Had an interesting one the other day – a bunch of clients at a new site were randomly picking up lease times of 5 minutes for no apparent reason. Out of 10 clients, maybe 2 were failing – sometimes it was as high as 5 clients. All the others would pick up their correct lease times. There seemed to be no rhyme or reason to the problem looking at the client side. The behavior exhibited 5 minutes of connectivity, followed by a few lost packets then another 5 minutes of connectivity, followed by a few lost packets, etc.
We tore down the clients – nothing. Checked the primary DHCP server – nothing. It wasn’t until we look a look at the secondary “failover” DHCP server that we noticed that the scope for that site was disabled. Once enabled, everything worked as expected.
It turns out that when you have Microsoft DHCP configured with failover, it will first hand a client a temporary lease of 5 minutes (5 minutes is the default setting for Max Client Lead Time). There will then be a decision made on which server will actually maintain the registration of the client – if the scope is disabled on the secondary server it will not be able to register and begin its 5 minute registration cycle again.
Interestingly, when creating this particular scope and configuring failover, the failover dialogue box did say that it had successfully activated the scope on the failover node. Clearly this was not the case.
Moral of the story – be sure to check the failover server when configuring new scopes to confirm.