昨天中秋节，本该是团圆的好日子，苦逼的运维我还要值班（哈哈，吐槽一下）本以为会没有啥事，谁知道比较重要的一台Oracle服务器突然报警，CPU 2个core都飙到100%，load average也比较高，如下图：

为襄城等地区用户提供了全套网页设计制作服务，及襄城网站建设行业解决方案。主营业务为网站设计、成都做网站、襄城网站设计，以传统方式定制建设网站，并提供域名空间备案等一条龙服务，秉承以专业、用心的态度为用户提供真诚的服务。我们深信只要达到每一位用户的要求，就会得到认可，从而选择与我们长期合作。这样，我们也可以走得更远！

AWS CloudWatch也可以看出来CPU长期使用率100%

从图可得：系统us比较高，sy基本可以忽略，Memory和IO都已经检查过，不存在瓶颈，根据以往经验，极有可能是Oracle数据库有SQL在长时间运行，并且没有释放，登录到数据库查看，可以看到sid为410，408，404进程执行的都是同一个SQL,

SYS@xxxxxx>SELECT b.sid oracleID,
       b.username,
       b.serial#,
       spid,
       paddr,
       b.machine,
       c.sql_text
FROM v$process a, v$session b, v$sqlarea c
WHERE a.addr = b.paddr AND b.sql_hash_value = c.hash_value;
 
