February 2011 - Ibr-dtn - lists.ibr.cs.tu-bs.de

API port - global listen
by Harsha Chenji 20 Feb '11

20 Feb '11

Hello all, Is there any way to make the dtnd API listen on 0.0.0.0:4550 instead of 127.0.0.1? If yes, is there a UCI option in openwrt? Thanks

1 0

Re: [ibr-dtn] Need your help about ibr_dtn. dtnping failed
by Johannes Morgenroth 18 Feb '11

18 Feb '11

Hello Ke Meng. One important thing with IBR-DTN is, that the node must have discovered each other. The discovery mechanism uses broadcast/multicast UDP beacons to detect another node. If this is not working, a bundle never gets forwarded. To check if the nodes has discovered each other check the log output. There have to be NodeEvent saying node xyz is available. You can force this by define a static node like in the example configuration. But be aware that this defines a static node and will not work disruption tolerant. To solve your issue, you should check if broadcast/multicast is working in your setup. You could use tcpdump to listen on UDP packets addressed to port 4551 and check if they arrive at the other host. That is the default discovery port in IBR-DTN. Lets say something to the "localhost dtnping"-thing. An alias like localhost do not exist in IBR-DTN. To use dtnping locally, you have to call dtnping with the correct local hostname AND the application to send the bundles to. In your case this would be "dtnping dtn://OpenWrt/echo". Best regards, Johannes Morgenroth Am 17.02.2011 20:22, schrieb Meng, Ke: > Hi Johannes, > > I'm trying to use ibr_dtn under openwrt, however failed to get through by dtnping. There are few guide about how to use ibr_dtn in the internet. Could you help me with that. > > I compiled the openwrt with ibr_dtn and used it in the virtualbox. Two virtual machines were created with bridge to connect each other. The ping command is successful. However, the dtnping always failed. Here is one of the ibr_dtn configuraton: > > local_uri = dtn://openwrt1.dtn > routing = default > blob_path = /tmp/ibrdtn/blobs > storage_path = /tmp/ibrdtn/bundles > statistic_type = stdout > statistic_interval = 2 > statistic_file = /tmp/ibrdtn.stats > net_lan0_type = tcp > net_lan0_interface = eth0 > net_lan0_port = 4556 > net_interfaces = lan0 > > > By the way, the dtnping also failed to the localhost(127.0.0.1). Do you think there are other things I need to consider? > > When I use debug mode, one side shows: > Thu Feb 17 18:43:14 2011 NOTICE: BundleReceivedEvent: Bundle received [351283394.7] dtn://OpenWrt/echo-client -> dtn://openwrt1.dtn > Thu Feb 17 18:43:14 2011 NOTICE: QueueBundleEvent: New bundle queued [351283394.7] dtn://OpenWrt/echo-client > Thu Feb 17 18:43:14 2011 NOTICE: BundleEvent: bundle [351283394.7] dtn://OpenWrt/echo-client received > Thu Feb 17 18:43:45 2011 NOTICE: BundleEvent: bundle [351283394.7] dtn://OpenWrt/echo-client deleted > Thu Feb 17 18:43:45 2011 NOTICE: BundleExpiredEvent: Bundle has been expired [351283394.7] dtn://OpenWrt/echo-client > > The other side shows nothing. > > > Thanks a lot. > > Ke Meng -- Johannes Morgenroth Institut fuer Betriebssysteme und Rechnerverbund Tel.: +49-531-391-3249 Muehlenpfordtstrasse 23 Fax.: +49-531-391-5936 TU Braunschweig D-38106 Braunschweig

1 0

Re: [ibr-dtn] Fwd: Neighbor discovery in IBR-DTN
by Johannes Morgenroth 10 Feb '11

10 Feb '11

Hi. I think the periodically sent data you had observed are tcp keepalive beacons. Send by the tcp protocol stack itself. Those beacons contains exactly one byte. The summary vector exchange is done by sending real bundles (over tcp, udp or whatever convergence layer you use) with an attached special extension block to the discovered peers once them has been discovered and then only if they change. The extension block itself has an identifier and a string with the phrase "epidemic routing" in it. Additionally, it contains a bloom-filter as summary vector, like proposed in the paper introducing epidemic routing. In future releases this behavior will change and summary vector will be requested by other peers. Then the bundle contains the data as payload, no longer as extension block. Best regards Johannes > Von:< > > > Datum: 9. Februar 2011 21:05:55 MEZ > > An: "Sebastian Schildt"<schildt(a)ibr.cs.tu-bs.de> > > Betreff: Re: [ibr-dtn] Neighbor discovery in IBR-DTN > > > > Hi, > > > > I was able to set up the discovery mechanism after making the > > changes you suggested in the configuration file. > > > > I wanted to know more about summary vectors.How are they > > exchanged?I turned on Wireshark and found that after UDP > > beacons are sent,there is some TCP data stream being sent > > periodically.Is that the summary vector exchange?When I try to > > monitor these streams closely,I get some information saying > > that 1 byte data was sent(character @).What does this mean? > > > > Thanks > > Neha Paranjape > > Graduate Student(Telecommunications) > > University of Maryland,College Park -- Johannes Morgenroth Institut fuer Betriebssysteme und Rechnerverbund Tel.: +49-531-391-3249 Muehlenpfordtstrasse 23 Fax.: +49-531-391-5936 TU Braunschweig D-38106 Braunschweig

1 0

Neighbor discovery in IBR-DTN
by parneha＠umd.edu 02 Feb '11

02 Feb '11

Hi, I added tcp convergence layer and routing = epidemic to my conf file.I did not add any static nodes.This was done on two laptops. Now,when I started both the daemons,I was expecting the NodeEvent log but I did not get any.This means that the two laptops have not discovered each other,correct? The next time I run both the daemons I add static connections between them in their conf files.Now I see the NodeEvent log and the connection event.What could be the reason for this?How did it give these logs when I added the static nodes? Does this mean that nodes with static connections to them are discovered as up or down and the others are not?I am kind of confused here. Neha Paranjape Graduate Student(Telecommunications) University of Maryland,College Park

2 1