Responses inline. On Mar 29, 2012, at 9:11 PM, terryxing wrote: > I am installing the NOX controller on Ubuntu10.10, but it turns out like the > current NOX has three tracks, NOX(C++), POX(Python) and NOX classic (C++ and > Python). I installed the NOX (C++ and Python) before by following the > instruction on NOX wiki. However, that wiki seems not be available any more. > The current instruction (https://github.com/noxrepo/nox-classic/wiki) is > incomplete. So, I am thinking I may want to install the POX instead of the > NOX. Before start installing the POX, I have following questions:
First, a note: the github wiki should have all of the content of the previous wiki. There may be a couple places where the links are broken (I've fixed all the broken ones I've found), but all the content should be there. > 1. When I finished installing the NOX and its dependency and tried to run the > nox controller. I can not successfully run the nox. I got the NOX from the > git, I think it was the latest version. Now I just tried the NOX destiny > branch, it works. Was this a question? destiny is the latest version of NOX Classic. > 2. Does POX introduce more limitation or restriction into the original NOX ? > I mean, is infrastructure which is doable on original NOX still doable on POX > ? For example, I am caring about if multiple POX can cooperate with Openflow > and Flowvisor like NOX did ? POX should not impose any additional limitations. However, it is a newer project and therefore has not yet been run in as many situations as NOX, so there may be problems that haven't been discovered yet. Bug reports on this sort of thing will be addressed. For example, I only just found out this week that HP switches weren't compatible (because of a case where they stray from the OpenFlow spec), since nobody I knew of had tried to run it on an HP switch until recently. (I'll be pushing fixes for this issue soon.) To take your specific example, I do believe it currently runs on FlowVisor okay (if it doesn't, it will in the next few days). > Thanks very much for your kind reply. Hope that clarifies some things. -- Murphy
