Weird IP address conflicts

Hello all!

Running into a situation that I've literally never seen before.

Running multiple Windows 10 boxes in a air-gapped Server 2012r2/2016 environment. Win10 boxes are connected to a Cisco 3850 with ipbasek9 license (15.2 OS) via copper SFPs. All systems are given a static IP.

Whenever we boot a system, it comes up normally. When the system reboots, it tells us that there's an IP conflict, and it gets an APIPA address. We've verified that there actually is no IP conflict on the network. When we reboot, the systems go back to normal, no APIPA address, and they're happy with their statically assigned IP. We reboot again, and it goes back to IP conflict mode. Reboot again, back to good. Repeat... forever.

Digging in a bit further, I was looking into the event log, I see that on every machine, there's a 4199 event for the IP conflict. Now here's where it gets weird.

Every box says:

The system detected an address conflict for the IP address 0.0.0.0 with the system having the network hardware address of XX XX XX XX XX XX. Network operations on this system may be disrupted as a result.

The MAC address is the MAC addy for the switch port that the PC is connected to. Somehow, the systems are having an IP conflict with their own switchports.

We've tried:

• Adding HKLM\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters | DWORD "ArpRetryCount" | Value= 0

• Changing HKLM\SYSTEM\CurrentControlSet\Services\Tcpip\Parameters\Interfaces<Interface GUID> | DWORD "IPAutoconfigurationEnabled" | Value= 0

• Issuing the "ip device tracking probe delay 10" command to the switch

• Switching SFPs

• Resetting the interface

Any advice would be greatly appreciated, because this is just boggling to me.