  ORACLEID USERNAME     SERIAL# SPID      PADDR       MACHINE
---------- ------------------------------ ---------- ------------------------ ---------------- ----------------------------------------------------------------
SQL_TEXT
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
       410 PRERNAP2 371 16743      00000002DEC84E60 Prernap2-mbr
with cte as(select distinct weekdate, week_month, week_year, SHIPTO, BUYERPARTNUMBER, dense_rank() over(partition by week_year,SHIPTO, BUYERPARTNUMBER, week_month order by weekdate) as rank_week FROM
(Select weekdate, to_char(weekdate, 'MM') as week_month,  to_char(weekdate, 'YYYY') as week_year, SHIPTO, BUYERPARTNUMBER from AUTO.ford_forecast_details  ) order by weekdate),  cte1 as (select foreca
stdate,forecast_month, forecast_year, shipto, buyerpartnumber from (select to_date(forecastdate, 'YYYYMMDD') as forecastdate, TO_CHAR(to_date(forecastdate, 'YYYYMMDD'), 'MM' ) as forecast_month,  TO_C
HAR(to_date(forecastdate, 'YYYYMMDD'), 'YYYY' ) as forecast_year,SHIPTO, BUYERPARTNUMBER from AUTO.ford_forecast_details  ) order by forecastdate ),  cte2 as (select cte.shipto, cte.buyerpartnumber,ct
e.week_month, cte.week_year, max(rank_week) as max_week  from cte group by cte.shipto, cte.buyerpartnumber,cte.week_month, cte.week_year),  cte4 as (select cte2.*, cte1.forecast_month, cte1.forecast_y
       408 PRERNAP21163 15129      00000002DEC916A0 Prernap2-mbr
with cte as(select distinct weekdate, week_month, week_year, SHIPTO, BUYERPARTNUMBER, dense_rank() over(partition by week_year,SHIPTO, BUYERPARTNUMBER, week_month order by weekdate) as rank_week FROM
(Select weekdate, to_char(weekdate, 'MM') as week_month,  to_char(weekdate, 'YYYY') as week_year, SHIPTO, BUYERPARTNUMBER from AUTO.ford_forecast_details  ) order by weekdate),  cte1 as (select foreca
stdate,forecast_month, forecast_year, shipto, buyerpartnumber from (select to_date(forecastdate, 'YYYYMMDD') as forecastdate, TO_CHAR(to_date(forecastdate, 'YYYYMMDD'), 'MM' ) as forecast_month,  TO_C
HAR(to_date(forecastdate, 'YYYYMMDD'), 'YYYY' ) as forecast_year,SHIPTO, BUYERPARTNUMBER from AUTO.ford_forecast_details  ) order by forecastdate ),  cte2 as (select cte.shipto, cte.buyerpartnumber,ct
e.week_month, cte.week_year, max(rank_week) as max_week  from cte group by cte.shipto, cte.buyerpartnumber,cte.week_month, cte.week_year),  cte4 as (select cte2.*, cte1.forecast_month, cte1.forecast_y
18 PRERNAP2 311 19710      00000002DEC948B0 Prernap2-mbr
with cte as(select distinct weekdate, week_month, week_year, SHIPTO, BUYERPARTNUMBER, dense_rank() over(partition by week_year,SHIPTO, BUYERPARTNUMBER, week_month order by weekdate) as rank_week FROM
(Select weekdate, to_char(weekdate, 'MM') as week_month,  to_char(weekdate, 'YYYY') as week_year, SHIPTO, BUYERPARTNUMBER from AUTO.ford_forecast_details  ) order by weekdate),  cte1 as (select foreca
stdate,forecast_month, forecast_year, shipto, buyerpartnumber from (select to_date(forecastdate, 'YYYYMMDD') as forecastdate, TO_CHAR(to_date(forecastdate, 'YYYYMMDD'), 'MM' ) as forecast_month,  TO_C
HAR(to_date(forecastdate, 'YYYYMMDD'), 'YYYY' ) as forecast_year,SHIPTO, BUYERPARTNUMBER from AUTO.ford_forecast_details  ) order by forecastdate ),  cte2 as (select cte.shipto, cte.buyerpartnumber,ct
e.week_month, cte.week_year, max(rank_week) as max_week  from cte group by cte.shipto, cte.buyerpartnumber,cte.week_month, cte.week_year),  cte4 as (select cte2.*, cte1.forecast_month, cte1.forecast_y
       404 PRERNAP2 665 21911      00000002DEC95960 Prernap2-mbr
with cte as( select distinct weekdate, week_month, week_year, SHIPTO, BUYERPARTNUMBER, dense_rank() over(partition by week_year,SHIPTO, BUYERPARTNUMBER, week_month order by weekdate) as rank_week FROM
 (Select weekdate, to_char(weekdate, 'MM') as week_month,  to_char(weekdate, 'YYYY') as week_year, SHIPTO, BUYERPARTNUMBER from AUTO.ford_forecast_details  ) order by weekdate ),  cte1 as ( select for
ecastdate,forecast_month, forecast_year, shipto, buyerpartnumber from (select to_date(forecastdate, 'YYYYMMDD') as forecastdate, TO_CHAR(to_date(forecastdate, 'YYYYMMDD'), 'MM' ) as forecast_month,  T
O_CHAR(to_date(forecastdate, 'YYYYMMDD'), 'YYYY' ) as forecast_year,SHIPTO, BUYERPARTNUMBER from AUTO.ford_forecast_details  ) order by forecastdate ),  cte2 as ( select cte.shipto, cte.buyerpartnumbe
r,cte.week_month, cte.week_year, max(rank_week) as max_week ,  cte1.forecast_month, cte1.forecast_year from cte inner join cte1 on cte1.forecast_year>=cte.week_year  and cte.shipto = cte1.shipto and c
22 SYS 447 23888      00000002DEC96A10 ec2-admart-01
SELECT b.sid oracleID,      b.username, b.serial#,   spid,paddr,      b.machine,c.sql_text FROM v$process a, v$session b, v$sqlarea c WHERE a.addr = b.paddrAND b.sq
l_hash_value = c.hash_value
       387 PRERNAP2 313 24261      00000002DEC97AC0 Prernap2-mbr
with cte as( select distinct weekdate, week_month, week_year, SHIPTO, BUYERPARTNUMBER, dense_rank() over(partition by week_year,SHIPTO, BUYERPARTNUMBER, week_month order by weekdate) as rank_week FROM
 (Select weekdate, to_char(weekdate, 'MM') as week_month,  to_char(weekdate, 'YYYY') as week_year, SHIPTO, BUYERPARTNUMBER from AUTO.ford_forecast_details  ) order by weekdate ),  cte1 as ( select for
ecastdate,forecast_month, forecast_year, shipto, buyerpartnumber from (select to_date(forecastdate, 'YYYYMMDD') as forecastdate, TO_CHAR(to_date(forecastdate, 'YYYYMMDD'), 'MM' ) as forecast_month,  T
O_CHAR(to_date(forecastdate, 'YYYYMMDD'), 'YYYY' ) as forecast_year,SHIPTO, BUYERPARTNUMBER from AUTO.ford_forecast_details  ) order by forecastdate ),  cte2 as ( select cte.shipto, cte.buyerpartnumbe
r,cte.week_month, cte.week_year, max(rank_week) as max_week ,  cte1.forecast_month, cte1.forecast_year from cte inner join cte1 on cte1.forecast_year>=cte.week_year  and cte.shipto = cte1.shipto and c
6 rows selected.


