i/o slave wait

最新推荐文章于 2021-09-30 21:48:58 发布

最新推荐文章于 2021-09-30 21:48:58 发布 · 343 阅读

文章标签：

#数据库

今日接到报告，某作业运行异常缓慢，通过监控发现I/O资源接近饱和，进一步排查发现是SYS用户正在进行RMAN备份操作导致。本文讨论如何避免数据库备份期间影响正常业务。

今天Dev team给我报告说有个JOB 跑了2个小时还没跑完，平时只需要10多分钟，叫我check一下。

数据库版本：

SQL> select * from v$version;

BANNER
----------------------------------------------------------------------------
Oracle Database 11g Enterprise Edition Release 11.1.0.7.0 - 64bit Production
PL/SQL Release 11.1.0.7.0 - Production
CORE 11.1.0.7.0 Production
TNS for HPUX: Version 11.1.0.7.0 - Production
NLSRTL Version 11.1.0.7.0 - Production

由于我是用toad监控的，所以不方便贴出具体信息，该SQL的等待事件是 i/o slave wait，所以我怀疑I/O出现了问题

登陆HP-UNIX，用glance监控，发现Disk Util 有很大部分时间是100%

ProcList CPU Rpt Mem Rpt Disk Rpt NextKeys SlctProc Help Exit
Glance C.05.00.100 02:31:24 bdhp4463 ia64 Current Avg High
------------------------------------------------------------------------------------------------------------------------------------------------------
CPU Util S SN NU U | 59% 52% 76%
Disk Util F F | 72% 78% 100%
Mem Util S SU UF F | 53% 57% 60%
Networkil U UR R | 29% 31% 33%
再看了看占用大量磁盘资源的 session，结果发现是SYS在用RMAN 备份。

赶紧给SYSDBA 发邮件说明情况，我不是SYSDBA啊，我必须推卸责任，于是SYSDBA回复了
Dear SYSDBAs,

Today, I was reported that we’re having a long running job on adw1u1.

It has running for about 156 minutes, wait event is i/o slave wait, I’ve asked DEV team, they said the job should be finished within few minutes.
According the wait event , I doubt we are lack of I/O resource, So I checked UNIX side , found that some session consumed lots of I/O resource
finally , I found that top I/O consumed user is SYS, they are backup database using RMAN.
Is there a way to avoid performance problem while you’re backup database?

SYSDBA 回复：

For non-prod we have weekly backups running.
Backups will need to read the data from disk and dump it to tape, that is indeed IO intensive.
But backups are needed and it’s not possible to run a backup without consuming IO.

i/o slave wait 是一个Idle的等待事件，通常出现该等待事件表示I/O资源不足