Hi, Thanks. It just works. I will take look at that function later. Is there any special reason which makes you set it to be false by default?
On Sun, Oct 26, 2014 at 2:54 PM, Murphy McCauley <murphy.mccau...@gmail.com> wrote: > Try modifying l2_multi so that the call to match.from_packet() passes the > keyword parameter spec_frags=True. > > -- Murphy > > On Oct 26, 2014, at 12:06 PM, tim huang <pds...@gmail.com> wrote: > > Hi, > > After I tested, I found that the packet needs to be fragmented will > cause _handle_PacketIn in l2_multi continually being called. And the packet > of the parameter 'event' will always be the fragmented packets. I did try > to track down into the event and handler system, but can't find the reason. > Is there any possible reason or clue for this? > > I just ping once from the host > > --- 10.0.0.3 ping statistics --- > 1 packets transmitted, 1 received, 0% packet loss, time 0ms > rtt min/avg/max/mdev = 0.068/0.068/0.068/0.000 ms > mininet> h1 ping h3 -c 1 -s 2000 > PING 10.0.0.3 (10.0.0.3) 2000(2028) bytes of data. > > --- 10.0.0.3 ping statistics --- > 1 packets transmitted, 0 received, 100% packet loss, time 0ms. > > > ------- --------- --------- > the code I add in handle_packet_In under l2_multi > > global handler_counter > handler_counter += 1 > print 'l2_multi_packetIn ' + str(handler_counter) > print packet > ---------- ------------- ------------- > The output from controller console > > > l2_multi_packetIn 8118 > [86:7d:e2:27:00:5c>da:22:ba:a5:76:a5 IP] > l2_multi_packetIn 8119 > [86:7d:e2:27:00:5c>da:22:ba:a5:76:a5 IP] > l2_multi_packetIn 8120 > [86:7d:e2:27:00:5c>da:22:ba:a5:76:a5 IP] > l2_multi_packetIn 8121 > [86:7d:e2:27:00:5c>da:22:ba:a5:76:a5 IP] > l2_multi_packetIn 8122 > [86:7d:e2:27:00:5c>da:22:ba:a5:76:a5 IP] > l2_multi_packetIn 8123 > > It's still increasing, even I stop the ping for around 3 minutes. > > > > On Fri, Oct 24, 2014 at 6:04 AM, Murphy McCauley < > murphy.mccau...@gmail.com> wrote: > >> Dump the control traffic using openflow.debug and then take a look at it, >> e.g., with the OpenFlow dissector for Wireshark. That may shed some light. >> >> You also might try enabling fragment reassembly by sending an >> OFPT_SET_CONFIG. >> >> -- Murphy >> >> On Oct 23, 2014, at 9:41 PM, tim huang <pds...@gmail.com> wrote: >> >> Sorry, please ignore last email. I made some mistakes for my statement. >> >> Hi, >> >> I find the situation is more interesting. At beginning, I thought it was >> caused by the fragmentation, then I can't ping with any fragmentation. >> However, when I tried several times.Actually, I can ping successfully with >> large size packet, if that's the first flow between 2 switches.After the >> flows timeouts on the switch, I can't ping any more through controller. >> >> Here is the output I get with same topology I described in my 2nd email. >> >> The 1st ping is successful. After the flows timeout on all the switches, >> I can't ping any more with large size packets, but I can still issue ping >> with small size packets. >> >> The outputs are the attempts with different sizes. >> >> And there are still some situations that I can't ping with large packets >> at the beginning when I use mininet. I'm thinking maybe there are some >> packets,like ARP or something else, had been forwarded between these 2 >> switches