SYS@xxxxxx>select b.sid,b.serial#,b.machine,b.terminal,b.program,b.process,b.status
from v$lock a, v$session b where a.SID = b.SID and username is not null and username not in ('SYS','SYSTEM');

       SID    SERIAL# MACHINE							       TERMINAL 		      PROGRAM					       PROCESS			STATUS
---------- ---------- ---------------------------------------------------------------- ------------------------------ ------------------------------------------------ ------------------------ --------
       387	  313 Prernap2-mbr						       unknown			      SQL Developer				       6246			ACTIVE
       387	  313 Prernap2-mbr						       unknown			      SQL Developer				       6246			ACTIVE
       404	  665 Prernap2-mbr						       unknown			      SQL Developer				       4145			ACTIVE
       387	  313 Prernap2-mbr						       unknown			      SQL Developer				       6246			ACTIVE
       387	  313 Prernap2-mbr						       unknown			      SQL Developer				       6246			ACTIVE
       408	 1163 Prernap2-mbr						       unknown			      SQL Developer				       3377			ACTIVE
       387	  313 Prernap2-mbr						       unknown			      SQL Developer				       6246			ACTIVE
       387	  313 Prernap2-mbr						       unknown			      SQL Developer				       6246			ACTIVE
       404	  665 Prernap2-mbr						       unknown			      SQL Developer				       4145			ACTIVE
       408	 1163 Prernap2-mbr						       unknown			      SQL Developer				       3377			ACTIVE
       410	  371 Prernap2-mbr						       unknown			      SQL Developer				       5691			ACTIVE
	18	  311 Prernap2-mbr						       unknown			      SQL Developer				       1497			ACTIVE
	20	  221 Prernap2-mbr						       unknown			      SQL Developer				       4689			ACTIVE
	20	  221 Prernap2-mbr						       unknown			      SQL Developer				       4689			ACTIVE
       387	  313 Prernap2-mbr						       unknown			      SQL Developer				       6246			ACTIVE

15 rows selected.

SYS@xxxxxx>select sid, username, blocking_session from v$session where blocking_session is not null;

       SID USERNAME			  BLOCKING_SESSION
---------- ------------------------------ ----------------
	18 PRERNAP2				       408
       387 PRERNAP2				       404
       410 PRERNAP2				       408

SYS@xxxxxx>select sid, serial#, username from v$session where sid='410';

       SID    SERIAL# USERNAME
---------- ---------- ------------------------------
       410	  371 PRERNAP2

解决方法

找到开发人员，询问原因，得到的反馈是在测试几条SQL（我擦，竟然在生产环境测试SQL,哎，一点敬畏之心都没有，可怕！）

kill掉blocked的进程，释放资源，再这么跑下去，系统随时可能崩溃，最后去优化一下的SQL,再去执行

alter system kill session '410,371';

......其他几个进程同理干掉即可

当前文章：Oracle%Cpu100us
转载源于：http://kswjz.com/article/jpidhj.html