Juniper virtual chassis member replace

From Wikitech

This page describes how to replace a member QFX or EX switch configured in a virtual-chassis.

Juniper process is described in more detail here:

https://www.juniper.net/documentation/us/en/software/junos/virtual-chassis-qfx/Other/virtual-chassis-ex4200-member-replacing-cli.html

Assumptions

This process assumes that the switch being replaced is dead, or already powered down.

It also assumes that the replacement switch is using the correct version of JunOS matching other devices in the virtual-chassis. If not it should be upgraded (not described here).

Procedure

Prep Member Switch

  1. Rack the switch and connect to serial console server.
  2. Connect to device over console and zeroize to remove any existing configuration:
    • CLI: "request system zeroize"
    • Device should reboot after this command has been issued
  3. When system reboots check for normal signs of health:
    • CLI: "show chassis environment"
    • All parameters show should 'OK' status
    • CLI: "show version"
    • "Junos" version reported should exactly match what is reported for other switches in VC.
  4. Set VC mode to "mixed" (allowing different models to be part of same VC)
    • CLI: "request virtual-chassis mode fabric mixed"
  5. Set ports that will be used to connect to other switches to virtual-chassis mode
    • These ports will be listed in netbox something like vcp-2/0/49.
      • In this case '2' is the member switch number
      • '0' is the pic-slot number
      • '49' is the port number.
    • For each "vcp" port on the switch issue the following command in ops mode (serial console to member switch still)
    • CLI "request virtual-chassis vc-port set pic-slot <pic-slot> port <port>"
  6. Reboot the switch
  7. Check all looks good [unsure about what commands / what status we should see on the member switch here at this stage]
  8. Verify serial number of replacement device
    • CLI: "show system license"
    • Update serial recorded in Netbox for the member device to this

Prep virtual-chassis config

  1. Connect to virtual-chassis master via SSH / MGMT.
  2. Update the serial number for the member device to that of the replacement switch
    • CLI (Config mode): "set virtual-chassis member <member_number> serial-number <serial>"
  3. Cable the VC-ports on replacement switch to other VC members as per Netbox
    • New switch should be detected and added to the virtual-chassis, check:
      • CLI: "show virtual-chassis" (status should be 'present')
      • CLI: "show virtual-chassis vc-port" (all port status should be "Up")
      • CLI: "show virtual-chassis device-topology" (all ports 'Status' should be 'Prsnt')
      • <more commands here???>


Enable revenue ports

At this point we can re-enable ports that had been disabled in Netbox ("enabled" flag cleared on them), bringing those back up on the newly replaced member switch and reconnecting adjacent servers, routers etc.

Tick the 'enabled' flag for all such ports in Netbox, and run homer as usual against the virtual-chassis hostname. Homer should add the required configuration for all ports re-enabled. Validate when complete ports are up, errors are not present etc. Verify connectivity to newly reconnected endpoints are normal, validate with service owners if things look ok to them etc.