03-31-2014 08:06 AM - edited 07-05-2021 12:35 AM
Hi all,
We are facing a strange problem at a client's wireless solution.
Solution with 4x WLCs 5508, running v7.3.112.0, spread across 2x DCs. All 4x are active (no HA) with two mobility groups. The SSIDs are associated with a group interface (vlan select). All configurations are identical, being the IP addresses and hostname more or less the only configurations that change between them. The LAN is constructed pretty much the same way on both DCs and there are no STP loops (all interfaces forwarding).
Recently we activated two external DHCP servers. On 2x of the WLCs we have an issue: for each DHCP request packet from the clients (DHCP DISCOVER) thousands of packets arrive at the dhcp server. And I really mean thousands ![]()
The captures from the dhcp servers show that the “hops” field in the dhcp header is increasing every dozen or so packets which means that the packet is relayed more than once but the relay agent IP is always the same (VLAN interface @ WLC). Every time the DHCP hops is increased, there is the addition of one option 82 field so the packets are getting larger and larger until they achieve and impressive 1500 bytes.
The WLC / DHCP don’t cope with this rate and at a certain point the interface is marked as "dirty" because no answers are received from the dhcp server and this stops the “loop”.
A thorough search on bug tool and the web didn’t retrieve any relevant information.
Rings any bell?
Regards,
Telmo Oliveira
03-31-2014 10:17 PM
DHCP proxy enabled or disabled ?
Have you configured primary & secondary DHCP server information in all dynamic interfaces of your WLCs ?
Also what's the reason using DHCP option 82 ? Is it something can you disable on WLC & test.
Also to what hardware (Nexus ? ) these WLC connnect to ?
On a side node your running 7.3.x software is defferred. Better to move onto 7.4.121.0 code.
HTH
Rasika
*** Pls rate all useful responses ***
04-01-2014 04:48 AM
Hi,
Thank you for the quick reply.
Answering to your questions: DHCP proxy is enabled, pri and sec dhcp configured on all dyn interfaces, wlcs connect to c3750G-12S (soon to be replaced) and 7.3 will be upgraded soon.
It is my pleasure to announce that the root cause was found and a workaround was implemented.
Root cause
For some unknown reason, the switch where both WLCs are connected (c3750G-12S) was timing out the entries in the CAM relating to the DHCP servers, even when receiving frames/packets from them.
Behavior description
When the wlcs generate dhcp packets to the dhcp server, they send it to the switch and the switch, without a known mac on the cam, was flooding the frame to all interfaces. The frames flooded to the uplink would reach the dhcp server eventually but some frames were injected back to the wlc, strangely enough. When receiving these frames, the wlc would process (adding another option 82 field and incrementing the “hops” field on the dhcp header) and relay them to the dhcp server, back to the switch which would feedback the whole process.
Workaround
Static mac entries were configured for both dhcp servers’ macs pointing to the uplink
Follow ups
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide