0

I have currently 1-node (it's a lab) Windows 2012R2 cluster with SQL server 2014 installed (patched to the last patch based on sqlserverupdates.com). The goal is to upgrade the cluster to something newer and supported. I found out that a node with youngest OS I can join to the cluster is 2016 (probably cannot skip OS version and joining 2019 has failed anyway). That is tolerable. When I tried to install SQL 2016,2019,2022, all of them (see screenshot) claim to not to see the cluster, only SQL 2014 being installed on the Win2016 node does. Reading forums throughout the whole day, I am convinced it is possible to join newer SQL(2016+) to existing cluster with older SQL(2014) without any hiccup.

Then, when failing over to newer SQL, I am aware I cannot go back to older SQL (SQL internally updates the files), and that's fine - goal is to gradually invite newer nodes to cluster and evict old and unsupported nodes. But why I can't join SQL2016+ to the cluster is a puzzle for me in the first place. Suggestions appreciated.

SQL 2016 node being added to 2014 cluster

Improve this question

1 Answer 1

Reset to default
3

When I tried to install SQL 2016,2019,2022, all of them (see screenshot) claim to not to see the cluster, only SQL 2014 being installed on the Win2016 node does.

Actually it's not claiming to not see the cluster, it's claiming to not see any failover cluster instances for that version of SQL Server.

This is expected for any version that isn't the major version of the currently installed failover cluster instance of which you're attempting to add a node. This is why the 2014 add node "sees" the failover cluster instance and none of the other versions do. The other versions will "see" it if you create a new failover cluster instance. You cannot, however, add a SQL Server higher/lower major version as a new instance to a higher/lower major version failover cluster instance.

If you want a new failover cluster instance of a different version then you should use new failover cluster instance from the setup.

Improve this answer
2
  • Hi Sean, thank you for your feedback. Your answer has shed a light onto the problem I am trying to solve. In this thread: reddit.com/r/SQLServer/comments/103w7fa/… Edwin M. Sarimento himself confirms it's possible to have 1 cluster with two different SQL server versions (2019 and 2017). So the question is, how do I squish e.g. 2017 and e.g.2019 to one cluster given (to migrate from old to new version forever) that you confirmed it's not possible to combine different major versions of SQL server. Commented Nov 2, 2023 at 7:31
  • You can either install a new FCI version and migrate the databases or add 2014 to the node and then do a rolling upgrade. learn.microsoft.com/en-us/sql/sql-server/failover-clusters/… Commented Nov 2, 2023 at 10:32

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.