Oracle kill session相关问题（上）

今天同事杀会话的时候采用了alter system kill session ‘xxx,xxx’的方式，结果杀完后，v$session中还可以查到，就求助我，因为我之前杀会话都是带的immediate的，杀完后会立刻释放，v$session中也查询不到，同事现在的情况就只能杀后台进程了，但paddr列关联不到后台进程，查了下MOS还是给出了一些办法，整理了一下，分享给大家。

一.3 相关知识点扫盲(摘自网络+个人总结)

一.3.1 得到当前会话的几个SQL

SELECT USERENV('SID') FROM DUAL;

SELECT SID FROM V$MYSTAT WHERE ROWNUM =1;

SELECT b.SID,

b.SERIAL#

FROM v$session b

WHERE b.SID = USERENV('SID');

SELECT a.SID,

b.SERIAL#

FROM v$mystat a,

v$session b

WHERE a.SID = b.SID

AND rownum = 1;

SELECT a.SID,

b.SERIAL# ,

c.SPID,

b.status

FROM v$mystat a,

v$session b ,

v$process c

WHERE a.SID = b.SID

and b.PADDR=c.ADDR

AND rownum = 1;

SELECT b.SID,

b.SERIAL# ,

c.SPID,

b.status

FROM v$session b ,

v$process c

WHERE b.PADDR=c.ADDR

AND b.sid=???;

一.3.2 Session 状态说明

Oracle session 有如下几种状态:

ACTIVE - Session currently executing SQL

INACTIVE

KILLED - Session marked to be killed

CACHED - Session temporarily cached for use by Oracle*XA

SNIPED - Session inactive, waiting on the client

有关状态的说明：

（1）active 处于此状态的会话，表示正在执行，处于活动状态。

官方文档说明：

Any session that is connected to the database and is waiting for an event that does not belong to the Idle wait class is considered as an active session.

（2）killed处于此状态的会话，被标注为删除，表示出现了错误，正在回滚。

当然，也是占用系统资源的。还有一点就是，killed的状态一般会持续较长时间，而且用windows下的工具pl/sql developer来kill掉，是不管用的，要用命令：alter system kill session 'sid,serial#' ;

（3）inactive 处于此状态的会话表示不是正在执行的

该状态处于等待操作(即等待需要执行的SQL语句)，通常当DML语句已经完成。但连接没有释放，这个可能是程序中没有释放，如果是使用中间件来连接的话，也可能是中间件的配置或者是bug 导致。inactive对数据库本身没有什么影响，但是如果程序没有及时commit，那么就会造成占用过多会话。容易是DB 的session 达到极限值。

一般不处理inactive 状态的session，如果达到了session 的最大值，就增加processes 和 sessions 参数。对于Inactive 状态的session，可以设置过期时间：

（1）sqlnet.ora文件中设置expire_time 参数

（2）设置用户profile的idle_time 参数

（3）找到很久没有响应的会话然后kill掉

当设置了resource_limit=true 。通过idle_time限制session idle 时间。session idle超过设置时间，状态为sniped (v$session).,然而OS下的process并不会释放，当session(user process) 再次与server process 通讯，将关闭相应的server process.

sqlnet.expire_time 的原理不一样，Oracle Server 发送包探测dead connection ，如果连接关闭，或者不再用，则关闭相应的server process.

以上两者组合使用，减少server process,防止process超过init$ORACLE_SID极限值。

本文会对这3种方式分别说明的。

一.3.2.1 清理inactive会话的3种方式

一、设置sqlnet.expire_time

可以在sqlnet.ora文件里面加上sqlnet.expire_time这个参数来解决，设置一个分钟数，这是ORACLE建议的DCD解决方法。

在 sqlnet.ora文件中设置expire_time 参数

官网有关这个参数的说明：

http://download.oracle.com/docs/cd/B19306_01/network.102/b14213/sqlnet.htm

Oracle? Database Net Services Reference 11g Release 2 (11.2)E10835-10

SQLNET.EXPIRE_TIME

Purpose

Use parameter SQLNET.EXPIRE_TIME to specify a the time interval, in minutes, to send a probe to verify that client/server connections are active. Setting a value greater than 0 ensures that connections are not left open indefinitely, due to an abnormal client termination. If the probe finds a terminated connection, or a connection that is no longer in use, it returns an error, causing the server process to exit. This parameter is primarily intended for the database server, which typically handles multiple connections at any one time.

sqlnet.expire_time 的原理：Oracle Server 发送包探测dead connection ，如果连接关闭，或者不再用，则关闭相应的server process.

