FieldPoint Family

cancel
Showing results for 
Search instead for 
Did you mean: 

cFP-2220 DHCP fails when other devices do not

We have a strange DHCP situation with a cFP-2220;

 

There is a BOOTP server on another subnet. The cFP sits on a layer 3 port on a router - and the router forwards the DHCP dicover from the cFP-2220. The BOOTP server receives it, and sends back an offer. The router agent turns it into a unicast and the offer arrives at the router, but still the cFP-2220 acts as if it never sees a reply.

 

If we put a PC or other DHCP enabled device in the cFP-2220's place they get their address fine...Which should indicate that there is something about the network the cFP-2220 is not handling. BUT - if we introduce a switch between the router and the cFP-2220 (to try to sniff the traffic) - the cFP-2220 magically starts to get its dynamic address as it is supposed to(!).

 

Any clue as to what could cause this? Is there a difference in how the cFP-2220 treats DHCP (problem with unicast? I'm thinking about sections 3.1.1 and 4.1.2 in RFC 1542...  But why would the switch affect things?).  If we set the cFP to use static address we can communicate with it without the switch.

0 Kudos
Message 1 of 3
(6,205 Views)

Ok, so no answers here yet...but we've figured out what the cause is, and it might be of interest to others:

 

The DHCP client on cFP-2220 is simply too quick and impatient...the second the device gets power it sends out 4 DHCP DISCOVER messages in rapid (1 sec) succession. This happens so quickly that the routers on the network have not yet reached a state of convergence, so the offers from the DHCP server are dropped. A PC does this much slower and continues to send DISCOVER so the PC works fine. Adding a switch also fixes the issue because this keeps the interface seen by the router up before the interface on the cfP-2220 side...

 

Originally we were using LabVIEW RT 2011 SP2, so we got a tip over the phone to try 2012 as it is supposed to put 1,2 and 4 second delays between the discover packets. We upgraded and we now see 1,1 and 3 seconds between the packets. The last one is just slow enough for it to work in this setup...but we should really be able to increase it a bit more, or the first two should at least be a bit slower. So the question now is; can NI provide a solution for this?

Message 2 of 3
(6,154 Views)

Thanks for the follow-up.  This explains a recent problem we had where a cFP-2010 would sometimes startup up properly, sometimes not, solved by adding a switch.

 

Matt

0 Kudos
Message 3 of 3
(6,142 Views)