One week ago NSX-T version 2.4.1 (Build 13716575) was released. Dozens of resolved issues are listed in the release notes. The process of upgrading a deployment is depicted in this post.
First step is to download the 7,5 GB upgrade bundle file and upload it in the first screen of the NSX-T GUI’s Upgrade section:
After the upload is complete the bundle is extracted and its compatibility matrix is checked. Afterwards the upgrade process can be started:
The obligatory End User License Agreement has to be accepted as usual:
First step in the upgrade process is to upgrade the “Upgrade Coordinator” component:
When this step is completed three boxes with the current and new versions for the hosts, edges and management nodes are displayed:
It is recommended to run the pre-checks first, which check if the environment correctly configured for the further upgrade steps, e.g. whether the vSphere clusters are configured for DRS:
When the pre-checks are completed successfully you can proceed to the second step of the ugprade process which is upgrading the hosts. All of the hosts known to NSX via Fabric/Nodes are displayed and grouped according to their clusters in vCenter. The order of the hosts in each group can be changed, as can the upgrade order (parallel or one after the other). The upgrade mode “Maintenance” is recommended for productive environment, which evacuates (vMotion) each host while placing it in maintenance mode before installing the new NSX VIBs.
For test deployments the “In-place” upgrade mode can be selected, which might lead to service interuptions of the network functions offered by NSX to the running VMs.
The overall group upgrade order defines whether the host groups should be upgraded simultaneously:
During the upgrade the invidual status of each group can observed by clicking on it:
When all hosts are upgraded you can contine to the next step by clicking on “Next”:
All edge VMs have to be part of an edge cluster as those correspond to the edge groups, by which the edges are upgraded. During the upgrade the status reveals that a new operating system is installed on these:
When all edges are upgraded you can contine to the next step by clicking on “Next”:
With the NSX-T 2.4 upgrade the controller functionality was moved from the dedicated controller VMs to the manager, which was in turn changed from a single VM to a cluster, the fourth step is obsolete and can be skipped by clicking on “Next”:
The upgrade of the NSX-T manager cluster should be communicated to concerned parties (e.g. network admins) as functionality will not be available during the maintenance window:
The three manager VMs are upgraded in parallel:
By clicking on “More information” the detailed upgrade logs are displayed:
After completing the upgrade the manager VMs are rebooted. Until the services are available again this message is displayed:
With the management nodes being upgraded successfully the upgrade process is completed:
The upgrade history can be tracked by clicking on “Show Upgrade History”: