Rodi - anonymous P2P

Indeed in the current version filenames can not contain whitespace. This is a limitation i will fix.

You can download files you publish - localy. this is what i do for testing. Or you can start two Rodi Core using two different ports, for example 31211 and 31210 (command line flag LocalPort followed by port number will do the trick)
for example, ...\java -jar rodi.jar LocalPort 31210
make sure that in the log Rodi reports that it binds to the right port. Publish files on one Rodi Core and try to download using the other.
You can use different initialization scripts (command line flag ConfigurationScript followed by script name)

You can run up to 10 clients (i did not try more than 10) Every client needs about 25M of RAM. CPU consumption is zero if you have enough RAM.

I am not aware of anybody running Rodi, besides you.

Chat room in the design phase exactly for the security reasons you mention. I have some ideas, but it is going to be very very different from IRC or chat in eMule. Imagine that you send UDP packet somewhere and do not expect any answer and not even ack that your message received on the remote side. suddenly you see reply - another UDP packet from some other IP address. packets can be signed (packets from friends/trustees). Signature is the only way to find out who sent this packet. No support for this so far and not a single word in the online documentation.

BT exists without chat room and without search. It takes times - usually 2-3 years (if ever). i am not going to give up the project easily. some projects take tens of years, may be Rodi is one of them.

Meanwhile forums/mail lists is the way to exchange files.

P.S. this page is going to break 1,000 views. there are at least one or two people besides you who reads my musings. funny.

P.S.1 download support in GUI will be ready tomorrow. it's running, but i need time for some polish http://larytet.sourceforge.net/image...leDownload.PNG

Larytet, if we are the only 2 people running Rodi, is there any way we can download some test files from each other? I posted the test files that I am sharing - maybe you can do the same? Running 2 instances of the program and transferring files within my own computer is just not the same as actually downloading from someone else. I would like to download or upload something, even just once. I have yet to see if Rodi actually works.

If chat is hosted by a trusted 3rd party - sort of like an irc server - then it is not necessary for anyone to know the IP of anyone else, other than the server's IP. When SUMI was beta-testing last summer, I downloaded a few files from the Sumi developer through an irc channel (In Sumi, the irc server acts as the ACK proxy) and this IP-spoofing test worked quite well. I got to see IP-spoofing in action - and running a simulation on my own computer is simply no substitute for that kind of real-world test.

One other question: when I publish a file and a file hash gets listed, how can I tell what the block size is of that file hash? And what command can I submit that will set a different block size of my shared files?

And yes, a lot of people read the posts that appear in the forum, even if they do not post themselves.

i am behind corporate firewall and i need help from outside to break out. i am not sure i can access your PC. send me your IP or post your IP range i will try

word 3rd party points to the problem. i know how SUMI works. i even tried to contact the author(s) using SF.net email but failed. i suggested to join efforts or in Rodi or in Sumi. i personally think that Rodi (after all Rodi is MY project) is thought out better, but i can very well be wrong. The idea behind Sumi is very smart. There is only one problem - what happens if IRC bot got compromised and goes down. There is another backup IRC bot, right ? and so on ad infnitum. add to this logs collected by IRC server - there are some logs, right ?
And Sumi does not answer problem of trsuted publishers and faked data. or you are going to restrict access to IRC servers - create somethign like DC hubs ? there is no docs/specs on the website to find out what author(s) think about it.


different block sizes is not supported yet. i do not have time table for this one but i promise it is going to be implemented. it is right thing to do and it will be done.

P.S. the avatar on the left was intended to bring some colors to the board. Block images from geocities.com (do you use Firefox ?) will remove this thing.

Nice avatar. Dali had quite a creative mind. It's funny, the last time I was at a bookstore, I was looking through a book of his artwork. As long as you can shrink down the image to a manageable size - as this forum does not automatically do that.

Of course there are quite a few applications designed to tunnel P2P traffic under corporate firewalls - such as HTTP-Tunnel, I wonder if something like that works where you are? There is always a way around it I'm sure.

I only have Sygate Personal firewall - no router. I created a rule to open ports 31211 and 31213 - hopefully I did this correctly.

Code:

Look Req             = 0

so I guess nothing is happening.

You were saying before that Rodi will be able to mimic other types of internet traffic, so when this feature is functional, getting out from behind a corporate firewall will be the next step - you probably already know what ports and protocols your firewall allows, designing this might not be too hard. I tend that you being stuck behind a firewall is actually an advantage in many ways. There is definitely a great need for a firewall-busting P2P.

the last one was

local port: 53
remote port:18228

Another interesting I just found is that when I installed CommView, it replaced the RAS drivers, so all the traffic is showing up as contacting tscomm.sys

tscomm.sys is a driver for dial up

i use Ethereal or Packetyzer, but i am not sure that they are going to work with PPP

good night

I was not aware, until now, that CommView overwrote that Windows driver.
I wonder if that might be why I have a buggy connection - traffic will stop after awhile (several hours) until I reboot pc.

