用于永远在线的DR练习 [英] DR drill for Always-On

问题描述

假设我总是有3个节点

主节点1 ：同步

辅助节点2：同步

辅助节点3：异步

对于DR钻取（非实际DR），手动故障转移到node3。并且节点3上的读写和节点1/2上的只读

因此这将影响生产，因为它是只读的。 !!!!!

任何实现node1 / 2和node3的方法都是读写的???是否需要重新配置AG？

解决方案

您好，

如果您有所有3个节点一个可用性组然后你应该继续使用你的主连接字符串（基于AlwaysOn组的VIP或DNS名称）。

这里最重要的一点就是失败了设置为手动故障转移的节点停止将数据同步到节点1和节点2.

因此，如果故障转移到节点3，那么生产读取和写入将转到（所以你仍然将此数据视为生产，即使您处于'DR'）。  

一旦您进行了故障转移并测试了连接是否正常，您需要恢复数据移动可用性组中的所有数据库，以便它复制回节点1和节点2. 如果您不希望DR数据更改将
复制回节点1和节点2，那么您不应该在这些数据库上恢复数据移动。

但是，如果你不这样做恢复数据移动，然后您可能会丢失自故障转移后添加到数据库中的所有数据。

总之 - 作为测试失败到DR，将导致生产流量转到DR节点。 因此，您需要在重新启动之前恢复数据移动到每个数据库（通过可用性组）。 或者您将丢失数据。

如果您有任何其他问题，请告诉我。

谢谢，

马特

Suppose I have 3 nodes always on

Primary node1 : Sync

Secondary node2 : Sync

Secondary node3: Async

For DR drill (Not actual DR) , manual failover to node3 . And read-write on node 3 and read-only on node 1/2

So this will affecting the production as this is read-only. !!!!!

Any method to achieve both node1 / 2 and node3 are read-write ??? Need reconfiguration of the AG ?

解决方案

Hi,

If you have all 3 nodes in an availability group then you should continue to use your main connection string (based on the VIP or DNS name of the AlwaysOn group).

The most important thing to mention here is that failing over to a node which is set to manual failover stops synchronising data to node 1 and node 2.

So if you failover to node 3 then that is where production reads and writes will go to (so you still treat this data as production, even though you are in 'DR').  

Once you have failed over and tested that connectivity works, you need to resume data movement for all databases in the availability group so that it replicates back to node 1 and node 2.  If you DO NOT wish for your DR data changes to replicate back to node 1 and node 2 then you should not resume data movement on these databases.

However, if you dont resume data movement on them then you risk losing all data that was added to the DB since the failover.

In summary - failing over to DR as a test, will cause production traffic to go to the DR node.  So you need to resume data movement to each DB (via teh availability group) before failing back over.  Or you will lose data.

Let me know if you have any further questions.

Thanks,

Matt

这篇关于用于永远在线的DR练习的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！