I was performing server reboots and manually restarting the Dgraph process. When starting up the Dgraph Alpha process, I accidentally mistyped the my=ipaddress. The Dgraph process connected to the Zero, but the other Alphas could not communicate with the Alpha instance. I realized the problem, stopped the dgraph process, and re-started the Dgraph Alpha with the correct IP address. The cluster seemed to be fine again the Alphas appear to be communicating without a problem.
However, I noticed when restarting the Zero instances, they were still trying to connect to that incorrect Alpha address. The bad IP address does not appear in the Cluster state details, but somehow they’re still aware of it, resulting in a failed connection attempt. Is there a way to clear the bad IP address out?