Limitations on using this terminated connection detection feature are:

（1）It is not allowed on bequeathed connections.

（2）Though very small, a probe packet generates additional traffic that may downgrade network performance.

（3）Depending on which operating system is in use, the server may need to perform additional processing to distinguish the connection probing event from other events that occur. This can also result in degraded network performance.

Default ：0

Minimum Value ：0

Recommended Value ：10

Example

SQLNET.EXPIRE_TIME=10

二、设置用户profile的idle_time 参数

Oracle 用户 profile 属性

http://blog.csdn.net/tianlesoftware/archive/2011/03/10/6238279.aspx

注意，要启用idle_time 要先启用RESOURCE_LIMIT参数。该参数默认是False。官网说明如下：

RESOURCE_LIMIT

Property	Description
Parameter type	Boolean
Default value	false
Modifiable	ALTER SYSTEM
Range of values	true \| false

RESOURCE_LIMIT determines whether resource limits are enforced in database profiles.

Values:

TRUE：Enables the enforcement of resource limits

FALSE：Disables the enforcement of resource limits

三、找到很久没有响应的会话然后kill掉

关于v$session中LAST_CALL_ET列的理解：http://blog.itpub.net/26736162/viewspace-1762403/

根据v$session中LAST_CALL_ET列的意义我们可以写出如下的SQL脚本，没一个小时清理10个小时没有响应的会话：

set sqlblanklines on

CREATE OR REPLACE PROCEDURE P_kill_session_LHR AS

-----------------------------------------------------------------------------------

-- Created on 2013-06-25 12:05:07 by lhr

--Changed on 2015-08-05 12:05:07 by lhr

-- function：杀掉10个小时之前的会话 ,告警日志中会记录被杀掉的会话信息

-----------------------------------------------------------------------------------

BEGIN

-- IF to_char(SYSDATE, 'HH24') >= '20' OR

-- TO_CHAR(SYSDATE, 'HH24') <= '08' THEN

FOR cur IN (SELECT A.USERNAME,

A.LOGON_TIME,

A.STATUS,

A.SID,

A.SERIAL#,

A.MACHINE,

A.OSUSER,

'ALTER SYSTEM DISCONNECT SESSION ''' || a.SID || ',' ||

a.serial# || ''' IMMEDIATE' kill_session

FROM v$session A

WHERE A.STATUS IN ('INACTIVE')

AND A.USERNAME IS NOT NULL

AND A.LAST_CALL_ET >= 60 * 60 * 10) LOOP

BEGIN

EXECUTE IMMEDIATE cur.kill_session;

EXCEPTION

WHEN OTHERS THEN

NULL;

END;

END LOOP;

-- END IF;

EXCEPTION

WHEN OTHERS THEN

NULL;

END P_kill_session_LHR;

BEGIN

--DBMS_SCHEDULER.drop_job('JOB_P_kill_session_LHR');

DBMS_SCHEDULER.CREATE_JOB(JOB_NAME => 'JOB_P_kill_session_LHR',

JOB_TYPE => 'STORED_PROCEDURE',

JOB_ACTION => 'P_kill_session_LHR',

repeat_interval => 'FREQ=MINUTELY;INTERVAL=60',

ENABLED => TRUE,

START_DATE => SYSDATE,

COMMENTS => '删除--每60分钟检查一次');

END;

SELECT d.job_name,d.job_action,d.run_count FROM dba_scheduler_jobs d WHERE d.job_name='JOB_P_KILL_SESSION_LHR';

SQL> col JOB_ACTION format a20
SQL> SELECT d.job_name,d.job_action,d.run_count FROM dba_scheduler_jobs d WHERE d.job_name='JOB_P_KILL_SESSION_LHR';

JOB_NAME JOB_ACTION RUN_COUNT
------------------------------ -------------------- ----------
JOB_P_KILL_SESSION_LHR P_kill_session_LHR 4

运行日志：

SELECT * FROM dba_scheduler_job_run_details d WHERE d.job_name='JOB_P_KILL_SESSION_LHR';

一.3.3 oracle死连接(DC)和不活动会话(INACTIVE)

This note explains the difference between a dead connection and an INACTIVE session in v$session. It also discusses the mechanisms provided to automate the cleanup of each.

