I have installed GNS3 v0.8.6 on my laptop, Dell Latitude 6530, i7, 16GB RAM. I also have VMWare Workstation 10 installed, if that matters. For some reason, although I get green links on the built-in Ethernet Switch, I cannot get any traffic through it. I can get router to router, cloud(loopback) to router, etc. but nothing through a switch. When I swap in a hub instead, things work fine. Thanks for any help someone may provide.

Mark

I have the same issue with GNS3 0.8.4 stable for Debian 7 x64 Wheezy with Kernel: 3.2.0-4-amd64

Since it establish an UDP port with Dynamips and VirtualBox for socket data exchange, conntrack flush won't help this at all of course :-(. I haven't tried to send a SIGHUP to dynamips, should this be the case?

It's quite frustrating, also in my case the router is connected to a cloud that establish a tap0 unix. I am running GNS3 in root privileges so this won't pose a problem.

Can't figure out why this isn't working. The only conclusion I can arrive is that data is not actually send to dynamips from VirtualBox at all, host in VBox reports "Host unreachable" for echo request but using netstat the UDP socket does not present any byte data traffic increase, is at 0 all the time. My guess is that is something before the is sent which resonates to the fact that there is the impression of not having traffic flowing.

Later edit: actually there is data on Recv-Q ("The count of bytes not copied by the user program connected to this
socket.")

I've also tried to reset the switches from Configure menu. Also tried from command line, also tried with debug mode. Nothing of interest unfortunately; nothing worked. Any suggestions?

Dynamips version is: version 0.2.8-community-amd64/Linux stable
VirtualBox: 4.1.18 for Debian (revision: 78361)

Addition 1
So ok, after debugging more, I sent a SIGUSR1 to dynamips since it didn't reload on SIGHUP successfully. This made it simulate a crash and the switches clearly report that server node is offline: I think is something in dynamips since the data on Recv-Q above also points to this. What does it make dynamips to not answer? I hope to come back with an answer.

Thanks for your efforts!

As a side note, separated from the above but I finally found out why a Router won't start after you save its config on a new start of GNS3. Dynamips binds to 0.0.0.0 for 7120 port even if in Configuration I have it set to 127.0.0.1 (running GNS3 in privileged mode). I think we should patch to add 0.0.0.0 as a global address which is acceptable for 127.0.0.1 to workaround this (may not be the best solution).

Traceback is:

dynamips state is:Note the 0.0.0.0 bind. Seems it wants to send to localhost but gets 0.0.0.0 returned so upon send() call it will return an error. I need to find the call function which calls this one to debug the data that it fetches or reads on. Probably is better to exlucde 0.0.0.0

Problem is after many tries you will have a CLOSE_WAIT socket (see the "1" above for CLOSE_WAIT on Recv-Q --- this is similar --very similar-- to what I have presented above regarding switches... this points to dynamips communication issues) and thus you won't be able to start at all the router instance






By the way, as a suggestion you could implement Sphinx ReST type of documentation in your Python code which you can automate and dynamically build HTML documentation for all of us instead of letting us to reverse engineer your code. It would be much more useful. Hope it helps :-).

The new GNS3 will have a different approach regarding this.



The new code does, please have a look we are on github :)

Thanks for the suggestions.

_________________
Jeremy, GNS3 Programmer & Benevolent Dictator for Life.