Oracle死锁一例(ORA-00060)，锁表导致的业务死锁问题
1、问题发现
检查客户数据库的时候发现存在大量死锁的情况
Thread advanced to log sequence (LGWR switch)
Current log# seq# mem# : /oradata/oracle/online_log/redo16_01.log
Current log# seq# mem# : /oradata/oracle/online_log/redo16_02.log
Tue Jul ::
Archived Log entry added for thread sequence ID 0x59dc8ffa dest :
Tue Jul ::
LNS: Standby redo logfile selected for thread sequence for destination LOG_ARCHIVE_DEST_2
Tue Jul ::
opiodr aborting process unknown ospid () as a result of ORA-
Tue Jul ::
ORA-: Deadlock detected. More info in file /u01/oracle/diag/rdbms/orcl/orcl/trace/orcl_ora_25846.trc.
Tue Jul ::
ORA-: Deadlock detected. More info in file /u01/oracle/diag/rdbms/orcl/orcl/trace/orcl_ora_14067.trc.
Tue Jul ::
ORA-: Deadlock detected. More info in file /u01/oracle/diag/rdbms/orcl/orcl/trace/orcl_ora_20781.trc.
Tue Jul ::
Thread cannot allocate new log, sequence
Private strand flush not complete
查看trace文件orcl_ora_25846.trc结果如下
Deadlock graph:
---------Blocker(s)-------- ---------Waiter(s)---------
Resource Name process session holds waits process session holds waits
TX-026e0020-000001a5 X S
TM-0007fd6c- X SX session : DID --000001FEsession : DID --
session : DID --00000014session : DID --000001FE Rows waited on:
Session : obj - rowid = 0007FD6C - AAAAAAAAAAAAAAAAAA
(dictionary objn - , file - , block - , slot - )
Session : no row ----- Information for the OTHER waiting sessions -----
Session :
sid: ser: audsid: user: /FD14
flags: (0x45) USR/- flags_idl: (0x1) BSY/-/-/-/-/-
flags2: (0x40009) -/-/INC
pid: O/S info: user: oracle, term: UNKNOWN, ospid:
image: oracle@dbserver1
client details:
O/S info: user: TL3050, term: TL3050-WZ, ospid: :
machine: WORKGROUP\TL3050-WZ program: CWV4.2.8.337_20131204.exe
application name: CWV4.2.8.337_20131204.exe, hash value=
current SQL:
insert into pzd2018
(UNI_NO,ORD,STYPE,STYPE2,SNO,SYEAR,SMONTH,RMONTH,SDAY,SABSTRACT,OPERATOR,J_AMOUNT,D_AMOUNT,SUBJ,SUBJNAME,
OPP_SUBJ,SRC_CODE,ECO_CODE,SRC_PAYTYPE,SRC_BUTYPE,ECO_TYPE,ECO_WARRANT,PRJ_ORDER,PRJ_NAME,OPP_PRJ,CLR_ORDER,
UNIT_CODE,SPECCODE,CONTRACT_NO,CAR_NO,OLPAY_SNO,schedule_date,WB_TYPE,WB_JNUM,WB_DNUM,WB_FACT,NUM_TYPE,
NUM_JNUM,NUM_DNUM,NUM_PRICE,CAP_NO,CAP_ORD,JSFS_CODE,ZPH,BUSS_DATE,OTHER_UNIT,ACNT,BANKNO,ADDRESS1,ADDRESS2,
TNO,ACT_NO,BU_CODE,T_CODE,RESBU_CODE,RESBU_AMT,SPECCODE1,SPECCODE2,SPECCODE3,SPECCODE4,RES_S1,
RES_S2,RES_S3,RES_S4,ASSET_SUBJ,TAX_NO,SRC_NAME,ADDITION,UNI_PRJ_ORDER,Clr_Bu_Code,
Source_Type,Source,SrKey,SMark,Uni_Prj_Name,clrsno,input_name,check_name,attach_act,
attach_act_no,pz_attr,src_type,zj_type,ref_uni_no,charge_sno,charge_name,src_lkx,order_type,c ----- End of information for the OTHER waiting sessions ----- Information for THIS session: ----- Current SQL Statement for this session (sql_id=9ktt36bsngnyx) -----
insert into pz2018
(UNI_NO,STYPE,STYPE2,SNO,SYEAR,SMONTH,RMONTH,SDAY,INPUT_NAME,CHECK_NAME,COMP_NAME,COMP_NAME2,
ADDITION,CHILDNUM,J_AMOUNT,D_AMOUNT,SSTATE,REMARK,PZ_ATTR)
values(:,:,:,:,:,:,:,:,:,:,:,:,:,:,:,:,:,:,:)
===================================================
2、问题分析
可以看出来241号会话持有一个TM锁，在执行insert into pzd2018语句在等待S锁
4468号会话持有一个TX锁，在执行insert into pz2018语句，在等待SX锁
通过与业务沟通与数据库查询发现了以下的锁表操作，并和业务确定了属于业务SQL
lock table pz2018 in exclusive mode
到这里问题已经清楚了，整个逻辑是这样的
241号会话将pz2018全表排他模式进行了锁定，导致4468会话无法对pz2018表进行insert操作，原因是无法在表上获取共享排它锁即SX锁，导致4468号会话进入等待模式
而4468号会话在等待前进行了insert into pzd2018操作，而241号会话在插入时存在唯一约束，导致241会话进行TX锁等待，等待4468号session数据提交或者回滚
这样一个环状等待就形成了即死锁