这里解释死连接和不活动会话的区别，也会讨论自动清除的机制。

Difference between INACTIVE sessions and Dead Connections

---------------------------------------------------------

Dead connections and INACTIVE sessions are different issues. Oracle provides separate mechanisms to automate the cleanup of each.

死连接和不活动会话的不同问题，oracle提供了分离的机制去自动清理。

(1) Dead connections:死连接

These are previously valid connections with the database but the connection between the client and server processes has terminated abnormally.

有些之前合法的连接，但是由于客户端和服务器进程的异常中断。

Examples of a dead connection:

- A user reboots/turns-off their machine without logging off or disconnecting from the database.

- A network problem prevents communication between the client and the server.

1.用户没有注销就关闭机器。

2.网络在客户端和服务器端终端连接。

In these cases, the shadow process running on the server and the session in the database may not terminate. To automate the cleanup of these sessions, you can use the Dead Connection Detection (DCD) feature of Net8.

这种情况下，后台进程跑在服务器端，而会话在数据库端不会中断。

When DCD is enabled, Net8 (server-side) sends a packet to the client. If the client is active, the packet is discarded. If the client has terminated, the server will receive an error and Net8 (server-side) will end that session.

当死连接启动，NET8(服务端)会发送一个包到客户端。如果客户端是活动的，这个包就被丢掉。如果客户端已经被中断，服务器端将接收一个错误，将会中断该会话。SQLNET.EXPIRE_TIME.

Refer to Note:151972.1: Dead Connection Detection (DCD) Explained, for details regarding DCD.

(2) INACTIVE Sessions:

These are sessions that remain connected to the database with a status in v$session of INACTIVE.

会话与服务器端保持连接，但是状态为inactive.

Example of an INACTIVE session:

- A user starts a program/session, then leaves it running and idle for an extended period of time.

用户开始一个会话，运行一段时间后，保持相当一段时间的空闲。

To automate cleanup of INACTIVE sessions you can create a profile with an appropriate IDLE_TIME setting and assign that profile to the users.

自动清理不活动的会话，你可以创一个profile.然后设置恰当的IDLE_TIME,分配给指定用户。

Note:159978.1: How To Automate Disconnection of Idle Sessions, outlines the steps to setup IDLE_TIME for this.

一.3.4 kill session相关内容

kill session 是DBA经常碰到的事情之一。如果kill 掉了不该kill 的session，则具有破坏性，因此尽可能的避免这样的错误发生。同时也应当注意，如果kill 的session属于Oracle 后台进程，则容易导致数据库实例宕机。

一、获得需要kill session的信息

SET LINESIZE 180

COLUMN spid FORMAT A10

COLUMN username FORMAT A10

COLUMN program FORMAT A40

SELECT s.inst_id,

s.sid,

s.serial#,

p.spid,

s.username,

s.program,

s.paddr,

s.STATUS

FROM gv$session s

JOIN gv$process p

ON p.addr = s.paddr

AND p.inst_id = s.inst_id

WHERE s.type != 'BACKGROUND';

INST_ID SID SERIAL# SPID USERNAME PROGRAM PADDR STATUS

---------- ---------- ---------- ---------- ---------- ---------------------------------------- ---------------- --------

