Hi Mike, Responses inline:
> I understand the logic of tying the secondary scopes to the primary at ehe > DHCP side, however if the giaddr always reflects the primary subnet, how the > the DHCP server ever know to hand out addrs from the other secondary scopes? On the DHCP server, one configures the S1, S2, and S3 scopes to be "related" to the P scope. The DHCP server then knows there are four different subnets on this interface are related through P. When the DHCP server receives a DHCP DISCOVER with P as the giaddr, the above linkage indicates to the DHCP server that four subnets are on the same router interface. Without any additional logic, the DHCP server could randonly pick a free IP address from any of these four scopes, and send the selected IP in the DHCP OFFER. Note, that the DHCP server will send the DHCP OFFER (and ACK) to the giaddr IP (P). The router receives the DHCP packet, knows what interface it's asociated with (P), and forwards out the inteface accordingly. Also note that the DHCP server will likely also return other DHCP information in the OFFER including default gateway, subnet mask, DNS server IPs and domain information. The default gateway and subnet mask will be specific to the scope from which the IP was selected. Now, one could configure extra smarts into the DHCP server so that based on the device making the DHCP request, the DHCP server could assign the device an address out of one specific scope. Some devices will use DHCP Option 60 to inform the DHCP server of its device type. The DHCP server can be configured to use this information to help it select which of the scopes on the interface are applicable for this device. There are other mechanisms that can also be used by the DHCP server to help determine how to select which scope the DHCP request should be mapped to (device MAC address or OUI, DHCP Option 82, etc.) > This "feature" you describe sounds pretty worthless. If the giaddr is > always from P, and rotates through S1, S2, S3, etc when the DHCP server > doesn't respond, then unless your DHCP server is down or all IPs have been > allotted for subnet P, then the DHCP request will always result in an IP > from the scope for P. I think the idea for this cycling feature is as follows: If one wanted to assign multiple subnets onto an interface and these subnets are configured to have their IPs assigned via DHCP, then you have the problem discussed in this thread. I expect that there are some off-the-shelf DHCP servers which didn't have the ability to logically associate multiple subnets together. That is, the DHCP server had the limitation that each scope had to appear as if it was on its own interface. In this sort of environment, the only way for the DHCP server to match any of the secondary subnets was if it saw a giaddr from one of these secondary subnets. If the router only ever inserted the P address in the giaddr, none of the other scopes would ever be matched. This cycling approach causes the giaddr to change and rotate through all the subnets on the interface. As mentioned before, cisco now has a command which allows one to specify the DHCP relay behavior (ie always use the primary address or cycle through all subnets on the interface). This command is called "ip dhcp smart-relay". http://www.cisco.com/univercd/cc/td/doc/product/software/ios122/122sup/122csum/csum1/122csip1/1sfdhcp.htm#xtocid1563023 So in answer to the original poster's question, this command could be used to solve his/her problem. Of course, one needs to be running the right IOS rev and this approach will take 10s of seconds or minutes for the device to come online (as the DHCP cycling happens). Configuring the interaces to be related on the DHCP server is really that way to go (IMO). Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=35626&t=35539 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
