Bugzilla – Bug 2911
Binary exponential backoff unlimmited counts problem
Last modified: 2018-05-13 05:05:53 UTC
Created attachment 3093 [details] It is some screenshots when this error happens. I find this problem when I try to do my simulations with aodv routing protocol, I set the simulation time 300s, sometimes it will ends my simulation in seconds, sometimes it takes long time to trigger this error, I have debug it in gdb, the error is in the attachment, with others help, I locate the position that triggers this error, it is at 1104 rows in src/aodv-routing-protocol.cc, when I add an if statement to limit the value of rt.GetRreqCnt () - 1, the program can runn correctly. This problem exists in ns3.26/ns3.27/ns3.28/ns3dev at least(I use ns3 simulation from 3.26 version, I hadn't checked other older version), as you can see in the attachment, I have tested my program with ns3.26 and ns3.28, in fact, I also have tested it with ns3dev. it is strange that when I also used aodv to do simulations last year, it happens incidently, very few times, and now it almost happens every time I run my program, maybe due to the node's movement. I can't give a certain simulation scene or script to demonstrate it will must happen, because the backoff times is not sure, but the realization of Binary exponential backoff is really a logic problem, in Binary exponential backoff algorithm, it has a limmitation about backoff exponential number, when it overs the limmitation, the backoff exponential number is a constant, when it overs another limmitation, it will stop the retry behaviors. What I can make sure is that this will happen in a frequently dynamic changed network with a high probability. Hope that you can fix this problem
Hi, this may be a duplicate of a bug that exists in our tracker (2517). Can you please try this patch and let us know if it fixes? https://www.nsnam.org/bugzilla/attachment.cgi?id=2710&action=diff
(In reply to Tom Henderson from comment #1) > Hi, this may be a duplicate of a bug that exists in our tracker (2517). > > Can you please try this patch and let us know if it fixes? > > https://www.nsnam.org/bugzilla/attachment.cgi?id=2710&action=diff Ok, thank you, I think this has fixed the problem.