工作中DBA经常会查看alert log来检查数据库后台都记录了些什么日志,如果只想看某一天或者某段时间范围的日志,能够把这些日志从大的alert log中截取下来放到一个单独的文件中,对于查看和下载都是很有意义的事,接下来附上脚本
script.sh
#!/bin/bash ################################################################################ # this script is to get the specified date log of oracle database alert log. # # author: yangbao # # usage: ./script.sh 2019-5-1 --> get the log on 2019-5-1 # # ./script.sh 2019-5-1 2019-5-3 --> get the log between 2019-5-1 and 2019-5-3 # ################################################################################ # 判断日期格式是否有效 check_date() { date -d "$1" &> /dev/null flag=$? if [ $flag -eq 1 ]; then echo "date is incorrect, please input the correct date formate like 2019-5-1." exit 1 fi } # 得到开始和结束的日期的指定格式 get_begin_time() { year1=`date -d "$1" +%Y` month1=`date -d "$1" +%b` day1=`date -d "$1" +%d` week1=`date -d "$1" +%a` } get_end_time() { year2=`date -d "$1" +%Y` month2=`date -d "$1" +%b` day2=`date -d "$1" +%d` week2=`date -d "$1" +%a` } # 判断输入的参数个数是否正确 cnt=$# if [ $cnt -eq 0 ]; then echo "you must input at least one date, formate like 2019-5-1." exit 1 elif [ $cnt -eq 1 ]; then check_date $1 get_begin_time $1 get_end_time $1 elif [ $cnt -eq 2 ]; then check_date $1 check_date $2 get_begin_time $1 get_end_time $2 elif [ $cnt -gt 2 ]; then echo "you input too much arguments, at most two arguments allowed." exit 1 fi # 查找alert日志所在的路径 sqlplus -s /nolog &> /dev/null << eof set feedback off heading off verify off trimspool on timing off set pagesize 0 linesize 300 conn / as sysdba; spool /tmp/tmpdir.txt select value from v\$parameter where name=‘background_dump_dest‘; spool off exit; eof errs=`grep ‘ERROR‘ /tmp/tmpdir.txt | wc -l` if [ $errs -gt 0 ]; then echo "query alert log direction run error, please check the /tmp/tmpdir.txt for details." exit 1 else dir=`cat /tmp/tmpdir.txt` fi # 得到日志开始和结束的行数 row1=`find $dir/alert_$ORACLE_SID.log | xargs grep -n "$week1 $month1 $day1" | grep "$year1" | head -n 1 | cut -d ":" -f 1` if [ "$row1" == "" ]; then echo "$1 is not found in alert log" exit 1 fi row2=`find $dir/alert_$ORACLE_SID.log | xargs grep -n "$week2 $month2 $day2" | grep "$year2" | tail -n 1 | cut -d ":" -f 1` if [ "$row2" == "" ]; then echo "$2 is not found in alert log" exit 1 fi row2=$(($row2+10)) # 截取日志 if [ $cnt == 1 ]; then sed -n "${row1},${row2}p" $dir/alert_$ORACLE_SID.log > ./alert_$ORACLE_SID.$1.log echo "alert_$ORACLE_SID.$1.log has created!" elif [ $cnt == 2 ]; then sed -n "${row1},${row2}p" $dir/alert_$ORACLE_SID.log > ./alert_$ORACLE_SID.$1_to_$2.log echo "alert_$ORACLE_SID.$1_to_$2.log has created!" fi exit 0
使用说明:
1.使用oracle用户执行,脚本会自动寻找alert日志的位置
2.如果只查一天的日志,如2019-5-1这天的日志,则使用./script.sh 2019-5-1
3.如果查一段时间范围的日志,如2019-5-1到2019-5-10这10天的日志,则使用./script.sh 2019-5-1 2019-5-10
4.执行完成之后会在当前目录生成一个新文件
5.输入的日期必须在alert日志中存在才会有新文件生成,否则会报错
原文地址:https://www.cnblogs.com/ddzj01/p/10851969.html
时间: 2024-11-19 23:19:07