Controllerless Networks

This may have been asked and answered already, but I haven't found it. Here's the situation - We have two IAP277's that we're using as a PTP bridge (which is a mesh with only two AP's and not officially a PTP bridge) for a backup link between two buildings. These are currently configured as CAP's and not IAP's. It is working, however, our issue is what to do when the wireless link is down and we need to manage/troubleshoot the remote AP (mesh point). The mesh portal AP is always available, but the point AP is going through the portal AP as a default gateway, so we have no way to manage the AP unless the wireless link is up. Have any of you tackled this scenario before? I don't care if the solution is CAP or IAP, as long as it works. Also, just to be thorough, this backup link will have a routing protocol going over it, so the vlan(s) that handles the setup/maintenance/troubleshooting for the wireless bridge itself doesn't necessarily need to be the same as the vlan we use to enable the routing protocol since we can send multiple vlans across the wireless link.

If the only path to the mesh point is over the mesh link and the mesh link is down, you will need presence at the mesh point side (specifically for CAP the console port) to troubleshoot. I'm unclear if you are always wanting the link active or only upon an uplink failure (since you mentioned it as a backup link). For AOS Mesh, the link would always need to be active, and you would have to enable some kind of blocking on your wired infrastructure on the remote side to prevent a loop. But I might be misunderstanding your use case.

We have redundant paths to the remote building today. The mesh link is a backup path, and our routing protocol (EIGRP) is what determines primary versus backup path. Under normal circumstances, both paths are up, and the routing protocol will use the wired connection. If the wireless is down, I am going to assume that the wired (primary) connection is up and running. So we have a network path that would allow us to potentially connect to the mesh point via IP. If there's a way to avoid it, we'd rather not be forced to send a tech out to the remote building anytime wireless were down, but the wired/primary connection were up, to do troubleshooting on the mesh point.

Unfortunately with AOS mesh, the only path to the AP for troubleshooting is via the wireless uplink on the point, or the console port on the AP itself. You can create a specific AP system profile for your mesh point, and then try to enable telnet on that AP system profile and try to telnet to the AP, but again, the IP is assigned to the wireless interface, not the wired interface, so I am not 100% sure if that will work, but I would guess that it's no.

If we were to convert the AP's to IAP for just these two AP's, would that make any difference? Same scenario, just IAP versus controller based.

Possibly, though you would have to make sure that the IAP mesh cluster is isolated and that the mesh point cannot pull an IP address off the wire. With IAP, if the IAP point is orphaned it could come up as it's own VC though, so it should be tested in your deployment to see if that works as you need.

The big limitation is our mesh is not L3 aware (as you well know) so it's tricky to get some of these solutions to 'behave', and this is one of the bigger issues.