You were exactly right Rob. The 192.x.x.x IPs were encapsulated inside the controller's IP addresses. But still I don't get response from the controller (Floodlight in this case). Controller is responding to the initial 'hello' messages but does not respond to the 2 packet_in messages that are sent to it (rest of packet_in's seem to have been blocked somewhere). I am using the pre-configured floodlight VM for this purpose and I know floodlight is running as I can do a successful 'pingall' in a minimal mininet topology.
Kindly help me resolve this problem. Thanks, Salman > Date: Thu, 31 May 2012 12:47:18 -0700 > Subject: Re: [openflow-discuss] cbench utility > From: rob.sherw...@bigswitch.com > To: salma...@live.com > CC: openflow-discuss@lists.stanford.edu > > Salman, > > What you're doing sounds completely normal. What controller are you using? > > Also, if you're using wireshark to dissect the packets, the IPs are a > bit confusing -- you're seeing the IPs of the data packets embedded in > the packet_in's, not the IPs of the outer transport layer packets. If > you explore deeper into the packet, you should be able to verify this. > > - Rob > . > > On Wed, May 30, 2012 at 1:24 PM, Salman Malik <salma...@live.com> wrote: > > Thanks for the quick reply Rob. > > In my configuration I am using two VMs connected in a host-only network mode > > and I can ping fine between the two VMs. And while using cbench I see the > > right src/dst IPs in the packets when the switches send 'Hello' messages to > > the controller but rest of the messages have 192.x.x.x type of addresses. > > Furthermore, even when using it with the local host I see some of the > > packets with these 192.x.x.x addresses. So I think its not a firewall issue. > > > > I am using the OpenFlow tutorial VM which has preinstalled cbench and I > > start it with "cbench --controller=<remote-IP>". May be I am doing something > > wrong here. > > > > Salman > > > > > >> Date: Wed, 30 May 2012 13:15:40 -0700 > >> Subject: Re: [openflow-discuss] cbench utility > >> From: rob.sherw...@bigswitch.com > >> To: salma...@live.com > >> CC: openflow-discuss@lists.stanford.edu > > > >> > >> On Wed, May 30, 2012 at 1:11 PM, Salman Malik <salma...@live.com> wrote: > >> > Hi All, > >> > > >> > I have been looking at the cbench utlility and found that it works > >> > almost > >> > fine when the controller lies on the localhost, but when controller lies > >> > on > >> > a remote host I see no replies from the controller. I can see the > >> > packet_in > >> > messages at the controller side using wireshark and it seems that that > >> > the > >> > source and destination IP addresses of the packets are somewhat hard > >> > coded > >> > to 192.168.0.40 and 192.168.1.40 respectively. Furthermore, I don't see > >> > any > >> > option/argument for changing these values. > >> > > >> > So perhaps my question is: Can I test a remote controller using cbench ? > >> > >> My high-level reply is that cbench definitely works with remote > >> controllers and people have used it to do that all of the time. My > >> guess is that you're having some sort of environmental problem, e.g., > >> a firewall blocking the connection. > >> > >> Hope this helps, > >> > >> - Rob > >> .
_______________________________________________ openflow-discuss mailing list openflow-discuss@lists.stanford.edu https://mailman.stanford.edu/mailman/listinfo/openflow-discuss