I guess I would like to know in advance when an application overwrites Windows system files - they never tell you

This trick with DNS - ISPs are not going to like it. + in the long term firewalls can fight this back and it's relatively easy to filter faked DNS packets. So, i guess, some statement is required.

I am fully aware that ISPs are our best friends and in the future allies. I am going to move ahead with DNS, but let me make it absolutely clear - this is a temporary solution. In the beginning of 2006 i will replace the scheme.

I wonder how a ISP/network admin would know if you are on a shared line (cable or LAN) and impersonating your neighbor's computer?

Or several neighbors might be able to form a neighborhood organization and buy a block of IPs, and then randomly distribute/share them? Then there might be collective guilt, but not any individual liability, perhaps, since no one could then be singled out.

I also wonder if Rodi would work with a DNS-forwarding service such as No-IP.com? Since I do not have a static IP, and my IP can change rapidly while I leave my computer running. I guess maybe we all need to get together and try to set a time when we can all be online and available.

Is there any other software/service that would notify people of a change of IP? I am thinking of those banner images that used to be common with eDonkey users that would give the person's actual ED2K server - in real time - that the user was connected to. I can leave Rodi running, but my IP might change every hour or so.

As to this being only a temporary solution until 2006, I think that anonymity would be very valuable in countries such as China, where people can be jailed for only reporting news to the outside world. It would be my guess that those regions might be lagging in technology capability (why spammers can get away over there) so spoofing might work much longer in developing countries (that often have repressive governments). IP spoofing would be a much better way to distribute big files such as news videos, for instance, than Freenet, which is painfully slow for big files. Right now governments (this even includes the USA) are going to great expense to track down the originators of things such as newsreel videos that get posted on the web. So I think that anonymity for sending large files can be an effective tool to fight political repression.

the strength of IP scan that you do not need DNS and services like No-IP

usually your dialp up connection will get IP address from the same pool of IP addresses installed on the DHCP server of your ISP. in most cases size of the range is well under tens of thousands.

you use look or nat open (i improved NAT penetration support this morning) to find a peer. No matter what IP address of the peer is eventually you will find it by scanning all IP addresses. this is where table of hosts with IP ranges comes really handy. no DNS is required, trust me.

if IP ranges are large and there are many of them and number of peers is realtively small it takes time to get connected, but once connected you get list of existing peers and can improve your next look session and so on

your ability to search the network is limited only by size of your RAM and your upstream.

Packets Rodi sends are very small - under 256 bytes in most of cases. IP scan of the whole IP address space (4bill addresses) will take about month if you have 2Mbit uplink. This is a perfect task for Rodi crawler. If you are lucky and find Rodi crawler quick you will get any number of IPs you want up to size of the whole network.
Rodi crawler can also serve small group of people (hub) and even can require your public key to be registered somewhere.

Shared line

FCC requires connection logs from service providesr (ISPs) Congress can (and apparently will in the near future) require wiretappin abilities on the routers. If ISP fails to keep logs ISP will be liable.

neighborhood organization and buy a block of IPs,

Even easier - buy fiber connection or a couple of E1s or satelite, install WiFi amplifiers and one Linux box or Windows with NAT and you run your own ISP and any Linksys modem in the area of up to 2 miles can be connected. There is a company in the US which provides full set of SW/HW (you need some login page running on your NAT machine, etc.) + ISP services. the service is high speed connection at $150+/month
Beware that the moment you start to provide services you can be hold liable (i am not a lawyer) for the infrigements of your customers unless you have reasonably reliable log proving that this is not you.
You do not need Rodi for this.
You can establish fan club "Fans of Record Industry and Cinema" (FRICs) and enjoy mmmm..... let's call it "group watching and discussion".

P.S. UDP based protocols (Rodi) will perform significantly better than TCP on high speed connections wiht significant packet loss rates like WiFi, WiMAX, etc. It is going to be especially noticeable when the swarm is relatively small - 2-3 up to 10 peers. Rodi is dreaming application for university campuses.

i think your range is smaller than 2x64K
it is two ranges, but probably divided in subranges
see Rodi How to page (scroll three or four pages down) "How to find out IP addresses of your ISP "
indeed US ISPs have large consequent IP ranges because they were first in line.
But even for 128K of ports we are talking about
128K x 100 = 12M of traffic. it will take 2.5 hours to find you using dial up modem. 5 hours if i am out of luck and one packet is lost.
if there are other peers in this range with the same port number i will find them too (and they will find me). More peers means faster connection.
Peers exchange IP lists (not full lists but randomly selected subsets). You found one you learned 3-5 more.


...and YOUR Rodi client is Crawler. (not supported by existing code) i will let you to choose how freely you distribute lists of IPs and how much badnwidth you are ready to allocate for this.

Rodi client constantly calculates for every peer what you gave to the peer and what you received from the peer and formula includes both bandwith and CPU/RAM costs. It makes senses in Rodi to be cooperative. Pure leachers are going to be blocked fairly fast.