DISCO fails to pick up changes to switch configuration

[nkinkade]

DISCO currently determines which ports to monitor on startup. If the switch configuration changes after DISCO is already running, then DISCO will fail to pick up the changes. As long a we are using Juniper QFX5100 switches, or some switch with 48 ports, then this will likely only ever affect monitoring of the uplink port. The M-Lab standard is to connect 1g uplinks to port 47, and 10g uplinks to port 45. We recently had a case where a site (TNR01) upgraded our uplink to 10g, and moved the uplink to port 45 of the switch. On startup, DISCO discovered that it should be monitoring port 47, but somewhere along the way port 47 was uplugged and all traffic began flowing through port 45. It was necessary to delete the DISCO pods for TNR01 so that it could rediscover the correct uplink port to monitor.