The SoS tool takes periodic backups of the Cloud Foundation racks in your environment. After you insert the imaged management switch in the physical rack, you can restore the backup configuration on the switch.

Prerequisites

Retrieve the following files.

  • Backup file of the failed management switch's configuration. This file is named cumulus-192.168.100.1.tgz.

  • The hms.tar.gz backup file of the rack on which the management switch is to be replaced.

For the location of these file within the SoS tool's output, see Back Up ESXi and Physical Switch Configurations.

Procedure

  1. Retrieve the password of the management switch to be replaced and note it down.
    1. SSH to the SDDC Manager Controller VM.
    2. Run the following command.

      #/home/vrack/bin/lookup-password

  2. Unplug the management switch you are replacing.
  3. Note down the current connections to TOR switches, rack-interconnect switches, and hosts in the rack.
  4. Disconnect all host OOB connections to management switch.
  5. Remove the management switch from the rack.
  6. Use PuTTY to log into the management switch IP address 192.168.100.1 with username cumulus and password CumulusLinux! .
  7. Add the following line to the end of the /etc/dhcp/dhcpd.config file:

    ping-check false;

    If this line already exists in the file, leave it as is.

  8. Use WinSCP to copy the hms.tar.gz and cumulus-192.168.100.1.tgz files to the /home/cumulus directory of the new management switch.
  9. Use PuTTY to log into the management switch IP 192.168.100.1 with username cumulus and password CumulusLinux! .
  10. Type the following command.

    sudo su

  11. Restore the backup configuration to the new switch.
    1. Change to the root directory.

      cd /

    2. Unpack the contents of the hms.tar.gz file.

      tar zxvf /home/cumulus/hms.tar.gz

    3. Unpack the contents of the cumulus-192.168.100.1.tgz file.

      tar zxvf /home/cumulus/cumulus-192.168.100.1.tgz

      The /etc/dhcp/dhcpd.conf file is restored.

    4. To ensure that new hosts added to the rack are not assigned IP addresses reserved for existing hosts, edit the DYNAMIC LEASES section of the dhcpd.conf file. For example, if you have 8 servers in the rack, specify the range as follows.
      # DYNAMIC LEASES
      option routers 192.168.0.1;
      range 192.168.0.60 192.168.0.100;

      If you have 15 servers in the rack, specify the range as follows.

      # DYNAMIC LEASES
      option routers 192.168.0.1;
      range 192.168.0.70 192.168.0.100;

    5. Navigate to the dhcpd.leases file and note down the host_id, oob ip, and idrac mac address mapping.
  12. Install the replacement management switch into the rack and wire it according to the wiring connections of the previous switch. Refer to your notes from step 3.
  13. Change the password of the new management switch to the current password for your Cloud Foundation system's management switches, as obtained from the lookup-password command.
  14. Reboot the new management switch.
  15. The Physical Resources > Rack Details page on the Dashboard displays a message Error loading rack details. Follow these steps to resolve this error.
    1. Collect the following information.

      Information Required

      Procedure

      SDDC Manager Controller VM IP address

      Log in to vSphere Web Client and note down the 192.168.x.x address for the SDDC Manager Controller VM

      PSC IP address

      Log in to vSphere Web Client and note down the PSC1 VM IP address

      Single Sign On and management switch passwords

      See Look Up Account Credentials.

    2. Using the root account, SSH to the SDDC Manager Controller VM.
    3. Run the following script with Python 2.7.

      #/opt/vmware/bin/python2.7 /opt/vmware/sddc-support/fru-mgmtsw.py

      Wait for the script to complete.

  16. Reboot the management switch and the SDDC Manager (VRM) VM.
  17. Confirm that the switches and hosts are displayed correctly on the Physical Resources page.