等待发生时会话的等待情况
SQL> select a.sample_time,
2 a.session_id,
3 a.session_serial#,
4 a.blocking_session bsession,
5 a.blocking_session_serial# bserial#,
6 a.event,
7 a.machine,
8 a.module,
9 a.sql_opname
10 from dba_hist_active_sess_history a
11 where a.session_id in (241, 4468, 6819, 10817)
12 and a.sample_time > to_date('', 'yyyymmddhh24')
13 and a.sample_time < to_date('', 'yyyymmddhh24')
14 and a.event is not null
15 order by a.sample_time
16 ;
SAMPLE_TIME SESSION_ID SESSION_SERIAL# BSESSION BSERIAL# EVENT MACHINE MODULE SQL_OPNAME
------------------------------ ---------- --------------- ---------- ---------- ------------------------------ ------------------------------ ------------------------------ ----------------------------------------------------------------
03-7月 -18 10.00.40.857 上午 241 425 6587 7875 enq: TM - contention WORKGROUP\TL3050-WZ CWV4.2.8.337_20131204.exe LOCK TABLE
03-7月 -18 10.49.45.384 上午 10817 97 12929 7665 read by other session webserver JDBC Thin Client SELECT
03-7月 -18 10.51.16.143 上午 241 425 4468 2029 enq: TX - row lock contention WORKGROUP\TL3050-WZ CWV4.2.8.337_20131204.exe INSERT
03-7月 -18 10.51.16.143 上午 4468 2029 241 425 enq: TM - contention ZDCW\WANGH88208561 XCV5（新5.24）.exe INSERT
03-7月 -18 10.52.46.903 上午 10817 121 null event dbserver1 SELECT
03-7月 -18 10.53.57.464 上午 6819 99 10817 133 enq: TX - row lock contention ZDCW\WANGWD88981612 CWV4.exe INSERT
03-7月 -18 10.53.57.464 上午 10817 133 6819 99 enq: TM - contention ZDCW\WANGH88208561 XCV5（新5.24）.exe INSERT
03-7月 -18 10.54.07.554 上午 10817 133 db file parallel read ZDCW\WANGH88208561 XCV5（新5.24）.exe INSERT
8 rows selected
3、锁等待的模拟，问题复现
---session1
SQL> lock table pz2018 in exclusive mode;
Table(s) Locked.
----session2
SQL> insert into pzd2018(OBJECT_ID,OBJECT_NAME) values(100000000,'PZD_Yong');
1 row created.
SQL> insert into pz2018(OBJECT_ID,OBJECT_NAME) values(100000000,'PZ_Yong');
----session2执行直接hang住无法完成，在等待TM锁
SQL> select s.SID,s.BLOCKING_SESSION bsid,s.EVENT,s.MACHINE,s.MODULE,s.STATUS,s.STATE from v$session s where s.EVENT is not null and s.STATUS='ACTIVE' and s.WAIT_CLASS<>'Idle';
SID BSID EVENT MACHINE MODULE STATUS STATE
---------- ---------- ------------------------------ ------------------------------ ---------------------------------------- -------- -------------------
17 143 enq: TM - contention 172-16-8-110 SQL*Plus ACTIVE WAITING
----session1执行
SQL> insert into pzd2018(OBJECT_ID,OBJECT_NAME) values(100000000,'PZD_Yong');
--session1 直接hang住，session2抛出错误发现死锁
SQL> insert into pz2018(OBJECT_ID,OBJECT_NAME) values(100000000,'PZ_Yong');
insert into pz2018(OBJECT_ID,OBJECT_NAME) values(100000000,'PZ_Yong')
*
ERROR at line 1:
ORA-00060: deadlock detected while waiting for resource
----检查数据库等待
SQL> set linesize 1000 pagesize 5000
SQL> col EVENT for a30
SQL> col MACHINE for a30
SQL> col MODULE for a40
SQL> select s.SID,s.BLOCKING_SESSION bsid,s.EVENT,s.MACHINE,s.MODULE,s.STATUS,s.STATE from v$session s where s.EVENT is not null and s.STATUS='ACTIVE' and s.WAIT_CLASS<>'Idle';
SID BSID EVENT MACHINE MODULE STATUS STATE
---------- ---------- ------------------------------ ------------------------------ ---------------------------------------- -------- -------------------
143 17 enq: TX - row lock contention 172-16-8-110 SQL*Plus
查看数据库alert日志发现
Thu Jul 05 11:40:40 2018
ORA-00060: Deadlock detected. More info in file /u01/app/oracle/diag/rdbms/dtstack1/dtstack1/trace/dtstack1_ora_29840.trc.
