See the below example you understand why helper address should be given on all routers.Actually if we give on all routers,They would be no duplicate packets created,Otherwise request would from all the routers on which ip heper address is not used.
Link for Below Solution
https://blog.initialdraft.com/archives/2239/
Credit - Daniel of Network Faculty. - https://blog.initialdraft.com
Think of this exam situation: You have a “client” (R1) behind two routers where you have to put HSRP (R2 and R3), and this client is going to ask an IP address via HSRP from a server on another network (R4). You may be happy because you finally got an easy task. But nothing comes easy in the CCIE exam, later in the task you are asked to avoid any duplicate/redundant request to the DHCP server. And now you got yourself into trouble.
When you have a helper address configured on the HSRP interfaces on the same broadcast domain, both devices are going to forward the DHCP request to the server no matter what device has the active HSRP state. If you think about it, there is nothing in the configuration that ties the HSRP state to the helper-address.
Lets see this on a typical configuration.
R2
interface FastEthernet0/0 ip address 10.0.123.2 255.255.255.0 ip helper-address 10.0.234.4 ip ospf 1 area 0 duplex auto speed auto standby 1 ip 10.0.123.254 standby 1 preempt
R3
interface FastEthernet0/0 ip address 10.0.123.3 255.255.255.0 ip helper-address 10.0.234.4 ip ospf 1 area 0 duplex auto speed auto standby 1 ip 10.0.123.254 standby 1 priority 90 standby 1 preempt
R4
ip dhcp excluded-address 10.0.123.0 10.0.123.50 ip dhcp pool TEST network 10.0.123.0 255.255.255.0 default-router 10.0.123.254
I turned on debugging for dhcp server packets and events on R4 to have as much information as possible from the situation and configured R1 to ask for the IP via DHCP (ip address DHCP):
*Mar 1 00:24:45.347: DHCPD: Adding binding to radix tree (10.0.123.52) *Mar 1 00:24:45.347: DHCPD: Adding binding to hash tree *Mar 1 00:24:45.347: DHCPD: assigned IP address 10.0.123.52 to client 0063.6973.636f.2d63.3230.302e.3133.6463.2e30.3030.302d.4661.302f.30. *Mar 1 00:24:45.351: DHCPD: Sending DHCPOFFER to client 0063.6973.636f.2d63.3230.302e.3133.6463.2e30.3030.302d.4661.302f.30 (10.0.123.52). *Mar 1 00:24:45.351: DHCPD: unicasting BOOTREPLY for client c200.13dc.0000 to relay 10.0.123.3. *Mar 1 00:24:45.355: DHCPD: Sending notification of DISCOVER: *Mar 1 00:24:45.355: DHCPD: htype 1 chaddr c200.13dc.0000 *Mar 1 00:24:45.355: DHCPD: remote id 020a00000a00ea0400000000 *Mar 1 00:24:45.355: DHCPD: circuit id 00000000 *Mar 1 00:24:45.355: DHCPD: DHCPDISCOVER received from client 0063.6973.636f.2d63.3230.302e.3133.6463.2e30.3030.302d.4661.302f.30 through relay 10.0.123.2. *Mar 1 00:24:45.359: DHCPD: Seeing if there is an internally specified pool class: R4# *Mar 1 00:24:45.359: DHCPD: htype 1 chaddr c200.13dc.0000 *Mar 1 00:24:45.359: DHCPD: remote id 020a00000a00ea0400000000 *Mar 1 00:24:45.359: DHCPD: circuit id 00000000 *Mar 1 00:24:45.359: DHCPD: Sending DHCPOFFER to client 0063.6973.636f.2d63.3230.302e.3133.6463.2e30.3030.302d.4661.302f.30 (10.0.123.52). *Mar 1 00:24:45.363: DHCPD: unicasting BOOTREPLY for client c200.13dc.0000 to relay 10.0.123.2. *Mar 1 00:24:45.527: DHCPD: DHCPREQUEST received from client 0063.6973.636f.2d63.3230.302e.3133.6463.2e30.3030.302d.4661.302f.30. *Mar 1 00:24:45.527: DHCPD: Sending notification of ASSIGNMENT: *Mar 1 00:24:45.527: DHCPD: address 10.0.123.52 mask 255.255.255.0 *Mar 1 00:24:45.527: DHCPD: htype 1 chaddr c200.13dc.0000 *Mar 1 00:24:45.531: DHCPD: lease time remaining (secs) = 86400
As you can see we are getting duplicate request/reply for the R1 initial attempt to obtain an IP address. If you are not aware of this in the exam you just lost “easy” points, and even if you know what is happening, you may not know about a cool feature we can put in place avoid this and fulfill the requirements.
This feature (UDP Forwarding Support for IP Redundancy Virtual Router Groups) ties together the active state on the HSRP to the helper address. In other words, only the HSRP group interface with the active state forwards packets to the DHCP, the other device will not forward packets even when it has the helper address configured. In our case only if R3 becomes the active device is going to start forwarding requests to the DHCP server (R4).
The configuration is very simple, and all we have to change is this:
R2
interface FastEthernet0/0 ip helper-address 10.0.234.4 redundancy CCIE standby 1 name CCIE
R3
interface FastEthernet0/0 ip helper-address 10.0.234.4 redundancy CCIE standby 1 name CCIE
As you can see I configured a name for the HSRP group, and tied it with the “redundancy” option on the helper-address. You can avoid the configuration of the name by looking at the default name provided by the router with the “show standby” command and look for the “Group name”.
And now on the the DHCP server, we only see the traffic forwarded by R2 (Active HSRP state), when R1 ask for a IP.
*Mar 1 00:32:20.623: DHCPD: Adding binding to radix tree (10.0.123.53) *Mar 1 00:32:20.623: DHCPD: Adding binding to hash tree *Mar 1 00:32:20.627: DHCPD: assigned IP address 10.0.123.53 to client 0063.6973.636f.2d63.3230.302e.3133.6463.2e30.3030.302d.4661.302f.30. *Mar 1 00:32:20.627: DHCPD: Sending DHCPOFFER to client 0063.6973.636f.2d63.3230.302e.3133.6463.2e30.3030.302d.4661.302f.30 (10.0.123.53). *Mar 1 00:32:20.627: DHCPD: unicasting BOOTREPLY for client c200.13dc.0000 to relay 10.0.123.2. *Mar 1 00:32:20.847: DHCPD: DHCPREQUEST received from client 0063.6973.636f.2d63.3230.302e.3133.6463.2e30.3030.302d.4661.302f.30.
