I have a strong feeling that what is meant by this statement is that if your helper address is a subnet, rather than a host, that the destination address ( I believe the Siaddr field? correct me if I'm wrong - I can't get to any of the RFC sites I know for some reason ) would be the directed broadcast address. E.g. if your ip helper is 192.168.1.0 then a DHCP request would be sent to 192.168.1.255 In this case, the first DHCP server on that subnet to respond would be the source of the scope from which IP's are assigned.
Chuck ""Elijah Savage"" wrote in message [EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > I have just spent my afternoon looking through my docs heheheheh. But I > remember I actually saw this on a exam when I was going through my ccnp. > It is in the routing book on page 87 very first paragraph. Exact quote, > "It is important to note that every broadcast gets sent to all helper > addresses, regardless of wheter the server will actually be capable of > helping for a certain port". > So this is wrong it sure seems to be? Any others out there see this type > of behavior? > > > I noticed this phenomenon in a production network where I once worked. > > We had two DHCP servers, but all machines in the network appeared to be > > hitting only the first one - all IP addresses issued company wide were > > from scopes defined on the first machine listed in the IP helper > > section. > > > > My conclusion - in the Cisco world, it is not sequential. The DHCP > > request is forwarded only to the first IP listed in the configuration. > > If you want failover, you have to use the subnet, not the specific IP. > > > > Makes sense rationally speaking. The router only forwards packets, it > > does not verify if the destination host is alive or not. > > > > HTH > > > > > > ""Elijah Savage"" wrote in message > > [EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > >> Seeing the other ip helper question made me think of what I was > >> working on in my lab. On Friday morning I get into work and there was > >> a severity 1 ticket where about 800 employees could not log in. We > >> discovered that one of the dhcp servers was down but we have 2 so in > >> theory all should have been fine, on all of our routers we have both > >> dhcp servers for ip helper. From reading some place in my long journey > >> I am sure I read that ip helper would take a broadcast and change it > >> to unicast and send traffic to all ip helper addresses regardless if > >> it is down or not. But in this case that did not happen. To get > >> everything back up I actually had to change the order that I had the > >> ip helper addresses in. The server that was down I put it last and put > >> the server that was up first and then everything started to work. So > >> it seems as if some primary secondary thing is going on here. We are > >> running ospf on our backbone with a variety of equipment > >> configurations 6500's 5500's 3600's 2600's. All routers has a > >> different version of IOS we have not had a chance to bring them all up > >> to the same code what is similiar is they all have at least 12.0 on > >> them. I want to try and figure this out myself so I started playing > >> with this in the lab with 2600's running 12.1(5) IOS and I came across > >> the same exact thing. Did this change with IOS 12 or something has > >> anyone else experienced this? > > Nondisclosure violations to [EMAIL PROTECTED] Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=45059&t=45045 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
