Finally! After all of the noice VMware made at VMware World (USA), NSX-T 2.5 is finally GA and is now available for download! I’ve dedicated this blog post describing the process and pit-falls while updating NSX-T from version 2.4.2 up to version 2.5.0.
Some (free) hints before starting the update:
- Make sure the all NSX-T enabled clusters are running DRS and have sufficient capacity to put at least one host into “maintenance mode” .
- Unmount all mounted CD/DVD/ISO drives from the VM’s.
- Make sure a cluster(3) NSX-T controllers are up and running (Lab setup usually can do with one, but you can run into trouble with a GENEVE backed vMotion network).
- If a physical firewall is in place between NSX Managers and ESXi Hosts/Transport nodes add port TCP/1234 to the allowed rules.
Preparation: Download the latest NSX-T update package “.mub” file from the VMware download page and upload it into the NSX-T Manager.
After the .mub bundle is uploaded and extracted an overview is presented:
Notice the upgrade order has changed since the latest NSX-T version:
Earlier:Upgrade Coordinaton > Configuring & Upgrading Hosts > Upgrade NSX Edge Cluster > Upgrading Management Plane > Upgrade Policy Manager
Updating the edges before upgrading the transport nodes means an easier rollback, seeing replacing edges is usually a simpler process than replacing entire hosts.
I highly recommend running the “Post Checks” after this check for issues and notice this important change (one of the free hints from the beginning of this blog post):
Starting with NSX-T Data Center 2.5, the NSX Messaging channel TCP port from all Transport and Edge nodes to NSX Managers has changed to TCP port 1234 from port 5671. With this change, make sure all NSX-T Transport and Edge nodes can communicate on both TCP ports 1234 to NSX Managers and TCP port 1235 to NSX Controllers before you upgrade to NSX-T Data Center 2.5. Also make sure to keep port 5671 open during the upgrade process.
Now its time to get to business, when more edge clusters are in place running the upgrade “Parallel” across groups con save you some time:
In my lab setup i experienced no issues what so ever, flawless update of the NSX-T Edges. (This is of-course the ultimate redundancy and fail-over test):
After the edges the (transport)hosts are updated, note that certain clusters can be disabled/excluded from update and can also be updated in serial and parallel.
I experienced some errors like mounted drives and hosts not automatically coming out of maintenance mode, but overall the proces was pretty smooth.
Then lastly the Manager/Controller cluster is updated, after which the update is complete and all new features released in NSX-T 2.5 are at you’re disposal!