Unfortunately oflops needs some love and neither myself nor any of the other contributors have been able to prioritize it. In particular, it has been hacked on by a number of folks for a few specific measurements and scenarios seemingly at the expense of some of the more general functionality.
In other words, if you can figure out what's wrong, we would definitely appreciate the contribution. - Rob . On Feb 8, 2013, at 7:45, Allan Vidal <[email protected]> wrote: > Hi, > > I installed oflops from: > http://yuba.stanford.edu/git/gitweb.cgi?p=oflops.git;a=summary. > > I followed the instruction in the README file (exactly as they were), but > neither the libof_port_stat test nor the libof_packet_in tests/modules worked. > > I tried to run on Ubuntu 11.04 with the reference 1.0 software switch. > > I ran the commands as in: > sudo oflops -c lo -d veth1 -d veth3 /usr/local/lib/libof_packet_in.so > > And the errors are: > > For libof_port_stat: > ... > Found unmarked probe.Found unmarked probe.Found unmarked probe.Found unmarked > probe.Found unmarked probe.---- Sending port_stats_requests at rate 1024 per > second for 5 seconds > Found unmarked probe.Found unmarked probe.Found unmarked probe.Found unmarked > probe.Found unmarked probe.Found unmarked probe.Experiment has 10236 packets > sent and 0 received -- 1.000000 dropped (i.e., loss = 10236) with average > delay of -nan us. > lo:162:0 > > And for libof_packet_in: > ... > oflops: msg.c:243: get_next_msg: Assertion `(__extension__ ({ register > unsigned short int __v, __x = (unsigned short int) (ofp->length); if > (__builtin_constant_p (__x)) __v = ((unsigned short int) ((((__x) >> 8) & > 0xff) | (((__x) & 0xff) << 8))); else __asm__ ("rorw $8, %w0" : "=r" (__v) : > "0" (__x) : "cc"); __v; }))' failed. > > > When I inspect the stats requests in Wireshark, I can see the port stats > request is missing its body, and ofdatapath complains: > Feb 08 13:44:18|00291|datapath|WARN|stats request type 4 with bad body length > 0 > > For the packet in I don't know what is causing the error. > > What could be wrong? > > > Allan > _______________________________________________ > openflow-discuss mailing list > [email protected] > https://mailman.stanford.edu/mailman/listinfo/openflow-discuss
_______________________________________________ openflow-discuss mailing list [email protected] https://mailman.stanford.edu/mailman/listinfo/openflow-discuss