1 125 5 14029 SYS sqlplus@orcltest (TNS V1-V3) 0000000077C98660 INACTIVE

1 9 15 14274 SYS sqlplus@orcltest (TNS V1-V3) 0000000077C99710 INACTIVE

1 17 5 14078 LHR sqlplus.exe 0000000077CA5F50 INACTIVE

1 144 31 14645 SYS sqlplus@orcltest (TNS V1-V3) 0000000077CA7000 ACTIVE

1 20 7 14647 SYS plsqldev32.exe 0000000077CA80B0 INACTIVE

1 145 23 14651 SYS plsqldev32.exe 0000000077CA9160 INACTIVE

二、使用ALTER SYSTEM KILL SESSION 命令实现

语法：

SQL> ALTER SYSTEM KILL SESSION 'sid,serial#';

SQL> ALTER SYSTEM KILL SESSION 'sid,serial#' IMMEDIATE;

对于RAC环境下的kill session ,需要搞清楚需要kill 的session 位于哪个节点，可以查询GV$SESSION视图获得。11g杀掉集群环境下的某个会话：

alter system kill session'1228,42549,@实例号';

例如：alter system kill session '1228, 42549, @2'；

10g下应登录到某个特定的实例才可以。

Kill session 命令实际不会kill session，比如等待远程数据库的反应或者回滚事务，那么session 就不会立即kill其必须等待当前的操作结束才能执行，在这种情况下，session 就会被标记为killed 状态。

我们可以在kill 命令中添加immediate，语法如下：

SQL> ALTER SYSTEM KILL SESSION 'sid,serial#' IMMEDIATE;

这个命令不会影响性能，但它会立即返回到当前的session，处理kill操作，而不是等待其他的信息完成。如果session 一直处于killed 状态，那么可以考虑在操作系统级别kill掉相关的进程。不过在操作之前，要先确认session 是否在执行rollback 操作。可以使用如下SQL 来确认。

SET LINESIZE 200

COLUMN username FORMAT A15

SELECT s.username,

s.sid,

s.serial#,

t.used_ublk,

t.used_urec,

rs.segment_name,

r.rssize,

r.status

FROM v$transaction t,

v$session s,

v$rollstat r,

dba_rollback_segs rs

WHERE s.saddr = t.ses_addr

AND t.xidusn = r.usn

AND rs.segment_id = t.xidusn

ORDER BY t.used_ublk DESC;

如果有我们的session，那么就要等rollback 先完成，然后才能在操作系统级别kill session。

kill session 的时候仅仅是将会话杀掉。在有些时候，由于较大的事务或需要运行较长的SQL语句将导致需要kill的session并不能立即杀掉。对于这种情况将收到 "marked for kill"提示(如下)，一旦会话当前事务或操作完成，该会话被立即杀掉。

alter system kill session '4730,39171'

ERROR at line 1:

ORA-00031: session marked for kill

在下面的操作中将杀掉会话146，144

sys@AUSTIN> alter system kill session '146,23';

System altered.

sys@AUSTIN> alter system kill session '144,42';

System altered.

sys@AUSTIN> select inst_id,saddr,sid,serial#,paddr,username,status,program from gv$session where username is not null;

INST_ID SADDR SID SERIAL# PADDR USERNAME STATUS PROGRAM

---------- -------- ---------- ---------- -------- ---------- -------- ---------------------------------------------

1 4C70BF04 144 42 4C6545A0 SCOTT KILLED sqlplus@oracle10g (TNS V1-V3)

1 4C70E6B4 146 23 4C6545A0 TEST KILLED sqlplus@oracle10g (TNS V1-V3)

1 4C71FC84 160 17 4C624174 SYS ACTIVE sqlplus@oracle10g (TNS V1-V3)

SQL> select sid,serial#,server,status from v$session where sid=22;

SID SERIAL# SERVER STATUS

---------- ---------- --------- --------

22 7 PSEUDO KILLED

注意：在查询中可以看到被杀掉的会话的PADDR地址发生了变化，参照查询结果中的红色字体。如果多个session被kill 掉，则多个session的PADDR被改为相同的进程地址，被杀掉的会话的server列变为PSEUDO。

一.3.4.1 oracle killed会话不释放的问题

一般情况下，在杀一个会话的时候，直接执行alter system kill session ‘sid,serial#’;

Administrator's Guide说，当session是active的时候，alter system kill session 只是将session的状态标识为killed，server变为pseudo状态，并不会释放session持有的资源，所以我们在执行完alter system kill session 后，看会话还是一直存在。

这种情况下可以使用 immediate选项，强制立即Kill会话，如下：

SQL> alter system kill session '3964,51752' immediate;

SQL Language Reference（http://docs.oracle.com/cd/B28359_01/server.111/b28286/toc.htm#BEGIN）

里对Immediate的解释是：IMMEDIATE Specify IMMEDIATE to instruct Oracle

Database to roll back ongoing transactions, release all session locks, recover the entire session state,and return control to you immediately.

