Network interface starts "flapping" after reprovisioning of the VPX

[rein-tollevik]

If the 'netscalersdx_provision_vpx' resource is used to reprovision the cpu or memory configuration of an existing VPX its channel interfaces will end up in a "flapping" state where it continuously goes down and up again. The undocumented 'device_channel_name' property of the network_interfaces must be set to the same value as the 'port_name' property to avoid the "flapping" state, which is what the GUI does when it changes the configuration.

It is tempting to always set the 'device_channel_name' to the 'port_name' value in the provider, but I haven't tested it on VPXs that don't use channel interfaces. And its name indicates that is should only be used with channels.

# Read the existing VPXs to get the mac addresses of the network interfaces, as they are
# required when changing an existing VPX.
data "http" "ns" {
  url = "https://${var.sdx_hostname}/nitro/v1/config/ns"

  request_headers = {
    "X-NITRO-USER" = var.sdx_username
    "X-NITRO-PASS" = var.sdx_password
  }
}

locals {
  vpx = [for n in jsondecode(data.http.ns.response_body).ns : n if n.name == var.vpx_name] 
  mac = length(local.vpx) == 0 ? {} : { for n in local.vpx.network_interfaces : n.port_name => n.mac_address }
}

resource "netscalersdx_provision_vpx" "vpx" {
  ...
  network_interfaces = [
    {
      port_name           = "LA/1"
      device_channel_name = "LA/1"
      mac_address         = try(local.mac["LA/1"], null) # Value only required when changing an existing VPX
      ...
    },
    {
      port_name           = "LA/2"
      device_channel_name = "LA/2"
      mac_address         = try(local.mac["LA/2"], null)
      ...
    },
  ]
}