Controllerless Networks

We have an Instant cluster with a virtual controller composed of 18 APs of various models. The whole cluster is running version 6.5.4.22 of Aruba OS (the latest all can run). The issue we are seeing is that users connected to our nonguest SSID, when running nslookup from a command prompt, it says all our servers (including all DCs) are non-existent. The name server it returns when you start the nslookup command is correct (it says that server is non-existent as well). Running the same tests with the users' computers connected to Ethernet works correctly and returns the correct responses in nslookup. There is nothing different about the network setup between WiFi and Ethernet. The SSID is setup for Employees. Client IP assignment is set to Network Assigned with Client VLAN Assignment set to Default. Access is set to Unrestricted. I have tried flushing the DNS cache on the users' computers and reregistering DNS but that doesn't help. I have also taken note of what AP the computers happened to be attached to and rebooted just that AP. That fixes the issue but it will reappear within several days. I have also rebooted the entire cluster, with the same result. Anybody have any idea what might be going on or how to fix it?

While you're experiencing this issue, is DNS the only thing impacted? Can you ping the gateway of its subnet? or something a hop or two away?
Have you double checked the VLAN assignments and .Q tags for all APs in the cluster? 

We have an Instant cluster with a virtual controller composed of 18 APs of various models. The whole cluster is running version 6.5.4.22 of Aruba OS (the latest all can run). The issue we are seeing is that users connected to our nonguest SSID, when running nslookup from a command prompt, it says all our servers (including all DCs) are non-existent. The name server it returns when you start the nslookup command is correct (it says that server is non-existent as well). Running the same tests with the users' computers connected to Ethernet works correctly and returns the correct responses in nslookup. There is nothing different about the network setup between WiFi and Ethernet. The SSID is setup for Employees. Client IP assignment is set to Network Assigned with Client VLAN Assignment set to Default. Access is set to Unrestricted. I have tried flushing the DNS cache on the users' computers and reregistering DNS but that doesn't help. I have also taken note of what AP the computers happened to be attached to and rebooted just that AP. That fixes the issue but it will reappear within several days. I have also rebooted the entire cluster, with the same result. Anybody have any idea what might be going on or how to fix it?

While you're experiencing this issue, is DNS the only thing impacted? Can you ping the gateway of its subnet? or something a hop or two away?
Have you double checked the VLAN assignments and .Q tags for all APs in the cluster? 

We have an Instant cluster with a virtual controller composed of 18 APs of various models. The whole cluster is running version 6.5.4.22 of Aruba OS (the latest all can run). The issue we are seeing is that users connected to our nonguest SSID, when running nslookup from a command prompt, it says all our servers (including all DCs) are non-existent. The name server it returns when you start the nslookup command is correct (it says that server is non-existent as well). Running the same tests with the users' computers connected to Ethernet works correctly and returns the correct responses in nslookup. There is nothing different about the network setup between WiFi and Ethernet. The SSID is setup for Employees. Client IP assignment is set to Network Assigned with Client VLAN Assignment set to Default. Access is set to Unrestricted. I have tried flushing the DNS cache on the users' computers and reregistering DNS but that doesn't help. I have also taken note of what AP the computers happened to be attached to and rebooted just that AP. That fixes the issue but it will reappear within several days. I have also rebooted the entire cluster, with the same result. Anybody have any idea what might be going on or how to fix it?