Spanning Tree / DHCP

[mo_zubair01]

We have around 10 separate buildings all connecting back to the main server room over fibre connections. In each building on average we have 10 access switches and 1 Edge switch (Distribution layer but routing is done on our core switches) which we trunk all our access switches to and have 2 fibre connections from our edge switches to each core switch which is operating HSRP between the 2 core switches (Core 1 Active, Core 2 Standby).

The 2 core switches have been configured with rapid spanning tree with Core01 being the root switch and Core02 being the secondary. Both core switches have DHCP configured on them to hand out DHCP addresses. My thinking is that if we have spanning tree configured, the 2 trunk links we have coming from our edge switches from other buildings back to our core switches, one of them must be in blocking mode or a loop will occur. The link to Core1 as it being the root switch must be in forwarding mode and the link to Core 2 is in blocking mode. So that means all traffic is going through to Core 1. My question is if that is the case then how is Core 2 handing out DHCP addresses when the all traffic is going through Core 1 first. Surely Core 1 should be handing out DHCP addresses because the DHCP request was first received on Core 1?

Kind Regards

Zubair

[Wilson Bonilla]

Hello Zubair.

So if I understand correctly your topology is something like this:

rootsw             secrootsw

   Core1 ==== Core2

           \           /blocking port

     DISTRIBUTION SW

      |         |          |

site1   site2    site3

DHCP users Discover, Offer, Request and Acknowledge. Remember that when a host in site1 sends a Discover packet this packet has as a destination the broadcast address of the network where it belongs to, because this is a broadcast it will go all the way up to Core1 and Core1 will forward the broadcast across all interfaces in the same broadcast domain reaching finally  Core2. Core2 replies back with a Offer and the negotiation goes on...

Should be handing out DHCP addresses because the DHCP request was first received on Core 1?

Not necessarily, what if the the link between distribution and core1 goes down, will you move your dhcp pool back to core2 just because now it hits core2 first.... ? ... anyways due to this is a broadcast traffic it will still hit core1 and core2 everytime there's a request.

Regards.

Wilson B.