SQL群集的就地升级始终在2016年 [英] In-Place Upgrade of SQL cluster always on 2016

问题描述

大家好，

我在Windows 2012 R2操作系统上有一个AlwaysOn 2016 MSSQL双节点群集。

I have a AlwaysOn 2016 MSSQL Two node cluster on Windows 2012 R2 Operative System.

我需要升级Windows 2019 Server（就地）的节点。 SQL无需升级。只有操作系统。

I need upgrade the nodes to Windows 2019 Server (In-Place). No upgrade required for SQL. Only OS.

仅支持操作系统的就地升级？

In-Place upgrade of only Operative system is a supported scenario?

最佳问候。

推荐答案

嗨Marco，

Hi Marco,

如果这是我，我不会这样做 我不喜欢普通服务器的操作系统就地升级，因为它在文件系统上留下了很多混乱。 在混合中添加SQL让我更喜欢它。

If this was me, I would not do this.  I don't like in-place upgrades on the OS for normal servers as it leaves a lot of mess behind on the file system.  Adding SQL to the mix makes me like it even less.

微软实际上在Server 2016中进行了一次大的改变，称为混合模式集群，这使得升级变得更加容易。

Microsoft actually made a BIG change in Server 2016 called mixed mode clusters to make upgrades easier for this exact reason.

如果您有可用的硬件，这将是我的建议。 如果安装2台运行Server 2019的新服务器，则可以创建混合模式群集，从而允许您将2台新的2019服务器添加到现有的Windows 2012 R2故障转移
群集中。 然后，您可以将这两个新节点作为附加副本添加到AlwaysOn配置中，并在数据同步后将它们转移到新的2019服务器。

Here would be my advice, if you have the available hardware.  If you install 2 NEW servers running Server 2019 then you have the ability to create a mixed mode cluster, which allows you to add 2 new 2019 servers into the existing Windows 2012 R2 Failover Cluster.  You can then add these 2 new nodes as additional replicas to the AlwaysOn configuration and fail them over to the new 2019 servers once data has been synchronised.

执行上述操作需要2台新服务器建立，但它消除了完整的AlwaysOn迁移的潜在痛苦过程（这是我假设你试图避免的部分？）并且是一个远程清洁过程。 您只需将SQL超过
及其在新的（干净的）2019服务器上失败： - ）

Doing the above does require 2 new servers to be built, but it takes away the potentially painful process of doing a full AlwaysOn migration (which is the part I assume you're trying to avoid?) and is a FAR FAR cleaner process.  You just fail SQL over and its on the new (clean) 2019 servers :-)

执行上述操作后，您可以将2012 R2节点保留一段时间因为你想确保运行正常。 然后，当您对它感到满意时，您需要运行1个PowerShell命令将群集功能升级到2019（可以随时执行
，但是一旦执行此操作，您将无法再回到2012 R2功能）并删除旧节点。

Once you do the above, you can leave the 2012 R2 nodes in for as long as you want to make sure things run OK.  Then when you're happy with it, you need to just run 1 PowerShell command to upgrade the cluster functionality to 2019 (which can be done at any time but you cannot fail back to 2012 R2 functionality once you do this) and remove the old nodes.

这提供了一个很好的干净升级过程！

This gives a nice clean upgrade process!

如果这听起来是合适的选择，请告诉我？

Let me know if this sounds a suitable alternative?

谢谢，

马特

这篇关于SQL群集的就地升级始终在2016年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！