Search Mailing List Archives
[mininet-discuss] TCP-OUT-OF-ORDER segments on wireshark
andreasimeoni84 at gmail.com
Wed Oct 12 02:38:04 PDT 2011
Thank you for answering,
My setting is the following:
> I have a flow that is installed on a path, with one simple forwarding
action in each switch (forward on the route).
> Sometimes happens that I wish to inspect traffic flowing by that path. To
this aim, I install in the first and last switches on the path, one more
action: explicitly send packets of this flow to controller. When a packet
enters one of these two switches, both actions are applied together, then
the flow is forwarded on the route, and a certain number of bytes is
forwarded to the controller.
Maybe the problem is in the way I install the second action; since I send an
Flow-mod message to switches with two associated actions (as defined
before). In such a way I overwrite the existing single action, with the
previously defined pair, instead of simply adding one more action.
Maybe it would be solved by simply adding the second action instead of
overwriting with a new pair. Is this possible with OpenFlow?
2011/10/12 Bob Lantz <rlantz at cs.stanford.edu>
> I'm not entirely sure I understand your setup, but I don't think it's a
> mininet issue.
> Ordering is not preserved if some packets go through the switch and some
> take a
> detour to the controller. This frequently happens when flow entries expire.
> On Oct 11, 2011, at 6:14 AM, andrea simeoni wrote:
> > Hi guys,
> > I'm running a mininet with an OF switch having a flow entry with a pair
> of actions:
> > > forward packets of the flow A on the route.
> > > forward packets of the flow A to the controller.
> > In my test I'm using iperf and wireshark on the controller interface to
> inspect OF traffic. In this setting I can observe all packets that are
> forwarded to the controller.
> > The problem is that on the monitored interface, I observe a big amount of
> TCP-OUT-OF-ORDER segments. In my setting the controller is running on the
> HOST machine, then is outside the mininet virtual machine.
> > Do you have suggestions? For you is a normal behavior?
> > best regards.
> > --
> > Andrea Simeoni
> > _______________________________________________
> > mininet-discuss mailing list
> > mininet-discuss at lists.stanford.edu
> > https://mailman.stanford.edu/mailman/listinfo/mininet-discuss
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the mininet-discuss