routing issue

jimcalano · ‎09-29-2007

I inherited a wireless network with multiple wireless AP's connected to two GW's in a mesh configuration. The GW's are wireless AP's as well and are connected to two separate 2950 switches. Each switch is connected to 2621 routers.

The AP's hand out IP's on the 10.0.x.x /23 network and funnel the packets across a 10.1.1.0 /24 network to the routers, which are on a 172.16.0.0 /16 network. All traffic is then passed through a 172.16.0.0 gateway to the Internet.

The switches that the wireless GW's connect to are on the 10.0.x.x network. I can access all switches and all routers, but not the wireless AP's or GW's. I've tried adding secondary IP addresses to include the 10.1.x.x network, but still am unable to connect.

Because the wireless network is meshed I cannot subnet the GW's and build a route to each gateway on the separate networks.

My questions are, first, is it necessary to configure the switches with the 10.0.x.x IP address? Second, how would I build a route to the 10.1.1.0 network in two directions off separate interfaces from the main router? I need to be able to access all devices from the central location.

Thanks in advance for your help.

Richard Burts · ‎09-29-2007

Jim

There are some things about the topology of the network that are not clear to me. Some additional information might help us to give better answers.

But based on what is here I have a couple of comments:

- you ask if it is necessary to configure switches in the 10.0.x.x network. But you have not told us what addresses they currently have.

- you state that you can access all switches and routers. If you can access all switches why would you be asking about what addresses they should have? It seems that the addresses that they already have are good enough.

- the 2950 is a layer 2 switch. A layer 2 switch has an IP address so that you can manage it. The address (and what subnet it is in) have nothing to do with the switch forwarding frames.

- you ask how to build a route to the 10.1.1.0 network. But you have not told us whether you are using static routes, running a dynamic routing protocol, or what. With static routes it is easy: you configure 2 static routes for 10.1.1.0 with 1 static route having a next hop out one interface and the other static route having a next hop out the other interface. If it is a dynamic routing protocol then how to do it depends on which protocol and what the topology looks like. More information needed before we go any where with this one.

You list a couple of questions, but from what you have posted it looks like you did not get to the most important one: why can you not access the wireless APs or GWs? My first thing would be to ask you to verify that they are configured with appropriate IP addresses, correct subnet masks, and correct default gateways. I believe that these are the most likely causes of that problem.

HTH

Rick

HTH

Rick

jimcalano · ‎09-29-2007

Rick,

Thank you for responding so quickly and sorry for the lack of information.

I don't know why I can't access the GW's, I don't have physical access to them yet to look at their default GW. The net work was functional up to a couple of days ago when it experienced an ISP outage.

Anyway, the switches are configured with an IP address of 10.0.42.5 and 10.0.44.5. No dynamic routing in the inside of the access layer. The wireless AP are giving out addresses in the 10.0.42.0 /23 and 10.0.44.0 /23 address ranges.

Switch 10.0.42.5 has GW's attached to it with IP's of 10.1.1.50 and 10.1.1.51. Switch 10.0.44.5 has GW 10.1.1.60 attached to it.

Switch 10.0.42.5 is connected to switch 10.1.1.9, which in turn is directly connected to the gateway at IP address 172.16.0.1. Switch 10.0.44.5 connects to the 10.0.44.10 interface of an external 2621 router. This router is connected to the main router through it's 192.168.100.2 interface to the main router's 192.168.100.1 interface. The 172.16.1.2 interface of the main router is connected to the 10.1.1.9 switch.

That's basically the complete layout of the network.

So what I'm reading from your information is that it's most likely that the GW's aren't communicating with the switch interface? And if they were then I should be able to access them anyway?

When I try to configure the static routes out one interface, ip route 10.1.1.0 255.255.255.0 172.16.1.2, and the other interface, ip route 10.1.1.0 255.255.255.0 192.168.100.1, I receive an error stating that the route is already defined out the first interface.

Sorry for the rookie questions.

Jim

jimcalano · ‎09-29-2007

Thanks Rick. I think I figured it out. I think the Belair AP's reconfigured their default gateways when the circuit went down.

Richard Burts · ‎09-30-2007

Jim

I am puzzled why the Belair APs would reconfigure their default gateway. But that certainly would explain the symptom.

I am curious about this in your response:

When I try to configure the static routes out one interface, ip route 10.1.1.0 255.255.255.0 172.16.1.2, and the other interface, ip route 10.1.1.0 255.255.255.0 192.168.100.1, I receive an error stating that the route is already defined out the first interface.

What kind of router is this? I have never seen a router that would complain about a second static route to a destination using a different path. Could you perhaps give us information about the addresses of the interfaces on this router?

HTH

Rick

HTH

Rick

jimcalano · ‎09-30-2007

It turns out I have a bad network map with incorrectly labeled interfaces on neighboring devices. Either that, or I simply couldn't read it correctly. Issuing the sh cdp nei det command helped to identify the correct IP addresses assigned to the correct ports. The error was "%Invalid next hop address (it's this router)" which indicates that I was inputting the incorrect next hop address. Replacing "172.16.1.2" with "FA0/0" took care of the problem. I mistook that ip address as the next hop address when in fact it was the address of the fa0/0 interface on the router I was trying to build the routes for. My notes were incorrect. Rookie mistake....(and lack of adequate sleep).

Thanks again Rick. Your comments helped me to think about this problem more clearly.

Jim

P.S. It was explained to me that the Belair routers will automatically reconfigure their gateways if there is too much traffic from one AP to another AP or if the link between two APs drop. In the case of the link dropping, the AP will locate another gateway device to use and will automatically reconfigure to that device. I'm not too familiar with those devices so I can't say for sure that's what happened. All I know at this point is that there is no path from the switches to the Belair GW's, which are nothing more than AP's.

Richard Burts · ‎09-30-2007

Jim

Thanks for the additional information. This error message (%Invalid next hop address (it's this router)) makes much more sense than a message that the static route exists on another interface.

I am concerned about the solution with a static route which points at the destination via interface FA0/0. This solution depends on the next hop router enabling proxy arp. If the next hop router does not enable proxy arp then the static route will fail. And if the next hop does enable proxy arp then the router will be doing more work than it needs to. If a static route points to an Ethernet interface then the router must ARP for each destination address, and must maintain an ARP table which contains each of the deatination addresses. It is much better to do static routes for LAN interfaces which specify next hop addresses. Based on the output of show cdp neighbor detail you should know the address of the neighbor. It would be much better to use it in the static route.

HTH

Rick

HTH

Rick

jimcalano · ‎09-30-2007

It's a very small network anyway, but I like the idea of optimizing the tables. I'm thinking of rebuilding the network so everything is on one network. I think that's more efficient.

There are a total of about 30 AP's connecting to 4 AP GW's. Two GW's per switch. One switch connected to one router which is connected to another router, then on to a switch. The other side has a switch connected to a switch, which then connects into the Internet gateway. I don't know why the previous network engineer set it up that way. I don't know why this was split into a network with three layer 3 paths to the same end points. This just doesn't make sense to me. It seems like a lot of administrative overhead for such a simple network.

The main server that's acting as both the accounting server and the gateway to the Internet is on the 172.16.0.0 /16 network. It's connected to a switch that is passing traffic for the 172.168.0.0 /16, 10.0.40.0 /23, and 10.1.1.0 /24 networks. I really do think that this network is more complicated then it needs to be.