before I ping. Usually, the first ping can be successful on my >> esxi platform where I deploy the openvswitches and linux machines. >> >> ------------------------------------------------- >> mininet> h1 ping h3 -s 5000 >> PING 10.0.0.3 (10.0.0.3) 5000(5028) bytes of data. >> 5008 bytes from 10.0.0.3: icmp_req=1 ttl=64 time=94.5 ms >> 5008 bytes from 10.0.0.3: icmp_req=2 ttl=64 time=0.161 ms >> 5008 bytes from 10.0.0.3: icmp_req=3 ttl=64 time=0.092 ms >> 5008 bytes from 10.0.0.3: icmp_req=4 ttl=64 time=0.155 ms >> 5008 bytes from 10.0.0.3: icmp_req=5 ttl=64 time=0.130 ms >> ^C >> --- 10.0.0.3 ping statistics --- >> 5 packets transmitted, 5 received, 0% packet loss, time 4001ms >> rtt min/avg/max/mdev = 0.092/19.012/94.526/37.757 ms >> mininet> h1 ping h3 -s 5000 >> ^CPING 10.0.0.3 (10.0.0.3) 5000(5028) bytes of data. >> >> --- 10.0.0.3 ping statistics --- >> 212 packets transmitted, 0 received, 100% packet loss, time 211623ms >> >> mininet> h1 ping h3 >> PING 10.0.0.3 (10.0.0.3) 56(84) bytes of data. >> 64 bytes from 10.0.0.3: icmp_req=2 ttl=64 time=0.330 ms >> 64 bytes from 10.0.0.3: icmp_req=1 ttl=64 time=1166 ms >> 64 bytes from 10.0.0.3: icmp_req=3 ttl=64 time=0.093 ms >> 64 bytes from 10.0.0.3: icmp_req=4 ttl=64 time=0.095 ms >> 64 bytes from 10.0.0.3: icmp_req=5 ttl=64 time=0.081 ms >> 64 bytes from 10.0.0.3: icmp_req=6 ttl=64 time=0.083 ms >> ^C >> --- 10.0.0.3 ping statistics --- >> 6 packets transmitted, 6 received, 0% packet loss, time 5002ms >> rtt min/avg/max/mdev = 0.081/194.523/1166.459/434.663 ms, pipe 2 >> mininet> h1 ping h3 >> PING 10.0.0.3 (10.0.0.3) 56(84) bytes of data. >> 64 bytes from 10.0.0.3: icmp_req=1 ttl=64 time=76.2 ms >> 64 bytes from 10.0.0.3: icmp_req=2 ttl=64 time=0.083 ms >> 64 bytes from 10.0.0.3: icmp_req=3 ttl=64 time=0.095 ms >> 64 bytes from 10.0.0.3: icmp_req=4 ttl=64 time=0.061 ms >> 64 bytes from 10.0.0.3: icmp_req=5 ttl=64 time=0.079 ms >> ^C >> --- 10.0.0.3 ping statistics --- >> 5 packets transmitted, 5 received, 0% packet loss, time 4001ms >> rtt min/avg/max/mdev = 0.061/15.313/76.250/30.468 ms >> >> mininet> h1 ping h3 -s 5000 >> ^CPING 10.0.0.3 (10.0.0.3) 5000(5028) bytes of data. >> >> --- 10.0.0.3 ping statistics --- >> 42 packets transmitted, 0 received, 100% packet loss, time 41045ms >> -------------------------------------------------- >> >> I really want to fix this problem for pox to do some contribution, >> because it really helps me a lot. I do love the concept of pox than >> other similar controller,like Ryu. Can anybody give me some clue about this >> problem? >> >> On Fri, Oct 24, 2014 at 12:25 AM, tim huang <pds...@gmail.com> wrote: >> >>> Hi, >>> >>> I find the situation is more interesting than I thought. At beginning, >>> it's just caused by the fragmentation,I can't ping with any fragmentation. >>> however, when I tried several times. I found something new. I can ping >>> successfully with large size packet, it that's the first flow between 2 >>> switches,after the flows timeouts on the switch, I can't ping any more. >>> >>> Here is the output I get with same topology I described above. >>> >>> The 1st ping is successful. After the flows timeout on all the switches, >>> I can't ping any more with large size packets, but I can still issue ping >>> with small size packets. >>> Here are the attempts with different