I am having a DHCP related issue. Please help.
Our main office has a SN2100 (core-02, Onyx 3.7). We have a few VLAN and there is a single DHCP server (dhcpd on CentOS, Host IP 10.36.24.2) handle multiple subnet. For example 10.36.19.0/24 for VLAN19. The subnet for main office are all in 10.36.0.0/16. Everything works fine for now.
Now we are expend to a new site in a different location – let’s say Colo01. I set up a Mellanox SB1012 (core-01, Onyx 3.6) and connected it to main office SN2100 via a dark fiber trunk connection. In this Colo01 we use subnet 10.38.0.0/16. And there we also have multiple VLAN just like main office. have I built a DHCP server (IP: 10.38.24.2) in Colo01 to handle all local DHCP requests and assign to clients in the subnet 10.38.X.y. (here X is the VLAN ID as well).
The idea is let the local DHCP server handle it’s own local DHCP requests. Now when I test the Colo01 setup, I see that the DHCP broadcast request reached on both DHCP server all the time! I think the reason is that on both sites we have the same VLAN setup so the DHCP request will travel to reach all the DHCP server.
I checked on both switches and I have the following setting:
core-02 (mail office)
core-02 [standalone: master] # show ip dhcp relay
Instance ID 1:
VRF Name: default
DHCP Servers:
10.36.24.2
DHCP relay agent options:
always-on : Enabled
Information Option: Disabled
UDP port : 67
Auto-helper : Enabled
core-01 (Colo01)
core-02 [standalone: master] # show ip dhcp relay
Instance ID 1:
VRF Name: default
DHCP Servers:
10.38.24.2
DHCP relay agent options:
always-on : Enabled
Information Option: Disabled
UDP port : 67
Auto-helper : Enabled
The above setting seems to me are correct. It relay the DHCP request to its local DHCP server respectively. But when I test it, for example starting up a PC in VLAN19 at Colo01, some how is reaches DHCP server on main office so it get a IP of 10.36.19.x, which is not as planned.
Another thing I noticed is that when I monitor on both DHCP server with tcpdump, I only see requests from Colo01 to arrive on DHCP server at main office, not the other direction. I don’t understand what happened on the SN1012 at Colo01.
Anyway may I get help so I can stop the DHCP request broadcast across sites on the Colo01 switch(core-01)?
Thank you in advance.
Jian