线上Oracle准备实现类似MySQL slow query的监控脚本，把查询时间超出定值的SQL定时的发送邮件告警，实现过程记录如下：

主要思路是通过DBA_HIST的几个视图来获取每小时快照中慢SQL的情况，为了不影响线上环境，这里把脚本部署在了自己的监控端，通过DBLINK定期的抓取线上生产库的数据到监控数据库，并简单的处理后获得csv格式的报表，发送报表至邮箱。

定时脚本每小时查询一次

00 * * * * /opt/scripts/oracle/get_slow_query.sh

脚本内容如下

[oracle@59-Mysql-Test ~]$ cat /opt/scripts/oracle/get_slow_query.sh

#!/bin/bash

errlog="/opt/scripts/oracle/sqlerror.log"

sq_data="/opt/scripts/oracle/slow_query_data.xls"

check_file="/opt/scripts/oracle/slowsql_check.log"

send_mail_check="/opt/scripts/oracle/send_mail.chk"

export ORACLE_BASE=/u01/app/oracle

export ORACLE_HOME=/u01/app/oracle/product/11.2.0/db_1

export ORACLE_SID=oramon

export PATH=/usr/sbin:$PATH

export PATH=/u01/app/oracle/product/11.2.0/db_1/bin:$PATH

export LD_LIBRARY_PATH=$ORACLE_HOME/lib:/lib:/usr/lib

export CLASSPATH=/u01/app/oracle/product/11.2.0/db_1/JRE:/u01/app/oracle/product/11.2.0/db_1/jlib:/u01/app/oracle/product/11.2.0/db_1/rdbms/jlib

cd /opt/scripts/oracle/

$ORACLE_HOME/bin/sqlplus -S sqmon/oracle @main > ${errlog}

cat ${errlog} | grep -v 'Call completed.' | grep -v '' > ${check_file}

[ -s ${check_file} ] && /bin/mail -s "Oracle slow query check error" xxx@xxx.com < ${check_file}

cat ${sq_data} | grep -v '<' >${send_mail_check}

[ -s ${send_mail_check} ] && /bin/mail -a ${sq_data} -s "OracleDB find slow query,please check" xxx@xxx.com,xxx@xxx.com

[oracle@59-Mysql-Test oracle]$ cat main.sql

call pro_get_slow_query();

set linesize 5000

set term off verify off feedback off pagesize 999

set markup html on entmap ON spool on preformat off

spool slow_query_data.xls

@get_tables.sql

spool off

exit

[oracle@59-Mysql-Test oracle]$ cat get_tables.sql

select sql_id,elapsed_time,cpu_time,iowait_time,gets,reads,rws,clwait_time,execs,elpe,machine,username,dbms_lob.substr(sqt,4000) from DBA_ORA_SLOW_QUERY where elpe > 10 and machine not in ('rac01','rac02');

存储过程pro_get_slow_query内容如下

CREATE OR REPLACE PROCEDURE SQMON.pro_get_slow_query

AS

BEGIN

/**********delete old data on sqltext*************/

delete from local_dba_hist_sqltextas;

commit;

insert into local_dba_hist_sqltextas select * from dba_hist_sqltext@dg2;

commit;

insert into DBA_ORA_SLOW_QUERY_HISTORY select a.*,sysdate from DBA_ORA_SLOW_QUERY;

commit;

delete from DBA_ORA_SLOW_QUERY;

commit;

/*

select * from DBA_ORA_SLOW_QUERY;

select * from DBA_ORA_SLOW_QUERY_HISTORY;

*/

/************insert new date ********************/

insert into DBA_ORA_SLOW_QUERY

select v_1.sql_id,

v_1.elapsed_time,

v_1.cpu_time,

v_1.iowait_time,

v_1.gets,

v_1.reads,

v_1.rws,

v_1.clwait_time,

v_1.execs,

v_1.elpe,

v_2.machine,

v_2.username,

v_1.sqt

from (select s.sql_id,

elapsed_time / 1000000 elapsed_time,

cpu_time / 1000000 cpu_time,

iowait_time / 1000000 iowait_time,

gets,

reads,

rws,

clwait_time / 1000000 clwait_time,

execs,

st.sql_text sqt,

elapsed_time / 1000000 / decode(execs, 0, null, execs) elpe

from (select *

from (select sql_id,

sum(executions_delta) execs,

sum(buffer_gets_delta) gets,

sum(disk_reads_delta) reads,

sum(rows_processed_delta) rws,

sum(cpu_time_delta) cpu_time,

sum(elapsed_time_delta) elapsed_time,

sum(clwait_delta) clwait_time,

sum(iowait_delta) iowait_time

from dba_hist_sqlstat@HUBSDG2

where snap_id >=

(select max(snap_id) - 1

from dba_hist_snapshot@DG2)

and snap_id <=

(select max(snap_id)

from dba_hist_snapshot@DG2)

group by sql_id

order by sum(elapsed_time_delta) desc)

where rownum <= 20) s,

local_dba_hist_sqltextas st

where st.sql_id = s.sql_id) v_1

left join (select distinct a.sql_id, a.machine, b.username

from dba_hist_active_sess_history@DG2 a

left join dba_users@DG2 b

on a.user_id = b.user_id

where a.snap_id >=

(select max(snap_id) - 1 from dba_hist_snapshot@DG2)

and a.snap_id <=

(select max(snap_id) from dba_hist_snapshot@DG2)) v_2

on v_1.sql_id = v_2.sql_id

order by elpe desc;

commit;

END;

/

上面标绿加粗部分的表不再列出，可以直接通过CTAS格式去创建即可。

Oracle慢SQL监控脚本实现