一.3.4.2 killed状态的会话如何找到spid列

一般情况下我们查询会话的后台进程是通过如下的SQL，即通过v$session的paddr列关联v$process的addr列，但是killed状态的v$session的paddr列都变成了一样的，所以已经没有办法通过如下的SQL去查询了。

SELECT b.SID,

b.SERIAL# ,

c.SPID,

b.status

FROM v$session b ,

v$process c

WHERE b.PADDR=c.ADDR

AND b.sid=???;

如果会话已经在v$sesion里是killed状态，那么我们通过上面的SQL已经查不出spid,可以用下面的SQL查出SPID：

-----方法1

select spid, program from v$process

where program!= 'PSEUDO'

and addr not in (select paddr from v$session)

and addr not in (select paddr from v$bgprocess)

and addr not in (select paddr from v$shared_server);

select INST_ID, spid, program,'kill -9 '|| spid kill9

from gv$process a

where program != 'PSEUDO'

and (INST_ID, addr) not in (select INST_ID, paddr from gv$session)

and (INST_ID, addr) not in (select INST_ID, paddr from gv$bgprocess)

and (INST_ID, addr) not in (select INST_ID, paddr from gv$shared_server)

and a.PNAME is null;

-----方法2 我自己常用的方法

set line 9999

col sessionid format a20

col sessionid_killed format a20

col kill_session format a60

SELECT a.INST_ID,

a.SID || ',' || a.SERIAL# || ',' ||

(select spid

from gv$process b

where b.INST_ID = a.INST_ID

and A.creator_addr = b.ADDR --and decode(a.status,'KILLED',A.creator_addr,A.PADDR) = b.ADDR

) sessionid,

a.PADDR,

a.STATUS,

a.PROGRAM,

'alter system disconnect session ''' || sid || ',' || serial# || ''' immediate;' kill_session

FROM gv$session a

WHERE a.USERNAME = 'SYS'

and a.STATUS = 'KILLED';

As a result of the bug, 2 additional columns have been added to V$SESSION from 11g on:

V$SESSION

CREATOR_ADDR - state object address of creating process

CREATOR_SERIAL# - serial number of creating process

CREATOR_ADDR is the column that can be joined with the ADDR column in V$PROCESS to uniquely identify the killed process corresponding to the former session.

Following the previous example, this would identify the killed session

-----方法3

SELECT a.SID || ',' || a.SERIAL# || ',' ||

(select spid

from gv$process b

where b.INST_ID = a.INST_ID

and A.pid = b.pid) sessionid,

'alter system kill session ''' || sid || ',' || serial# || ''';' kill_session

FROM gV$DETACHED_SESSION a;

-----方法4

SELECT INST_ID, spid, program, 'kill -9 ' || spid kill9

FROM gv$process a

WHERE (a.INST_ID, a.addr) in (select INST_ID, p.addr

from gv$process p

where pid <> 1

minus

select INST_ID, s.paddr

from gv$session s)

and a.PNAME is null;

-----方法4

SELECT s.SID, s.username,s.status,

x.ADDR,x.KSLLAPSC,x.KSLLAPSN,x.KSLLASPO,x.KSLLID1R,x.KSLLRTYP,

decode(bitand(x.ksuprflg,2),0,null,1)

FROM x$ksupr x,v$session s

WHERE s.paddr(+)=x.addr

and bitand(ksspaflg,1)!=0

;

一.3.4.3 为何killed状态的进程一直在v$session中能查询到？

[ID 100859.1]这篇文章中提到pmon如何清理killed的会话：

PMON will not delete the session object itself until the client connected to

that session notices that it has been killed. Therefore, the sequence of

events is:

1) alter system kill session is issued - the STATUS of the session object in

V$SESSION becomes KILLED, its server becomes PSEUDO.

