Restricting True-P2P traffic to a local network
 

I am trying to find a quick/easy/idiot-proof way to set up a true P2P network such that all traffic is restricted to a local subnet. This should preferrably be done from the client-side with a config file.

I am looking for a solution that is true P2P (that is, doesn't require a centralized host that everyone on the network has to connect to), and that is not much more difficult for users to setup than something like "Download this package, run the setup, select which files you want to share." This solution should preferrably be OS-independent. (or at least work on windows and mac). Has anyone encountered something like this before? Wanting to be able to share, and easily search for files on a LAN? (Not simply windows filesharing) Perhaps a gnutella client packaged with a config file?

LimeWire multicasts queries to the local network but it's neither quick, nor easy nor fool-proof to configure it to run without any connections at all.

If you have an almost static local host to connect to, you could use LimeWire with some modifications to the default configuration. The default configuration - if your users should be able to use LimeWire out-of-the-box - can be found in the source code and requires compiling & packaging the modified LimeWire version yourself.

Gnucleus LAN
 

Gnucleus LAN does exactly what you want without having to recompile or deal with configuration scripts. Just download a copy of Gnucleus LAN create something similar to a workgroup they call it a network. As long as your machines join the same network they will all be able to see one another.

I don't know if this specific client will meet your cross-platform needs as I think it primarily runs on windows. I have used it to build a private P2P community.

Hope this helps.

GnucleusLAN will only run on Windows.

Thanks... quick question about configuration of GnucleusLAN. I specify the workgroup at the beginning. Then, if I want it to be private, do I have to restrict the IP range in preferences?

Also, will this autodetect which nodes on the LAN are running the client, or should I just set it up on a few machines first and add those hosts manually?

Thanks in advance. If this works I'll be working on a MAC port for a few days... heh.

Also...
 

Also... will this keep people with another gnutella client out? I.E. someone inside the network connects to it with another gnutella client. Then someone outside the network connects to them. Will this keep the person outside the network out? Or will only people with this client be able to connect?

You could also use phex and tell it to cennect to a differently named network. (switch network).

Only Phex clients who know the network name will then be able to connect. What exactly it does in a LAN, I doN't know (there is an option to connect to the LAN, which is standardly enabled. I don't know, if only setting this and not telling it to connect will only connect to clients in the LAN).

It still uses GWebCaches.

Phex is written in Java. so usable on Mac and PC, I use it myself in OSX.

http://phex.kouk.de

You might want to ask the developer to create a special version for you in the phex forum: http://www.gnutellaforums.com/forumd...?s=&forumid=16

Gnucleus LAN Solution
 

Yes, you could in fact use both of the methods you eluded to protect your network from outside connections. However, I am assuming that your subnet is not directly connected to the public Internet. If this is true, you don't need to worry about it.

Even if you're subnet is connected to the public Internet people wishing to connect to your network would have to know the name exactly, they would also have to specify the correct port. Finally, they would also need the Gnucleus client to connect to your network.

Clients that are physically located on the same LAN i.e. Broadcast domain will automatically see the network specified by the first client on the P2P network via the Gnucleus LAN client.

Basically this is how it works. I have approximately 400 machines located in a community based network. 200 machines in one building and 200 machines in another Building.

In Building #1 I create a network called P2P-BLDG I do this by creating a network using the first machine containing the Gnucleus LAN client. Next, I distribute the client to the other 199 computer users. After they start their clients they would have two choices

#1 join an existing network (they will see the name of the network that I created--because the computers are all in the same broadcast domain)

OR

#2 Create a new network yes, you can create multiple networks and the users can select the networks they want to join.

note that all computers in building #1 are on the same subnet 192.168.0.0/24

The computers in Building #2 are on a different subnet.
192.168.1.0/24 Therefore when I distribute the Gnucleus client to those users, they have to specify that they want to connect to IP_ADDRESS and PORT# + NETWORKNAME of the main computer running the P2P-BLDG network.

Each person would have to do this. But there's a solution. To simplify connectivity for the users in Building #2

The first machine that starts in Building #2
creates a network called P2P-GW which will be available to all of the computer in that building/subnet and have the P2P-GW computer establish a connection to the main computer on the network hosting the P2P-BLDG network. Obviously, the two subnets need to be able to reach one another via a router.

BLDG#1 BLDG#2

P2P-BLDG <=router=> P2P-GW
200 users 200 users

hope this helps.

Ok, one more question...
 

Ok, one more question (i hope this is it)... say I've got a network started with GnucleusLAN, and I've got several hosts connected to that network. If the node that started the network (in the example above, the node that created P2P-BLDG) gets shut down, what happens to the network? Are the other nodes still connected to each other and the network? Can new nodes see the P2P-BLDG network? Thanks again!

Gnucleus LAN Solution
 

It's been a while but I think that if the main machine shuts down another machine will take its place. This I believe is automatic. I think that the new clients will continue to see the network P2P-BLDG as long as there is at least one node on the same LAN signed into that group/network.

If you have three machines you should be able to verify this.