There are several customer-side failure modes Sauce Connect can encounter, based on changing tunnel hosts, routing and load balancing peculiarities, and high availability setups:
Load Balancers may direct some test traffic to faulty test instances, or instances with different code versions
SC may sometimes be started on a machine using different DNS or Routing, leading to different SUTs (Systems Under Test)
This can impact unidentified or singly-identifier tunnels if the tunnel host machine changes between tests.
It can be more endemic when using High Availability tunnels, as individual SC Hosts may route to different SUTs. As tests are assigned to tunnels in a HA pool individually, this leads too weird, intermittent failures. From the customer's POV, they are using "the same tunnel", even though the network environment for each tunnel in the pool can be quite different.
Include the Hostname of the SC machine in the Job Details page. This would make it easier for customers and support staff to spot changes if a SC tunnel changes location
Include an advisory message in builds if the tunnels used by tests in that build suddenly change. This would be more useful for single tunnels, rather than pools.
Include an advisory message on tests if the SC hostname is usually one thing, and suddenly becomes another. Something like "The network path to the SUT has changed. This may influence your test results."
Include similar messages if the SUT's routing or IP changes.
Have HA tunnels compare routing information for hosts, ensuring they route to the same (set of) instances at a given point in time. Optionally, allow users to configure allowable SUT Addresses and raise an error if others are requested/resolved
Return IP address information for SUT hostnames to Sauce Labs for analysis purposes. This may also improve Performance Testing.
Store IP address information for SUT hostnames locally to determine if sudden changes occur.