG" ]; then
echo "Listener Log Does Not Exist: $LSNRLOG"
# See if the listener log has been cut today.
elif [ -f $LSNRLOG.$TODAY ]; then
echo "Listener Log Already Cut Today: $LSNRLOG.$TODAY"
# Cut the listener log if the previous two conditions were not met.
else
# Remove old 11g+ listener log XML files.
if [ ! -z "$LSNRDIAG" ] && [ -d "$LSNRDIAG" ]; then
echo "Cleaning Listener Diagnostic Dest: $LSNRDIAG"
find $LSNRDIAG -type f -name "log\_[0-9]*.xml" -mtime +$NDAYS -exec $RM {} \; 2>/dev/null
fi
# Disable logging.
$LSNRCTL <<EOF
set current_listener $LSNRNAME
set log_status off
EOF
# Cut the listener log file.
f_cutlog $LSNRLOG
# Enable logging.
$LSNRCTL <<EOF
set current_listener $LSNRNAME
set log_status on
EOF
# Delete old listener logs.
f_deletelog $LSNRLOG $NDAYS
fi
done
echo "`basename $0` Finished `date`."
exit
在crontab中設置一個作業,每天晚上凌晨零點運行這個腳本,日志文件保留31天。
00 00 * * * /home/oracle/_cron/cls_oracle/cls_oracle.sh -d 31 > /home/oracle/_cron/cls_oracle/cls_oracle.sh.log 2>&1
如下所示,非常自動化的維護、清理了監聽日志文件(listener.log),又能保留一段時間以便查找、跟蹤問題