Can you help me with a suitable example to verify the working of set_field action?
On Wed, Apr 8, 2015 at 11:02 AM, Ruchika Luthra < h2013...@pilani.bits-pilani.ac.in> wrote: > Thank you Nicholas. > > You are quite right. Actually my current point was just to experiment and > see if the change done by set_field action can be reflected in the packet. > You are right to point out my error about MPLS. I gave the example without > thinking for which i apologize. > if considering any other field, my point is just to see a packet with the > changes reflected on it. > > On Wed, Apr 8, 2015 at 10:55 AM, Nicholas Bastin <nick.bas...@gmail.com> > wrote: > >> On Tue, Apr 7, 2015 at 10:35 PM, Ruchika Luthra < >> h2013...@pilani.bits-pilani.ac.in> wrote: >> >>> I am trying to modify a header field by using set_action instruction. I >>> have been able to successfully add a flow table entry corresponding to >>> this. But how can I actually verify if the change has taken place >>> successfully in the packet. >>> >>> I have tried doing this: >>> >>> sudo dpctl unix:/tmp/s1 flow-mod cmd=add,table=0 >>> in_port=1,eth_dst=00:00:00:00:00:02 apply:output=2,set_field=eth_type=0x8848 >>> >>> The flow table entry has been successfully inserted. But how to make >>> sure that the change has been successfully applied? >>> >> >> Primarily the way that you would see this change is via a tap on the >> output port or running tcpdump on the end host in question. >> >> However, what you are asking the datapath to do is undefined, and, by the >> rules of the MPLS spec, illegal. You are asking the datapath to match any >> packet incoming on port 1 with a specific destination MAC, and to change >> the ethertype of this packet. How do you imagine this will work? >> Obviously the incoming packet has a specific format based on the original >> ethertype - do you imagine that the datapath will do whatever magic mapping >> is necessary to convert the packet? It most certainly will not do this - >> it will, at best (assuming it does not just drop the packet), simply >> overwrite the 16 bits of the ethertype field with your new value, >> essentially corrupting your packet. Also, 0x8848 is reserved for MPLS >> packets that are carried in multicast ethernet frames, but your match >> specifies a destination MAC address that is *not* multicast and thus would >> create a packet that will get dropped by any legacy MPLS hardware in your >> path (including host NICs, in a lot of cases). >> >> Perhaps you need to step back and explain what you are trying to >> accomplish at a high level, before asking how very specific things work. >> >> -- >> Nick >> > > > > -- > Ruchika Luthra > M.E (CS) 4th Sem > BITS Pilani > > -- Ruchika Luthra M.E (CS) 4th Sem BITS Pilani
_______________________________________________ openflow-discuss mailing list openflow-discuss@lists.stanford.edu https://mailman.stanford.edu/mailman/listinfo/openflow-discuss