查看死锁trace
Deadlock graph:
---------Blocker(s)-------- ---------Waiter(s)---------
Resource Name process session holds waits process session holds waits
TM-0001554c-00000000 28 17 X 27 143 SX
TX-00040008-000002d9 27 143 X 28 17 S session 17: DID 0001-001C-00000024 session 143: DID 0001-001B-00000155
session 143: DID 0001-001B-00000155 session 17: DID 0001-001C-00000024 Rows waited on:
Session 17: no row
Session 143: obj - rowid = 0001554C - AAAAAAAAAAAAAAAAAA
(dictionary objn - 87372, file - 0, block - 0, slot - 0) ----- Information for the OTHER waiting sessions -----
Session 143:
sid: 143 ser: 905 audsid: 610017 user: 85/DTYONG
flags: (0x45) USR/- flags_idl: (0x1) BSY/-/-/-/-/-
flags2: (0x40009) -/-/INC
pid: 27 O/S info: user: oracle, term: UNKNOWN, ospid: 29846
image: oracle@172-16-8-110 (TNS V1-V3)
client details:
O/S info: user: oracle, term: pts/1, ospid: 29845
machine: 172-16-8-110 program: sqlplus@172-16-8-110 (TNS V1-V3)
application name: SQL*Plus, hash value=3669949024
current SQL:
insert into pz2018(OBJECT_ID,OBJECT_NAME) values(100000000,'PZ_Yong') ----- End of information for the OTHER waiting sessions ----- Information for THIS session: ----- Current SQL Statement for this session (sql_id=7jbgf8fc4cac0) -----
insert into pzd2018(OBJECT_ID,OBJECT_NAME) values(100000000,'PZD_Yong')
===================================================
PROCESS STATE
-------------
Process global information:
process: 0x914c4ed0, call: 0x910d2008, xact: 0x90390710, curses: 0x916aa020, usrses: 0x916aa020
in_exception_handler: no
----------------------------------------
SO: 0x914c4ed0, type: 2, owner: (nil), flag: INIT/-/-/0x00 if: 0x3 c: 0x3
proc=0x914c4ed0, name=process, file=ksu.h LINE:12721, pg=0
(process) Oracle pid:28, ser:23, calls cur/top: 0x910d2008/0x910d2008
flags : (0x0) -
flags2: (0x0), flags3: (0x10)
intr error: 0, call error: 0, sess error: 0, txn error 0
intr queue: empty
ksudlp FALSE at location: 0
(post info) last post received: 138 0 2
last post received-location: ksl2.h LINE:2374 ID:kslpsr
last process to post me: 0x914b3298 1 6
last post sent: 0 0 26
last post sent-location: ksa2.h LINE:285 ID:ksasnd
last process posted by me: 0x914b3298 1 6
(latch info) wait_event=0 bits=0x0
Process Group: DEFAULT, pseudo proc: 0x915515e8
O/S info: user: oracle, term: UNKNOWN, ospid: 29840
OSD pid info: Unix process pid: 29840, image: oracle@172-16-8-110 (TNS V1-V3)
----------------------------------------
SO: 0x916aa020, type: 4, owner: 0x914c4ed0, flag: INIT/-/-/0x00 if: 0x3 c: 0x3
proc=0x914c4ed0, name=session, file=ksu.h LINE:12729, pg=0
(session) sid: 17 ser: 683 trans: 0x90390710, creator: 0x914c4ed0
flags: (0x45) USR/- flags_idl: (0x1) BSY/-/-/-/-/-
flags2: (0x40009) -/-/INC
Oracle死锁一例(ORA-00060)，锁表导致的业务死锁问题的更多相关文章
- 查看oracle的sql语句历史记录和锁表的情况
查看oracle的sql语句历史记录和锁表的情况 (2012-01-04 20:59:59) 转载▼ 标签: 杂谈 分类: database 查询sql的历史记录 select * from v$sq ...
