Before describing my problem I'd just like to say that I've read through most of the similiar threads here on the forums (there are quite a few, yes), but the other threads either doesn't solve the problem or isn't connected to my problem.
The Setup:
-Pure WDS on Windows Server 2012, no SCCM or anything else.
-No subnets
-One DHCP, without configuring options 66 and 67.
-WDS configured to respond to all clients.
My problem is that after about 24 hours, my Windows Server 2012 and it's WDS stops responding to PXE-boots and the only way to solve the issue is to restart the WDS server itself. I can think of nothing that triggers this. It happens every day, regardless of if if I'm using the WDS or not. Server shows no other symptoms of being ill, and is fully functional in all other ways. When trying to PXE-boot, this would happen after a long timeout:
PXE-E53: No boot filename received
PXE-M0F: Exiting Broadcom PXE ROM
What's even more confusing is that on some machines, e.g. virtual Hyper-V Machines (on server 2008R2 SP1) would sometimes still be able to PXE-boot, even though it is very very slow before showing up the boot menu (20 seconds+). Restarting the WDS server will make these virtuals find the WDS very fast during boot, and all physical machines will find the WDS-server imediately aswell.
Simply restarting the WDS-service doesn't help. Disabling firewall doesn't solve it either. WDS shows no errors in the logs. The PXE-E53 error is as I understand the clients reaction when it finds a DHCP, but doesn't get an answer from PXE-server in the network? Is there any way to track why PXE didn't respond to (some of the) clients? Any solution on this? I can't keep on rebooting the server like this since I want this put into production.