Vision Client not finding master gateway after failover

When I force a failover from the master gateway webpage, the vision client is able to find the backup gateway with no issue. However, when I have the master gateway regain control, the vision client is unable to find it until I restart the client. I get the “Gateway connection lost” popup, but it says that both nodes are available and inactive. The IP address is shown as identical on both (the IP address shown belongs to the backup) so that makes me think that the client doesn’t know to change which IP it’s looking at. The gateways themselves seem to have no issue failing back and forth, judging from the status webpages.

We recently upgraded from 7.9 to 8.1.
In our case it would occur when testing redundancy. The Vision Client presented the same behavior as described here when failing over to the backup gateway and when giving control back to the Master. It also occurred when a network interruption was emulated: disconnect client host from the network and reconnect it. Check the following setting on both gateways which should be set to true (default):
Config -> Networking -> Web Server -> Public HTTP Address -> Auto Detect HTTP Address
In our case the Master was set manually with the backup ip address hence the error/behavior.