On the 27th of Februari 2019 VMware announced the newest version of NSX-T: version 2.4. Which (arguably) is one of the most anticipated versions of all NSX-T versions yet! NSX-T is the go-to version of VMware, the promise of this version is that it delivers an equivalent version compared to NSX-V (for vSphere). On the question if this version can deliver this anticipate promise? I will owe you my answer, as I leave my opinion in the middle. I first need to check NSX-T 2.4 out for myself 🙂
This blog is about how I upgraded my homelab from NSX-T version 2.3 to version 2.4.
First I downloaded the NSX-T 2.4 upgrade bundle (the MUB-file) from the VMware download portal (all 7.34 Gigabytes of them).
When checking the prerequisites I hit the first bump, the compatibility with vSphere.
Only vSphere 6.5 P03+ and vSphere 6.7 EP06+ (and later versions) are supported!
These are the latest ESXi builds, so an vSphere-upgrade is almost mandatory for all NSX-T upgrades.
The second bump was that the NSX-Manager VM minimum requirements went up (from 2 vCPUs) to 4 vCPUS and (from 8 GB of RAM) to 16 GB of RAM.
So in my case I had to shut down the NSX-T Manager and upgrade the specs.
After fulfilling these requirements I could finally start the upgrade-process:
The Upgrade Process
Upload the MUB file to the NSX Manager,
After uploading the MUB file to the NSX Manager, the file is validated and extracted and you are ready to GO!
Downloading, uploading, validating and extracting the upgrade bundle takes about 30 – 45 minutes (so take you time for this step).
Accept the EULA ..
and click ‘Begin Upgrade’.
First start with running some pre-checks.
I absolutely advice you to run these checks, as you cannot foresee all compatibility issues yourself.
The sequence of the update is as follows: Hosts > Edges > Controllers(*) > Management Nodes. Which itself is a straightforward process.
Select the hosts and click on Start to start the hosts upgrade process.
After a few minutes it stated that the upgrade status is successful, but I received an indescribable error-message.
The upgrade interface offers a “post-check” option, to check if the upgrade has successfully been executed.
In my case, the hosts had successfully been upgraded.
In my current lab configuration where no edges configured, but the process is exactly the same as for the hosts upgrade.
I will continue with the controller upgrade .. which is exactly not an upgrade.
With NSX-T 2.4 the Management and the Controller nodes are merged together. In post NSX-T version separate manager and controller nodes had to be deploy. With NSX-T 2.4 a total of 3 appliances are deployed which al share a piece of the manager and controller nodes. A positive result is that less appliances have to be maintained and that the management plane has been made high-available.
In my case I only have 1 Manager appliance and 1 controller appliance, which posses a problem from a high-availability point of view. But the upgrade process is fully supporting this upgrade/merging process.
It allows you to create a single node cluster.
When starting the non-reversibel management and controller upgrade an explicit “start upgrade” message is presented. When upgrade/merging the NSX-T Manager and Controller manually disallow access to them, changes made during the upgrade-process can break the upgrade process. Leaving you with a bricked NSX-T environment ( #backup #restore )
Halfway the upgrade except an error message that the status cannot be fetched, because the NSX-T Manager is rebooted.
I was a bit impatient, but fortunately patient enough: I reloaded the NSX-T Manager website, but I received a small text message box stating that the Manager status was unknown for some time .. and even after a few reloads the status went from “unknown” to “down”. But after a few minutes I received a glorious new UI of NSX-T version 2.4!
I re-checked if the upgrade has successfully been executed:
Et voila .. a successful NSX-T upgrade