故事发生在2010年4月8日下午,Ex2007 CCR集群节点中的被动节点虚拟机磁盘空间不足。
导致主节点的日志无法重播到被动节点,造成主节点日志盘迅速膨胀。
结果可想而知了,所有库都unmount了。
全公司所有用户的邮件应用中断,幸好有OC顶着,中断时长大约30分钟至1小时。
要不麻烦大了,现将解决方案分享如下:
解决方案分享:
一、测试部分
1、 在虚拟机上新建一个测试存储组,mount一个测试数据库
2、 将几个测试账号新建在该测试存储组合测试数据库下,开始向这几个测试账号发送大邮件
3、 将测试存储组复制中止,然后再向这几个测试账号发送大邮件
4、 Dismount 数据库,停掉复制及Information store 服务
5、 手工清除活动节点的测试日志
6、 手工清除被动节点的所有测试日志和测试数据库
7、 启动主节点information store 及复制服务,
8、 进入群组管理器,将所有应用联机
9、 重新mount 测试数据库
10 查看测试用户的邮件是否丢失?若无丢失,可进入真实环境操作。
二、真实环境部分
1、 先Dismount其中一个数据库,然后重复4-10步
2、 将所有的数据库都Dismount,然后重复4-10步
3、 发送测试邮件,看邮件服务器是否正常
4、 进行被动节点CCR角色脱离,进入添加删除程序,直接将Exchange server 2007 卸载即可。
5、 关闭被动节点,将主节点重新启动。
6、 发送测试邮件,检查邮件服务器是否正常
参考文档:
1、日志清除文档
We have an Exchange 2007 CCR Cluster that spans two seperate subnets. Our data has gotten really out of sync and the logs are taking up a TON of space. On the passive side the disk is completely out of space which is why I believe I cannot re-initialize. So here is what I thought was a good idea.. I want to somehow stop the replication, run a full backup on the active side to clear the logs(as of right now, they will not clear and I read thats because they havent been copied), completely clear the passive side and then start the replication again. Is their anyway to get this accomplished?
You are correct that the logs will not clear since the logs haven't got copied across. What you could do is, run the following command.
Suspend-StorageGroupCopy -identity ServerName/SGName
http://technet.microsoft.com/en-us/library/aa998182.aspx
Dismount Mailbox and PF Stores.
Stop Replication Service, Information Store Service.
Remove all logs from within log files folder (on Active Node).
Restart Replication Service, Information Store Service.
Mount Mailbox and PF Stores.
Start Backup of Information Stores.
On Passive Node, please empty database and log files from target folders.
Run the following command to initiate re-seed.
Update-StorageGroupCopy --identity ServerName/SGName
http://technet.microsoft.com/en-us/library/aa998853.aspx
That should be it.
2、CCR角色脱离文档