VMware NSX

  • 1.  NSX-T Upgrade on Cisco Hyperflex Cluster

    Posted Aug 22, 2023 10:04 AM

    We recently onboarded a setup that has NSX-T version 3.1 running on a 14-node Cisco Hyperflex cluster. This particular setup hasn’t been upgraded since its inception, and we're now planning its first upgrade to NSX-T version 3.2.

    While I've done my best to understand the intricacies of both NSX-T and Hyperflex, I acknowledge I might be misinterpreting some aspects. Please correct me if that's the case.

    My specific concerns are:

    1. Upgrading Nodes the Right Way: NSX-T's approach seems to involve sequentially restarting nodes during an upgrade. However, given Hyperflex's architecture, each node requires a preparatory phase, typically managed via HX Connect and followed by a storage rebalance. How can we ensure NSX-T’s upgrade process respects these nuances?

    2. Cluster Health Checks Between Node Upgrades: After upgrading a single node in NSX-T, it's critical to ensure our Hyperflex cluster's health before moving on to the next node. Does NSX-T provide a way to pause between node upgrades or offer an opportunity for manual intervention?

    3. Maintaining Communication: Post-upgrade, the nodes need to maintain seamless communication with each other and align with NSX-T's central controls. What provisions does NSX-T have in place to ensure this?

    Given the above, we'd appreciate guidance on:

    • Best practices for upgrading NSX-T within a Hyperflex environment, taking storage rebalancing into account.

    • Potential challenges we might encounter due to the intertwined nature of NSX-T and Hyperflex.

    • Tips or tools to monitor the health and connections of NSX-T and Hyperflex during the upgrade.

    • Options within NSX-T to either manually intervene or synchronize with Hyperflex's health checks during the process.

    As we navigate this initial upgrade, your expert insights will be instrumental in ensuring a smooth and efficient transition.



  • 2.  RE: NSX-T Upgrade on Cisco Hyperflex Cluster

    Broadcom Employee
    Posted Aug 22, 2023 11:52 AM

    I would recommend to run upgrade check tool - https://kb.vmware.com/s/article/87379 , followed by that you can plan the upgrade using NSX upgrade coordinator. Customization and pausing the upgrade is fully supported. If you need the flexibility to change the order of upgrade for your edge clusters and hosts it is supported only with NSX 4.0.1.1. You should speak with the vendor about the suggested host maintenance activity process based on your storage design.As described in the documentation(Please refer https://docs.vmware.com/en/VMware-NSX-T-Data-Center/3.2/nsxt_32_upgrade.pdf) . If you tailor the upgrade process and carry out serial upgrades , you must then execute the necessary post-checks to ensure the HX cluster is in a healthy state before going on to the next node.



  • 3.  RE: NSX-T Upgrade on Cisco Hyperflex Cluster

    Posted Aug 22, 2023 01:41 PM

    Thanks for your feedback. does the NSX Upgrade Evaluation Tool only focus on the NSX Manager's readiness, or does it take a holistic view, examining the entire setup including Edge nodes, hosts, and the NSX Manager?

    For the host upgrades, I'm leaning towards a hands-on approach: disabling the automated processes and taking charge of the upgrade manually.

    https://docs.vmware.com/en/VMware-NSX/4.1/upgrade/GUID-7C9F9EE1-7174-4855-823E-C2F61EFAD11C.html

    Once the VIB is installed, I will follow the standard reboot process for HX, login to the controller put whatever host you upgraded into maintenance and reboot it, once it's up exit maintenance and wait for it to rebalance.