- oracle 锁表的处理。
最近系统每天经常锁表,进程杀死后,很快再次锁住这个表. (一)先贴出现场处理死锁的步骤. 另外:有时候通过PL/SQL执行kill session可能仍然无法解锁,此时需要登陆到Oracle服务器将进 ...
- oracle 查看锁表情况并处理锁表
/* *locked *query locked object and analyse reason,kill it * */ select 'alter system kill session '' ...
- oracle锁表查询，资源占用，连接会话，低效SQL等性能检查
查询oracle用户名,机器名,锁表对象 select l.session_id sid, s.serial#, l.locked_mode, l.oracle_username, l.os_user ...
- informix 数据库锁表分析和解决方法
一.前言 在联机事务处理(OLTP)的数据库应用系统中,多用户.多任务的并发性是系统最重要的技术指标之一.为了提高并发性,目前大部分RDBMS都采用加锁技术.然而由于现实环境的复杂性,使用加锁技术又不 ...
- 针对MyISAM锁表的解决方案
最近服务器上经常出现mysql进程占CPU100%的情况,使用show processlist命令后,看到出现了很多状态为LOCKED的sql.使用show status like 'table%'检 ...
- ORACLE查询数据库的锁表情况
查询数据库的锁表情况语句如下: SELECT p.spid,a.serial#, c.object_name,b.session_id,b.oracle_username,b.os_user_na ...
- Java并发 行级锁/字段锁/表级锁 乐观锁/悲观锁 共享锁/排他锁 死锁
原文地址:https://my.oschina.net/oosc/blog/1620279 前言 锁是防止在两个事务操作同一个数据源(表或行)时交互破坏数据的一种机制. 数据库采用封锁技术保证并发操作 ...
- Oracle锁表与解锁
查看锁表语句:方法1: select sess.sid, sess.serial#, lo.oracle_username, lo.o ...
随机推荐
- ccf-201709-2 公共钥匙盒
问题描述 有一个学校的老师共用N个教室,按照规定,所有的钥匙都必须放在公共钥匙盒里,老师不能带钥匙回家.每次老师上课前,都从公共钥匙盒里找到自己上课的教室的钥匙去开门,上完课后,再将钥匙放回到钥匙盒中 ...
- Page Cache的落地问题
除非特别说明,否则本文提到的写操作都是 buffer write/write back. 起因 前几天讨论到一个问题:Linux 下文件 close成功,会不会触发 “刷盘”? 其实这个问题根本不用讨 ...
- Android常用的图片加载库
Android常用的图片加载库 前言:图片加载涉及到图片的缓存.图片的处理.图片的显示等.四种常用的图片加载框架,分别是Fresco.ImageLoader. Picasso. Glide. Uni ...
- python SQLAchemy常用语法
SQLAlchemy是Python编程语言下的一款ORM框架,该框架建立在数据库API之上,使用关系对象映射进行数据库操作,简言之便是:将对象转换成SQL,然后使用数据API执行SQL并获取执行结果. ...
- 替换空格（C++和Python 实现）
(说明:本博客中的题目.题目详细说明及参考代码均摘自 “何海涛<剑指Offer:名企面试官精讲典型编程题>2012年”) 题目 请实现一个函数,把字符串中的每个空格替换为 "%2 ...
- C++：关于委托类
转自:http://blog.csdn.net/dadalan/article/details/4041931.vs2010已经支持function/bind,能很好实现委托. ［说明］ 本文不仅介绍 ...
- AJAX原生JavaScript写法
GET方式 //创建XMLHttpRequest对象,为考虑兼容性问题,老版本的 Internet Explorer (IE5 和 IE6)使用 ActiveX 对象 var ajax = windo ...
- CentOS7 安装tomcat为系统服务器 Systemctl管理Tomcat,并设置开机启动
本文转载:http://blog.chinaunix.net/uid-24648266-id-5729891.html CentOS7开始,从/etc/init.d脚本改为了systemctl管理服务 ...
- python全栈学习笔记（三）网络基础之网络设备及架构介绍
- python 面向对象：类，作用域
类(Class)和实例(Instance) 定义类是通过class关键字:class Student(object): pass class后面紧接着是类名,即Student接着是(object),表 ...