sizes. >>> >>> And there are still some situations that I can't ping with large packets >>> at the beginning. I'm thinking that maybe there are some packets had been >>> forwarded between these 2 switches before I ping. >>> >>> I want to fix this problem for pox to contribute some code for this >>> project, because it really helps me a lot. I do love the concept of pox >>> than other similar controller,like Ryu. Can anybody give me some clue about >>> this problem? >>> ------------------------------------------------- >>> mininet> h1 ping h3 -s 5000 >>> PING 10.0.0.3 (10.0.0.3) 5000(5028) bytes of data. >>> 5008 bytes from 10.0.0.3: icmp_req=1 ttl=64 time=94.5 ms >>> 5008 bytes from 10.0.0.3: icmp_req=2 ttl=64 time=0.161 ms >>> 5008 bytes from 10.0.0.3: icmp_req=3 ttl=64 time=0.092 ms >>> 5008 bytes from 10.0.0.3: icmp_req=4 ttl=64 time=0.155 ms >>> 5008 bytes from 10.0.0.3: icmp_req=5 ttl=64 time=0.130 ms >>> ^C >>> --- 10.0.0.3 ping statistics --- >>> 5 packets transmitted, 5 received, 0% packet loss, time 4001ms >>> rtt min/avg/max/mdev = 0.092/19.012/94.526/37.757 ms >>> mininet> h1 ping h3 -s 5000 >>> ^CPING 10.0.0.3 (10.0.0.3) 5000(5028) bytes of data. >>> >>> --- 10.0.0.3 ping statistics --- >>> 212 packets transmitted, 0 received, 100% packet loss, time 211623ms >>> >>> mininet> h1 ping h3 >>> PING 10.0.0.3 (10.0.0.3) 56(84) bytes of data. >>> 64 bytes from 10.0.0.3: icmp_req=2 ttl=64 time=0.330 ms >>> 64 bytes from 10.0.0.3: icmp_req=1 ttl=64 time=1166 ms >>> 64 bytes from 10.0.0.3: icmp_req=3 ttl=64 time=0.093 ms >>> 64 bytes from 10.0.0.3: icmp_req=4 ttl=64 time=0.095 ms >>> 64 bytes from 10.0.0.3: icmp_req=5 ttl=64 time=0.081 ms >>> 64 bytes from 10.0.0.3: icmp_req=6 ttl=64 time=0.083 ms >>> ^C >>> --- 10.0.0.3 ping statistics --- >>> 6 packets transmitted, 6 received, 0% packet loss, time 5002ms >>> rtt min/avg/max/mdev = 0.081/194.523/1166.459/434.663 ms, pipe 2 >>> mininet> h1 ping h3 >>> PING 10.0.0.3 (10.0.0.3) 56(84) bytes of data. >>> 64 bytes from 10.0.0.3: icmp_req=1 ttl=64 time=76.2 ms >>> 64 bytes from 10.0.0.3: icmp_req=2 ttl=64 time=0.083 ms >>> 64 bytes from 10.0.0.3: icmp_req=3 ttl=64 time=0.095 ms >>> 64 bytes from 10.0.0.3: icmp_req=4 ttl=64 time=0.061 ms >>> 64 bytes from 10.0.0.3: icmp_req=5 ttl=64 time=0.079 ms >>> ^C >>> --- 10.0.0.3 ping statistics --- >>> 5 packets transmitted, 5 received, 0% packet loss, time 4001ms >>> rtt min/avg/max/mdev = 0.061/15.313/76.250/30.468 ms >>> >>> mininet> h1 ping h3 -s 5000 >>> ^CPING 10.0.0.3 (10.0.0.3) 5000(5028) bytes of data. >>> >>> --- 10.0.0.3 ping statistics --- >>> 42 packets transmitted, 0 received, 100% packet loss, time 41045ms >>> -------------------------------------------------- >>> >>> >>> On Tue, Oct 21, 2014 at 7:08 AM, Lucas Brasilino <lr...@cin.ufpe.br> >>> wrote: >>> >>>> > Yup, dart. It's actually more than due to get rolled over to eel, >>>> but it's >>>> > waiting on me to have some time to dedicate to POX, which hasn't >>>> happened >>>> > for a while. :) >>>> >>>> eel ? I was about to suggest 'eager' name :-D >>>> >>>> >>>> -- >>>> Att >>>> Lucas Brasilino >>>> MSc Student @ Federal University of Pernambuco (UFPE) / Brazil >>>> twitter: @lucas_brasilino >>>> >>> >>> >>> >>> -- >>> Thanks >>> Tim >>> >> >> >> >> -- >> Thanks >> Tim >> >> >> > > > -- > Thanks > Tim > > > -- Thanks Tim
