知

【MVS】如何完全kill rman 备份进程

性能分析和调优

吴成井

2025-04-04 发表

问题描述

现象

需要停止RMAN任务，直接杀掉rman备份主进程后，启动rman备份新任务后，RMAN任务卡顿。

解决方法

查分析

前一次RMAN进程可能未完全释放数据库资源（如锁、内存结构）、数据库后台进程可能仍在处理上次中断的备份操作等蒋导致数据库卡顿。

需要杀掉rman备份脚本的PID和各个通道的PID；

处理过程：

1.查看rman 进程信息；

SQL> set linesize 300

SQL> select sid,spid,client_info from v$process p,v$session s where p.addr=s.paddr and client_info like '%rman%';

SID SPID CLIENT_INFO

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

1011 113286 rman channel=ORA_DISK_4

1024 113958 rman channel=ORA_DISK_7

1143 112812 rman channel=ORA_DISK_2

1215 112587 rman channel=ORA_DISK_1

41 113559 rman channel=ORA_DISK_5

207 114056 rman channel=ORA_DISK_8

350 113026 rman channel=ORA_DISK_3

795 94931 rman channel=ORA_DISK_1

806 94932 rman channel=ORA_DISK_2

818 94933 rman channel=ORA_DISK_3

831 94934 rman channel=ORA_DISK_4

SID SPID CLIENT_INFO

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

857 94936 rman channel=ORA_DISK_5

877 94937 rman channel=ORA_DISK_6

901 94940 rman channel=ORA_DISK_7

914 94942 rman channel=ORA_DISK_8

1035 113756 rman channel=ORA_DISK_6

16 rows selected.

2.根据当前rman 日志显示分配通道的SID，筛选出旧通道的SID；

SID SPID CLIENT_INFO

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

795 94931 rman channel=ORA_DISK_1

806 94932 rman channel=ORA_DISK_2

818 94933 rman channel=ORA_DISK_3

831 94934 rman channel=ORA_DISK_4

857 94936 rman channel=ORA_DISK_5

877 94937 rman channel=ORA_DISK_6

901 94940 rman channel=ORA_DISK_7

914 94942 rman channel=ORA_DISK_8

3.查看系统进程

查询旧通道的PID后，发现该进程成为孤儿进程，而新rman备份任务的通道归属于新rman备份任务。

旧任务进程：

[oracle@hisdb2 database]$ ps -ef |grep 94931

oracle 94931 1 0 21:00 ? 00:00:30 oraclehisdb2 (DESCRIPTION=(LOCAL=YES)(ADDRESS=(PROTOCOL=beq)))

oracle 126860 7987 0 22:15 pts/0 00:00:00 grep 94931

[oracle@hisdb2 database]$ ps -ef |grep 94932

oracle 94932 1 0 21:00 ? 00:00:28 oraclehisdb2 (DESCRIPTION=(LOCAL=YES)(ADDRESS=(PROTOCOL=beq)))

oracle 127039 7987 0 22:15 pts/0 00:00:00 grep 94932

新任务进程

[oracle@hisdb2 database]$ ps -ef |grep 113026

oracle 113026 111114 0 21:42 ? 00:00:03 oraclehisdb2 (DESCRIPTION=(LOCAL=YES)(ADDRESS=(PROTOCOL=beq)))

oracle 127080 7987 0 22:15 pts/0 00:00:00 grep 113026

[oracle@hisdb2 database]$ ps -ef |grep 111114

oracle 111114 111110 0 21:38 ? 00:00:04 rman app/oracle/db/11.2.0.4/bin/rman target / log /databk/log/incrbackup_hisdb2.log append

oracle 111244 111114 0 21:38 ? 00:00:01 oraclehisdb2 (DESCRIPTION=(LOCAL=YES)(ADDRESS=(PROTOCOL=beq)))

oracle 111662 111114 0 21:39 ? 00:00:00 oraclehisdb2 (DESCRIPTION=(LOCAL=YES)(ADDRESS=(PROTOCOL=beq)))

oracle 112587 111114 0 21:41 ? 00:00:05 oraclehisdb2 (DESCRIPTION=(LOCAL=YES)(ADDRESS=(PROTOCOL=beq)))

oracle 112812 111114 0 21:41 ? 00:00:05 oraclehisdb2 (DESCRIPTION=(LOCAL=YES)(ADDRESS=(PROTOCOL=beq)))

oracle 113026 111114 0 21:42 ? 00:00:04 oraclehisdb2 (DESCRIPTION=(LOCAL=YES)(ADDRESS=(PROTOCOL=beq)))

oracle 113286 111114 0 21:42 ? 00:00:06 oraclehisdb2 (DESCRIPTION=(LOCAL=YES)(ADDRESS=(PROTOCOL=beq)))

oracle 113559 111114 0 21:43 ? 00:00:03 oraclehisdb2 (DESCRIPTION=(LOCAL=YES)(ADDRESS=(PROTOCOL=beq)))

oracle 113756 111114 0 21:43 ? 00:00:04 oraclehisdb2 (DESCRIPTION=(LOCAL=YES)(ADDRESS=(PROTOCOL=beq)))

oracle 113958 111114 0 21:44 ? 00:00:02 oraclehisdb2 (DESCRIPTION=(LOCAL=YES)(ADDRESS=(PROTOCOL=beq)))

oracle 114056 111114 0 21:44 ? 00:00:02 oraclehisdb2 (DESCRIPTION=(LOCAL=YES)(ADDRESS=(PROTOCOL=beq)))

oracle 129056 7987 0 22:20 pts/0 00:00:00 grep 111114

4.杀掉旧rman任务的通道进程，

# kill -9 26224

再次查询SQL，发现通道任务消失,同时系统中备份过程的临时文件也自动消失。

总结