Maintain TPVM Versions After a Rollback in a Multi-Node Deployment

Both nodes in a multi-node deployment should have the same version of TPVM after an upgrade.

About this task

This procedure addresses a scenario in which TPVM2 (on SLX2) was upgraded, but TPVM1 (on SLX1) was rolled back to a previous version because of a failure during upgrade. To maintain the same version of TPVM on both nodes, you must downgrade, or roll back, TPVM2.

In this procedure, SLX1 and TPVM1 refer to the standby EFA node. SLX2 and TPVM2 refer to the active EFA node. This procedure references TPVM versions 4.2.4 and 4.2.5 for clarity in examples.

Procedure

  1. From the SLX-OS command line on SLX2, stop and start TPVM to force a failover.
    device# tpvm stop
    
    device# tpvm start
  2. When EFA synchronizes after the failover, view the output of the following commands to ensure that both nodes are in their proper state.
    1. Run efa status to verify that both nodes are up.
    2. Run efactl status to verify that all pods on the active node are in Running state.
    3. Run efactl db-status to verify that the MariaDB is active (running).
  3. From the EFA command line on TPVM1 (the active EFA), upgrade TPVM.
    efa inventory device tpvm-upgrade execute --ip <slx-hosting-stby-efa> 
    --firmware-host <firmware-host-ip> --tpvm-image <image-path-on-host> 
    
  4. From the EFA command line, verify the TPVM upgrade process.
    efa inventory device tpvm-upgrade show -–ip <SLX2-IP>
    
  5. If the upgrade process (step 3) fails, take the following steps.
    1. Delete the TPVM on both SLX devices.
      device# tpvm uninstall force
      In the sample scenario, you are deleting version 4.2.5 from the upgraded device and deleting version 4.2.4 from the device on which the TPVM was rolled back.
    2. Install the earlier version of the TPVM on both devices.
      In the sample scenario, you are installing version 4.2.4 on both devices, so that both devices have the same version of TPVM.
    3. Install EFA on the TPVM.