Linux下Oracle进程CPU使用率过高处理方式
一、构建一条比较耗费性能的SQL语句:SQL> set linesize 200SQL> select count(*) from all_objects a,all_objects b,all_objects c group by a.object_id;二、登陆Linux,使用top工具查看进程信息top - 14:39:46 up 9 min,3 users...
·
一、构建一条比较耗费性能的SQL语句:
SQL> set linesize 200
SQL> select count(*) from all_objects a,all_objects b,all_objects c group by a.object_id;
二、登陆Linux,使用top工具查看进程信息
top - 14:39:46 up 9 min, 3 users, load average: 0.41, 0.51, 0.32
Tasks: 381 total, 1 running, 380 sleeping, 0 stopped, 0 zombie
Cpu(s): 3.5%us, 2.2%sy, 0.0%ni, 64.5%id, 28.8%wa, 0.1%hi, 0.9%si, 0.0%st
Mem: 2941544k total, 2799796k used, 141748k free, 47072k buffers
Swap: 3145724k total, 0k used, 3145724k free, 1024096k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
5241 oracle 20 0 1502m 132m 49m S 8.6 4.6 0:04.28 oracle
2023 root 0 -20 0 0 0 S 4.6 0.0 0:05.85 iscsi_q_33
4816 oracle -2 0 1403m 16m 14m S 3.0 0.6 0:08.61 oracle
3865 root RT -5 693m 154m 63m S 2.0 5.4 0:10.37 ologgerd
4129 grid -2 0 1306m 15m 12m S 1.7 0.5 0:09.79 oracle
3774 root 20 0 1567m 38m 16m S 1.0 1.3 0:02.31 orarootagent.bi
4191 root 20 0 2205m 66m 26m S 1.0 2.3 0:07.08 crsd.bin
3393 root 20 0 1850m 55m 22m S 0.7 1.9 0:04.81 ohasd.bin
3740 grid 20 0 1836m 41m 16m S 0.7 1.5 0:02.40 oraagent.bin
3776 grid 20 0 559m 39m 12m S 0.7 1.4 0:02.96 gipcd.bin
3828 grid RT 0 1612m 123m 54m S 0.7 4.3 0:05.42 ocssd.bin
4317 root 20 0 962m 28m 14m S 0.7 1.0 0:03.49 orarootagent.bi
4 root 20 0 0 0 0 S 0.3 0.0 0:00.04 ksoftirqd/0
3928 root 20 0 751m 25m 10m S 0.3 0.9 0:01.99 octssd.bin
3959 grid 20 0 1093m 28m 13m S 0.3 1.0 0:02.04 evmd.bin
4141 grid 20 0 1314m 26m 17m S 0.3 0.9 0:01.03 oracle
4143 grid 20 0 1322m 30m 16m S 0.3 1.1 0:00.99 oracle
4145 grid -2 0 1321m 31m 17m S 0.3 1.1 0:01.87 oracle
4305 grid 20 0 565m 19m 10m S 0.3 0.7 0:00.31 scriptagent.bin
4479 grid 20 0 2755m 162m 16m S 0.3 5.6 0:11.50 java
4744 oracle 20 0 1103m 33m 16m S 0.3 1.2 0:02.07 oraagent.bin
4832 oracle 20 0 1418m 31m 21m S 0.3 1.1 0:00.83 oracle
4836 oracle -2 0 1418m 49m 36m S 0.3 1.7 0:02.26 oracle
[root@or11g1 ~]#
从中可以看出,进程号为5241的oracle进程占用CPU资源是最多的
三、查看5241这个oracle进程对应的会话信息
SELECT sql_text FROM v$sqltext a WHERE (a.hash_value, a.address) IN
(SELECT DECODE(sql_hash_value, 0, prev_hash_value, sql_hash_value),DECODE(sql_hash_value, 0, prev_sql_addr, sql_address) FROM
v$session b WHERE b.paddr =(SELECT addr FROM v$process c WHERE c.spid = '5241')) ORDER BY piece ASC;
或者:
SELECT sql_text FROM v$sqltext a WHERE (a.hash_value, a.address)
IN (SELECT DECODE(sql_hash_value, 0, prev_hash_value, sql_hash_value),DECODE(sql_hash_value, 0, prev_sql_addr, sql_address)
FROM v$session b WHERE b.paddr =(SELECT addr FROM v$process c WHERE c.spid = '&pid')) ORDER BY piece ASC;
查询结果如下:
SQL> SELECT sql_text FROM v$sqltext a WHERE (a.hash_value, a.address) IN
(SELECT DECODE(sql_hash_value, 0, prev_hash_value, sql_hash_value),DECODE(sql_hash_value, 0, prev_sql_addr, sql_address) FROM
v$session b WHERE b.paddr =(SELECT addr FROM v$process c WHERE c.spid = '5241')) ORDER BY piece ASC; 2 3
SQL_TEXT
----------------------------------------------------------------
select count(*) from all_objects a,all_objects b,all_objects c g
roup by a.object_id
SQL>
结果显示SQL语句就是我们之前构建的SQL。
在执行如下SQL语句,找出相应的会话信息:
select sid,serial# ,username,osuser,machine,program,process,to_char(logon_time,'yyyy-mm-dd hh24:mi:ss') logon from
v$session where paddr in ( select addr from v$process where spid in('5241'));
或者:
select id,serial# ,username,osuser,machine,program,process,to_char(logon_time,'yyyy-mm-dd hh24:mi:ss') logon
from v$session where paddr in ( select addr from v$process where spid in('&pid'));
执行的结果如下:
到这里,就可以根据SID和SERIAL#来杀掉用户会话了。
alter system kill session '76,21' immediate;
SQL> alter system kill session '76,21' immediate;
System altered.
SQL>
但是,这要是通过PMON进程来定期清除会话的,因此会有一定的时间间隔,如果情况很紧急,需要立即释放资源,那么就要杀操作系统进程了。
四、找出操作系统进程
select PADDR from v$session where SID=76 and SERIAL#=21;
找出拥有会话的进程地址
然后通过PAAD号找出系统进程号
然后在操作系统上kill -9 5241即可。
更多推荐
已为社区贡献1条内容
所有评论(0)