About Controller Clusters : Troubleshooting cluster connections
  
Troubleshooting cluster connections
These situations can cause a controller to become disconnected from the cluster:
• The cluster that your controller is connected to has become unreachable for some reason.
• The trust settings on your controller or the peer to which you’re connected have changed and no longer match. Check your SSL settings; see Basic steps for configuring HTTPS/TLS proxy support.
If your controller is disconnected from the cluster, and attempts to reconnect are denied, detach and rejoin the cluster. For details, see Joining a controller to a cluster.
Make sure that you have your logs configured at Error level. Cluster error messages appear at this level. For details on filtering log messages, see Viewing and downloading logs.
These topologies can cause problems with Client Accelerator connectivity:
• Firewalls between the endpoint and the controller—To manage the controller more easily, be sure to open the firewall to allow access to ports 22, 80, 443, and 7870.
• Mixed mode clustering—In this topology, the controllers use different versions of the software. Mixed mode clustering can occur when not all the controllers are updated to the latest software release.
Making policy, configuration, and cluster changes in mixed mode can be challenging. Therefore, we recommend that all the controllers be updated to the same version of the software.