Today I managed some lab time where I was able to sit down with Tavita to try and sort out DHCP with the topology. We had connected all switches and DHCP server, and configured them as per the config files we had stored and worked on getting DHCP addressing to host PCs on each layer 2 segment. At first we were unsure why the hosts were not receiving addresses, but it became clear once we started to use 'debug ip udp' and 'debug ip dhcp server packet' commands as to why it was not allocating addresses.
Dynamic Host Configuration Protocol (DHCP)
Basically, our topology did not give a method of returning the addresses to the network the DHCP request originated from. An easy fix for this issue was to apply the 'redistribute connected subnets' command to the 'router ospf 1' instances on BAS1 and BBS1. Currently, we have not set up HSRP, but I have already created a command list to enter on each router in order to create an instance of Hot Standby Router Protocol for each VLAN.
In order to get our DHCP working with host computers, the connected networks had to be redistributed into OSPF. The alternative would be to select the specific networks we are using 'network XXX.XXX.XXX.XXX YYY.YYY.YYY.YYY area Z' statements, but it was easier to add the 'redistribute connected subnets' to each border multilayer switch. The configuration for the OSPF protocol is now as follows
##BAS1
router ospf 1
log-adjacency-changes
redistribute connected subnets
passive-interface port-channel 4
passive-interface port-channel 5
network 10.0.2.0 0.0.0.63 area 1
##BAS2
router ospf 1
log-adjacency-changes
redistribute connected subnets
passive-interface port-channel 4
passive-interface port-channel 5
network 10.0.2.0 0.0.0.63 area 1
The above configuration is exactly the same as BBS1/BBS2, however the port-channel group numbers are different. At this time, I am unsure of whether the redistribution of connected subnets will be affected once HSRP is active, however further testing will show whether problems will arise.
An issue we did run into when configuring the DHCP, is the ability for DHCP clients to send ICMP packets (or pings) to the DHCP server once they had received addresses. This issue is due to the way routing was performed, and the fact that we were splitting a single DHCP pool between two local area networks. The routing tables suggested that the first place to send packets from the 10.2.0.0 network was BBS1, as it was the first configured destination for the 10.2.0.0 network. What ended up happening was a test DHCP client on the Building B network was able to freely ping the DHCP server, however when you tried to ping the DHCP server from a Building A client, it would respond with "destination unreachable". In order to remedy this issue, we are creating two pools to suit each group of host computers. Each pool will allocate addresses to a single site only, as the routing protocol shouldn't have issues when sending ICMP packets. An updated DHCP pool configuration is as follows:
ip dhcp pool STUDENTS-A
network 10.2.0.0 255.255.0.0
default-router 10.2.255.254
domain-name STUDENTS
!
ip dhcp pool STUDENTS-B
network 10.3.0.0 255.255.0.0
default-router 10.3.255.254
domain-name STUDENTS
!
ip dhcp pool STAFF-A
network 10.1.0.0 255.255.128.0
default-router 10.1.127.254
domain-name STAFF
!
ip dhcp pool STAFF-B
network 10.1.128.0 255.255.128.0
default-router 10.1.255.254
domain-name STAFF
!
ip dhcp pool REGISTRY
network 10.0.0.0 255.255.255.128
default-router 10.0.0.126
domain-name REGISTRY
!
ip dhcp pool PAYROLL
network 10.0.0.128 255.255.255.128
default-router 10.0.0.253 10.0.0.254
domain-name PAYROLL
!
ip dhcp pool GUEST-A
network 10.100.0.0 255.255.128.0
default-router 10.100.127.254
domain-name GUEST
!
ip dhcp pool GUEST-B
network 10.100.128.0 255.255.128.0
default-router 10.100.255.254
domain-name GUEST
Hot Standby Router Protocol (HSRP)
HSRP allows us to have redundant default-gateways for our host computers to use to exit their local area network. If say the switch BAS1 were to come down due to an error, host computers will still have the ability to exit the network as the gateway will still be active, instead choosing the secondary backup gateway, BAS2. The same effect will be seen on BBS1/BBS2.
The following command list is what I'll be using to configure HSRP on the VLANs that require a redundant default-gateway for VLAN 10 (STUDENT), 20 (STAFF) and 100 (GUEST)
interface vlan 10 ##STUDENT
ip address 10.2.255.252 255.255.0.0
standby version 2
standby 10 ip 10.2.255.254
ip helper 10.0.2.22
!
interface vlan 20 ##STAFF
ip address 10.1.127.252 255.255.128.0
standby version 2
standby 20 ip 10.1.127.254
ip helper 10.0.2.22
!
interface vlan 100 ##GUEST
ip address 10.100.127.252 255.255.128.0
standby version 2
standby 100 ip 10.100.127.254
ip helper 10.0.2.22
The 'standby version X' command determines the version of HSRP to be used on that instance. I am using version two of HSRP because:
- It learns and advertises millisecond timer values, which helps stability of the hsrp group.
- It's group number range is larger than in version one, allowing up to 4095 groups instead of 255.
- HSRP packets have changed to include a 6-byte identifier field which has management benefits such as allowing an administrator to identify the sender of the packet.
- The IP multicast address has changed from 224.0.0.2 in version one to 224.0.0.102 in version two. Version one's multicast address conflicted with CGMP (Cisco Group Management Protocol), which the change now allows CGMP and HSRP to operate without issue.
