查看: 2393|回复: 0

solution for undo tablespace 100% full in noarchive mode

[复制链接]
论坛徽章:
0
跳转到指定楼层
1#
发表于 2005-4-29 09:38 | 只看该作者 回帖奖励 |倒序浏览 |阅读模式
When I prepare a testing environment for data warehouse with huge data volume, A lots of DML and indexes should be done. There is not enough disk space and the UNDO is not as big as you wish. I found a way to solve it and I'd like to discuss with you.

====================
Problem Description
====================
The SYS_UNDO tablespace is 100% full.
We want to drop it and then recreate it to shrink the free space.
The rollback segment has a status of PARTLY AVAILABLE.  I have tried off-lining the rollback segment and then dropping it, but that does not work.
The instance is up and running and the alert log shows the following messages repeatedly:
SMON: mark undo segment %v as available
SMON: about to recover undo segment %v
The database is running in nonarchive log mode, so there is no way to perform a recovery on the datafile containing the problem rollback segment.
Rather than letting the rollback segment keep trying to repeatedly recover an unavailable segment, you need a mechanism to stop this process so you can drop the bad rollback segment and get normal processing resumed with minimal data loss.
====================
Solution
====================
To do this, you need to be able to shutdown (possibly using abort) and restart the database in restricted mode.  No one is to be processing any transactions.   Then off-line and bring back on-line all the user tablespaces.   Once this is done, the problem rollback segments status will change to on-line. Then do shutdown immediate on the database and startup normal.     You will then be able to drop this rollback segment, and subsequently drop its associated tablespace.  The error messages will have gone away completely, along with the I/O problems.  Processing will return to normal without having lost data or by going back to older cold backups of the database.        
====================
Explanation
====================
Doing this off-lining and bringing back on-line of the tablespaces will reset the need for SMON to keep trying to recover the information in the corrupted  rollback segment.

您需要登录后才可以回帖 登录 | 注册

本版积分规则 发表回复

TOP技术积分榜 社区积分榜 徽章 团队 统计 知识索引树 积分竞拍 文本模式 帮助
  ITPUB首页 | ITPUB论坛 | 数据库技术 | 企业信息化 | 开发技术 | 微软技术 | 软件工程与项目管理 | IBM技术园地 | 行业纵向讨论 | IT招聘 | IT文档
  ChinaUnix | ChinaUnix博客 | ChinaUnix论坛
CopyRight 1999-2011 itpub.net All Right Reserved. 北京盛拓优讯信息技术有限公司版权所有 联系我们 未成年人举报专区 
京ICP备16024965号-8  北京市公安局海淀分局网监中心备案编号:11010802021510 广播电视节目制作经营许可证:编号(京)字第1149号
  
快速回复 返回顶部 返回列表