I am currently running Ontap 8.3.2 on our clusters. We have a Microsoft SQL cluster that runs off of a SAN SVM. We need to upgrade Ontap and are planning on going to 9.4. The SQL cluster nodes are running snapdrive 7.1.3, unified utils 7.0.5, snap manager 7.2.2.
My question is how do I procede? Do I need to upgrade the client stuff first (snap drive, manager, unified utils) or do I upgrade them after I do my ontap upgrade(s)? Also, from reading some of the docs, I need to upgrade snapdrive to 7.1.5 to support ontap 9 or higher. Can I install the newer snapdrive on top of the old version, or do I need to uninstall the old version first? Is a reboot required for the client installs?
I feel this will be a "rip the bandaid off" approach for you. It really doesn't matter if you upgrade all the plugins before or after ONTAP, just know you'll lose funcationally during the upgrade(s). Also something to keep in mind; You'll need three hops to get from 8.3 to 9.4. -> 8.3 to 9.1 then 9.1 to 9.3 and finially 9.3 to 9.4. Which you can technically do in one go (just verify everything in your environment via the NetApp IMT), but that's a lot for an envrioment to go through at once IMHO.
Generally yes, you can just run the installer, it'll clean up what it needs to.
I don't see anything in the docs about reboots of hosts, and don't recall any doing so during any upgrades.
Thanks for the information. Yes its a lot. We will probably space out the steps for the Ontap upgrades. I guess my big concern was the client apps. I just wanted to make sure that I can upgrade the "inactive" node of the sql cluster, then failover to it, making it active, then upgrading the now "inactive" other node without breaking something. I did not build this sql cluster and just wanted to confirm what I was thinking before procedeing. Nobody likes downtime.
For client apps, we have done similarly and there was no downtime for the snap and host utilities upgrade. Even if you have active-passive or active-active, the reboot of the node if done manually post failover of services then there is no downtime on the application.