ORACLE清理、截断监听日志文件（listener.log）

超、凢脫俗 2022-05-23 01:19 363阅读 0赞

转自：https://www.cnblogs.com/kerrycode/p/4227579.html

在ORACLE数据库中，如果不对监听日志文件（listener.log）进行截断,那么监听日志文件（listener.log）会变得越来越大，想必不少人听说过关于“LISTENER.LOG日志大小不能超过2GB，超过会导致LISTENER监听器无法处理新的连接”，当然这个不是真理，不会绝对出现，只是发生在老旧的32bit Linux或Unix系统下面，真实的原因是一些32bit OS自带的文件系统不支持2GB以上的文件，导致监听服务进程（tnslsnr）append write日志文件出错。

那么是否不需要对监听日志文件进行截断维护呢？ 答案是否定的。当然要对监听日志文件（listener.log）进行定期清理，如果不定期清理，会遇到下面一些麻烦：

1：监听日志文件（listener.log）变得越来越大，占用额外的存储空间。（当然现在存储白菜价，不差那几G的空间。但是我们还是要本着工匠情怀，精益求精）

2：监听日志文件（listener.log）变得太大会带来一些问题：LISTENER.LOG日志大小不能超过2GB，超过会导致LISTENER监听器无法处理新的连接。

3：监听日志文件（listener.log）变得太大，给写入、查看带来的一些性能问题、麻烦。

也有人说是监听服务进程一般使用标准C函数Write写出到Listener.log，listener.log文件时使用的是O\_WRONLY|O\_CREAT|O\_APPEND，O\_APPEND即追加到文件的尾端，一般来说追加写方式不会因为文件越大写地越慢。撇开这个不谈，在一个很大的监听日志文件（listener.log）查找某一天或某一个错误，这个确实会带来一些性能问题。查找起来也相当麻烦。

所以应该定期对监听日志文件（listener.log）进行清理，另外一种说法叫截断日志文件。关于截断监听日志，要注意一些问题。初学ORACLE的时候遇到一个错误的截断监听日志的，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，这样截断监听日志（listener.log）后，监听服务进程（tnslsnr）并不会将新的监听信息写入listener.log，而是继续写入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]$

[![clip\_image001][clip_image001]][clip_image001_clip_image001]

**规范正确的流程应该这么处理：**

**Step 1：首先停止监听服务进程（tnslsnr）记录日志。**

[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

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：将监听日志文件（listener.log）复制一份，以listener.log.yyyymmdd格式命名**

\[oracle@DB-Server log\]$ cp listener.log listener.log.20150114

**Step 3：将监听日志文件（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：开启监听服务进程（tnslsnr）记录日志**

\[oracle@DB-Server log\]$ lsnrctl set log\_status on;

当然也可以移走监听日志文件（listener.log），数据库实例会自动创建一个listener.log文件。

% lsnrctl set log_status off

% mv listener.log listener.yyyymmdd

% lsnrctl set log_status on

当然这些操作应该通过shell脚本来处理，然后结合crontab作业定期清理、截断监听日志文件。例如网上的一个清理、截断监听日志文件的shell脚本。

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"

这样的脚本还没有解决一个问题，就是截断的监听日志文件保留多久的问题。比如我只想保留这些截断的监听日志一个月时间，我希望作业自动维护。不需要我去手工操作。有这样一个脚本cls\_oracle.sh可以完全做到这个，当然它还会归档、清理其它日志文件，例如告警文件（alert\_sid.log）等等。功能非常强大。

#!/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>

# 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

# Function used to check the validity of days.

f_checkdays(){

if [ $1 -lt 1 ]; then

echo "ERROR: Number of days is invalid."

exit $FAILURE

if [ $? -ne 0 ]; then

echo "ERROR: Number of days is invalid."

exit $FAILURE

}

# 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

}

# 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

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

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

done

if [ "$MATCH" = "N" ]; then

ARRAY[$ARRCNT]=$e

ARRCNT=`expr $ARRCNT+1`

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

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."

# 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

# Build list of distinct Oracle Home directories.

# Exit if there are not Oracle Home directories.

if [ -z "$OH" ]; then

echo "No Oracle Home directories to clean."

exit $SUCCESS

# 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`

# 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"

# 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"

# 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"

# 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"

done

# Do cleanup for each Oracle Home.

for ORAHOME in `f_getuniq "$OH"`

# Get the standard audit directory if present.

if [ -d $ORAHOME/rdbms/audit ]; then

ADUMPDIRS="$ADUMPDIRS $ORAHOME/rdbms/audit"

# Get the Cluster Ready Services Daemon (crsd) log directory if present.

if [ -d $ORAHOME/log/$HOSTNAME/crsd ]; then

CRSLOGDIRS="$CRSLOGDIRS $ORAHOME/log/$HOSTNAME/crsd"

# Get the  Oracle Cluster Registry (OCR) log directory if present.

if [ -d $ORAHOME/log/$HOSTNAME/client ]; then

OCRLOGDIRS="$OCRLOGDIRS $ORAHOME/log/$HOSTNAME/client"

# Get the Cluster Synchronization Services (CSS) log directory if present.

if [ -d $ORAHOME/log/$HOSTNAME/cssd ]; then

CSSLOGDIRS="$CSSLOGDIRS $ORAHOME/log/$HOSTNAME/cssd"

# Get the Event Manager (EVM) log directory if present.

if [ -d $ORAHOME/log/$HOSTNAME/evmd ]; then

EVMLOGDIRS="$EVMLOGDIRS $ORAHOME/log/$HOSTNAME/evmd"

# Get the RACG log directory if present.

if [ -d $ORAHOME/log/$HOSTNAME/racg ]; then

RACGLOGDIRS="$RACGLOGDIRS $ORAHOME/log/$HOSTNAME/racg"

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

done

# 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

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

# 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

done

# 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

done

# 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

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

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}'`

# 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

# 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

done

echo "`basename $0` Finished `date`."

exit

在crontab中设置一个作业，每天晚上凌晨零点运行这个脚本，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，非常自动化的维护、清理了监听日志文件（listener.log）,又能保留一段时间以便查找、跟踪问题

[![clip\_image002][clip_image002]][clip_image002_clip_image002]

[clip_image001]: https://images0.cnblogs.com/blog/73542/201501/160041289018802.png
[clip_image001_clip_image001]: /images/20220523/3d55b3f4b09e48a79b994d5675dd5667.png
[clip_image002]: https://images0.cnblogs.com/blog/73542/201501/160041312296947.png
[clip_image002_clip_image002]: /images/20220523/1ca47374a92b4049bd00f40da25fe8ce.png