2) PMON cleans up the *resources* allocated to the session

(i.e., rolls back its transaction, releases its locks, etc).

3) the entry in V$SESSION remains there until the client of that session (the

client is the process associated with the OSUSER,MACHINE,PROCESS columns in

the V$SESSION view) tries to do another request.

4) the client attempts another SQL statement and gets back ORA-28.

5) PMON can now remove the entry from V$SESSION.

This behavior is necessary because the client still has pointers to the

session object even though the session has been killed. Therefore, the

object cannot be deleted until the client is no longer pointing at it.

alter system kill session后，会话状态变为killed，pmon回收资源后，会话的信息仍然保留在v$session中，直到客户端再次请求，数据库返回ORA-28错误："your session has been killed"。

此时pmon才从v$session中移除这些会话的信息。而上述中，客户端机器重启，无法再向数据库发送请求，因此从v$session中一直可以查到，只有通过操作系统强制杀掉进程，才能触发pmon从v$session中清除。

然后Oracle就等待PMON去清除这些Session.所以通常等待一个被标记为Killed的Session退出需要花费很长的时间.如果此时被Kill的process，重新尝试执行任务，那么马上会收到进程中断的提示，process退出，此时Oracle会立即启动PMON来清除该session.这被作为一次异常中断处理.

一.3.4.4 为何kill session后，paddr变成同一个值了？

[ID 387077.1]这篇文章中提到：

Bug 5453737 WHEN A SESSION IS KILLED, PADDR CHANGES IN V$SESSION BUT ADDR NOT IN V$PROCESS closed as not a bug with the following explanation:

When a session is killed, the session state object(and all the child state objects under the session state object) move out from under the original parent process state object, and are placed under the pseudo process state object (which is expected, given the parent/child process mechanism on Unix). PMON will clean up all the state objects found under the pseudo process state object. That explains why PADDR changes in V$SESSION when a session is killed. New PADDR you are seeing in v$SESSION is the address of the pseudo process state object. This shows up in system state under PSEUDO PROCESS for group DEFAULT: V$PROCESS still maintains the record of the original parent process. This is expected.

使用以下sql查找killed状态的spid：

select spid, program from v$process

where program!= 'PSEUDO'

and addr not in (select paddr from v$session)

and addr not in (select paddr from v$bgprocess)

and addr not in (select paddr from v$shared_server);

通过底层表x$ksupr关联v$session也可以，11.1.0.6版本及更高在v$session增加了2个字段CREATOR_ADDR和CREATOR_SERIAL#用于标记这种情况。然后就可以通过操作系统命令kill -9来杀掉这些进程了。

一.3.4.5 ALTER SYSTEM KILL SESSION 权限不足

需要赋权限，dba角色不用，grant alter system to lhr;

About Me

..........................................................................................................................................................................................................

本文作者：小麦苗，只专注于数据库的技术，更注重技术的运用

本文在ITpub（http://blog.itpub.net/26736162）和博客园(http://www.cnblogs.com/lhrbest)有同步更新

本文地址：http://blog.itpub.net/26736162/viewspace-2121019/

本文pdf版：http://yunpan.cn/cdEQedhCs2kFz （提取码：ed9b）

小麦苗分享的其它资料：http://blog.itpub.net/26736162/viewspace-1624453/

联系我请加QQ好友(642808185)，注明添加缘由

于 2016-06-15 10:00~ 2016-06-26 19:00 在中行完成

上一篇： Oracle索引的监控
下一篇： Oracle kill session相关问题（下）

xiaoxiao树

关注

1
点赞
踩
5

收藏

觉得还不错? 一键收藏
0
评论
Oracle kill session相关问题（上）

全部分类数据库大数据开发技术操作系统网络技术信息化IT人生Oracle kill session相关问题（上）5670阅读　0评论2016-06-26　lhrbest分类：Oracle Oracle kill session相关问题（上）一.1 BLOG文档结构图Oracle kill session相关问题 - 3 -1.1 BLOG文档结构图
复制链接

扫一扫