Hello,
On http://www.aredn.org/content/device-device-linking-dtdlink, for linking three devices with a non-vlan switch, the graphic shows two of the nodes labeled in ref "DHCP: DISABLED" but at the bottom of the graphic it says "CAUTION: Do not disable DHCP on all nodes".
This is confusing especially when you look at the next graphic for 3 nodes linked with a vlan switch on the same vlan (2) and all devices show DHCP enabled.
I will be setting up two (maybe three) nodes for a temporary field deployment linked with a non-vlan switch so I need to know what is the proper configuration for DHCP? I can see the need to disable DHCP on all but one node for proper IP assignment on peripheral devices but what happens with IP addressing of the nodes themselves when two nodes are connected together via their lan ports and do I even need to be concerned?
Can someone please clarify?
--
Bob, W7REJ
So the second image the dotted line is untagged and is set to the PVID of the port which is different on each port of the managed switch. The second image actually has VLAN2 (dtdlink) along with VLAN 11 (connected to a node and camera as a lan network) and 13 (connected to a node and VoIP phone as a lan network)
In the case of the first image (unmanaged switch) the "LAN" ports do not communicate with each other however if the DHCP were enabled on all of them as you noted the proper IP assignment wouldn't be guaranteed (I could go pretty in depth on this one if needed but suffices to say in this case only having one master node control DHCP is the generally recommended method so that way you know the network layout will be constant.) With DHCP disabled on all but one you could still talk directly to each mesh node if you had the right LAN configuration on the devices plugged in (this is an advanced topic that I'm not going to go into) but without that it just passively sits there and doesn't do anything. All traffic instead goes over the tagged VLAN #2 which all the nodes are still listening to and responding to.
So referring to the first image, does network traffic flow between the three nodes on the lan ports (connected to the switch) as if they were linked through RF? The lan ports are tagged VLAN #2?
I need to read up on this more.
--
Bob, W7REJ
The only reason we need managed switches is to pull out these "trunked" ports into physicial ports that can be used by the user. That's why it is possible (if you don't need WAN) for these to be setup with an unmanaged switch because those are requested to pass this "trunked data" along with modification which all the other mesh nodes can understand and a regular computer will just discard the "tagged" data.
Perfect, that's what I needed to confirm. We're doing a field deployment and have no need for a WAN port. We'll have a couple locations that need more than one node so this will work.
Check out http://www.epicrides.com/events/24-hours-in-the-old-pueblo/event-guide/.
The course is grueling with a total distance of 16.1 miles. Each year there are injuries and even deaths. Imagine taking a serious tumble into a cholla. The course is not flat and has an overall difference in elevation of over 1200 feet. We are setting up a mesh network with a total of seven locations for nodes. We'll have node heights ranging from 15' to 35' and some located at places to take advantage of the location's elevation as it relates to the course. This will give us at least a 60% clearance in fresnel zone between neighbor nodes with some completely clear LOS.
Last year we did not have a viable network largely due to poor (or lack of) planning. This year we are spending lots of hours in the design and planning of the network and will start constructing it about 3 days prior to the event. If this goes off as we hope, it will be a huge accomplishment and lay groundwork for other future events and even emergencies. This is one of our practice EmComm events. We want to use the network for VoIP as that will give us concurrent communication capability that we just don't have with radio. We also plan on using it for video (useful for real time recording of accidents) and possibly hosting RFID applications.
Anyway, thanks for the information.
--
Bob, W7REJ
The caution "do not disable DHCP on all nodes" is valid. Yes you do want to make sure only one node has DHCP enabled (on a group of nodes connected via a unmanaged switch). The tricky part is once you disable DCHP on a node it becomes difficult to communicate to it again in order to re-enable it. However this can be done via another node that is RF connected to it or a node that is DtD connected to it.