环境一套x86_64平台的12.1的rac
报错信息如下:
Thu Aug 24 10:01:33 2017
Errors in file /oradata/diag/rdbms/cdb/cdb2/trace/cdb2_m000_3564.trc:
ORA-12850: 无法在所有指定实例上分配从属进程:需要2,已分配1
trace 文件部分输出输出:
*** 2017-08-24 10:01:33.324
*** SESSION ID:(1390.13783) 2017-08-24 10:01:33.324
*** CLIENT ID:() 2017-08-24 10:01:33.324
*** SERVICE NAME:(SYS$BACKGROUND) 2017-08-24 10:01:33.324
*** MODULE NAME:(MMON_SLAVE) 2017-08-24 10:01:33.324
*** CLIENT DRIVER:() 2017-08-24 10:01:33.324
*** ACTION NAME:(Automatic Report Flush) 2017-08-24 10:01:33.324
*** CONTAINER ID:(1) 2017-08-24 10:01:33.324
2、报错分析
在mos找到了2篇文章和此次报错相吻合
ORA-12850 is Raised by Mmon Slave Automatic Report Flush Action (文档 ID 2217969.1)
High CPU Usage and/or Frequent Occurrences of ORA-12850 For Monitor Queries by MMON From 12.1 (文档 ID 2102131.1)
文中提到引起问题的原因:
bug:24554937 ORA-12850 WHILE MMON SLAVE AUTOMATIC REPORT FLUSH ACTION
There is a new feature in 12C called "Automatic Report Capturing Feature". As part of this feature, some monitoring SQLs are executed by MMON_SLAVE to identify the resource-intensive SQLs and generate the SQL Monitoring report automatically for those SQLs. Those SQLs consume little more CPU and it is expected behavior being a new feature. Such monitoring queries can be identified from (G)V$SQLSTATS.
==》大致意思是说12.1,引入了 "Automatic Report Capturing Feature"这个新特性,monitoring SQLs被执行以便收集resource-intensive SQLs信息,生成SQL Monitoring report.当monitoring SQLs消耗过高的cpu时会报ORA-12850.
3.解决方案
文章中提供了2中解决方案:
1.Apply Patch 24554937 if it is available for your version and platform.
2.执行以下sql禁用新特性
SQL> alter system set "_report_capture_cycle_time"=0;
来自 “ ITPUB博客 ” ,链接:http://blog.itpub.net/25923810/viewspace-2144072/,如需转载,请注明出处,否则将追究法律责任。
转载于:http://blog.itpub.net/25923810/viewspace-2144072/