After all the agents have been upgraded we check that all of the agents can reach the controller. If there are enough agents some of the connections will get try again errors. We should try again when that happens (with exponential backoff?) rather than just reporting that the upgrade failed.
After all the agents have been upgraded we check that all of the agents can reach the controller. If there are enough agents some of the connections will get
try again
errors. We should try again when that happens (with exponential backoff?) rather than just reporting that the upgrade failed.