Recover from an Upgrade Failure

You can recover from an upgrade failure by rerunning the upgrade, by performing a fresh installation and then restoring the system from a backup, or by reverting a TPVM snapshot.

Procedure

  1. To rerun the upgrade, follow the steps for the type of upgrade you were attempting.
  2. To perform a fresh installation and restore the system backup, take the following steps.
    1. Uninstall EFA to remove any components that might have been installed before the upgrade failed.
    2. Follow the steps for the type of installation you need.
    3. Restore the EFA backup.
      efa system restore --backup-tar <filename>.tar.gz
      For more information about backup tar files, see the "EFA System Backup and Restoration" section of the Extreme Fabric Automation Administration Guide, 2.7.0 .
  3. (For SLX-OS 20.3.2a or later) To revert TPVM from a snapshot and then restore EFA in a multi-node deployment, take the following steps.
    1. On the TPVM, ensure that the status of both nodes is Up.
      efa status
    2. On the SLX device that hosts the standby TPVM, create a TPVM snapshot.
      tpvm snapshot create withefa true
      This step created a snapshot with all the data that EFA needs for restoring the database. Because this process stops the TPVM, you run it on the standby TPVM first.
    3. On the SLX device that hosts the active TPVM, create a TPVM snapshot.
      tpvm snapshot create withefa true
    4. Ensure that TPVM is running on both devices and that both nodes are Up.
      efa status
    5. Upgrade EFA as described in Upgrade EFA on TPVM in a Multi-Node Deployment.
      Tip

      Tip

      If this upgrade attempt fails, select No when prompted, to uninstall the upgrade. You can still continue with the next step.
    6. On each SLX device that hosts a TPVM, revert to the TPVM snapshot.
      device# tpvm snapshot revert
      The revert process can take about 10 minutes.
    7. Verify that both nodes are Up.
      efa status
    8. If both nodes are not up, take the following steps.
      • Run tpvm stop on both nodes.
      • Run tpvm start on both nodes.
      • Run efa status again.