cRIO failing to get a DHCP assigned address

[ErikL68]

I just went through heck with my IT deparment on this so thought I'd post it here.

I had four cRIO-9073s on a remote switch in one of our labs, that MAX (Measurement and Automation Explorer) was saying were using Link-local addresses.  In fact, all were using the EXACT SAME address:  169.254.62.215.  Very strange!  I was able to talk to them using MAX some times, doing remote restarts, etc., but other times not.  All were configured to grab a DHCP address or fall back to link-local.

As it turns out, the RIOs were attached to a Cisco switch that has a feature callled STP fully enabled on all ports.  STP is designed to prevent accidental loops in the network from downing your network, and that feature blocks all initial transmissions from attached devices for 30 seconds before letting them through, while it "listens" and "learns" about your device, analyzing to make sure that letting it connect won't create a loop.  Apparently this 30 second delay (forever in computer terms!) is too long for the cRIOs, so they take a link-local address, and since they can't even check the network for other link-local users because their transmissions aren't being forwarded, they all take the exact same address.  I'm still amazed MAX can find them under these conditions! 

The way to get it working is to turn on "PortFast" on the Cisco switches, which simply disables the steps in STP that cause the slowdown (blocking, listening, learning).  Fortunately this can be done on a port-by-port basis, as it's recommended to NOT turn it on if ports are going to another switch (as that then opens the door for a loop).

Here's more info:

http://www.cisco.com/en/US/products/hw/switches/ps708/products_tech_note09186a00800b1500.shtml

Here's hoping you don't need this!

[Frisbee]

Hello Erik,

Thank you very much for the post. Great work figuring out the root cause of your communication issue!