Dear,
I tried it, and i always get response from cbench like:
01:51:29.823 32 switches: fmods/sec: 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 total =
0.000000 per ms
and from pox:
INFO:openflow.of_01:[Con 30/11] Connected to 00-00-00-00-00-0b
INFO:openflow.of_01:[Con 41/24] Connected to 00-00-00-00-00-18
INFO:openflow.of_01:[Con 42/25] Connected to 00-00-00-00-00-19
INFO:openflow.of_01:[Con 43/26] Connected to 00-00-00-00-00-1a
INFO:openflow.of_01:[Con 44/27] Connected to 00-00-00-00-00-1b
INFO:openflow.of_01:[Con 45/28] Connected to 00-00-00-00-00-1c
INFO:openflow.of_01:[Con 46/29] Connected to 00-00-00-00-00-1d
INFO:openflow.of_01:[Con 47/30] Connected to 00-00-00-00-00-1e
INFO:openflow.of_01:[Con 48/31] Connected to 00-00-00-00-00-1f
INFO:openflow.of_01:[Con 49/32] Connected to 00-00-00-00-00-20
INFO:openflow.of_01:[Con 50/18] Connected to 00-00-00-00-00-12
I got the last oflops(with path)
I run pox as ./pox.py log.level --DEBUG openflow.of_01 --address=localhost
--port=6633
and cbench as /cbench -c localhost -p 6633 -m 10000 -l 10 -s 32 -M 1000000
-t -d
Sometimes i got this msg from cbench: make_tcp_connection: connect:
Operation now in progress
make_nonblock_tcp_connection
and this messages from pox : POX> INFO:openflow.of_01:[Con 2/None]
Connection reset
INFO:openflow.of_01:[Con 2/None] closing connection
INFO:openflow.of_01:[Con 9/None] Connection r
Anyone has any idea about it?eset
INFO:openflow.of_01:[Con 9/None] closing connection
INFO:openflow.of_01:[Con 17/None] Connection reset
Anyone has any idea about it?
Thks!
2012/10/26 Murphy McCauley <murphy.mccau...@gmail.com>
> Thanks for looking at this, you guys. cbench has, indeed, been run on
> POX, but it was a pretty long time ago -- either so long ago that it
> predated the barrier during connection or I just hacked POX's connection
> process to work at the time and forgot about it. It'd be great to have it
> work out of the box.
>
> (And yeah, I think 8k is probably reasonable with CPython. Under PyPy,
> you can get significantly better.)
>
> -- Murphy
>
> > Thanks, that worked a treat! I added the following patch to
> > fakeswitch.c, and now I'm able to get responses from my POX controller:
> >
> > Index: cbench/fakeswitch.c
> > ===================================================================
> > --- cbench/fakeswitch.c (revision 135)
> > +++ cbench/fakeswitch.c (working copy)
> > @@ -261,6 +261,7 @@
> > int count;
> > struct ofp_header * ofph;
> > struct ofp_header echo;
> > + struct ofp_header barrier;
> > char buf[BUFLEN];
> > count = msgbuf_read(fs->inbuf, fs->sock); // read any queued data
> > if (count <= 0)
> > @@ -357,6 +358,14 @@
> > echo.xid = ofph->xid;
> > msgbuf_push(fs->outbuf,(char *) &echo, sizeof(echo));
> > break;
> > + case OFPT_BARRIER_REQUEST:
> > + debug_msg(fs, "got barrier, sent barrier_resp");
> > + barrier.version= OFP_VERSION;
> > + barrier.length = htons(sizeof(barrier));
> > + barrier.type = OFPT_BARRIER_REPLY;
> > + barrier.xid = ofph->xid;
> > + msgbuf_push(fs->outbuf,(char *) &barrier,
> sizeof(barrier));
> > + break;
> > case OFPT_STATS_REQUEST:
> > stats_req = (struct ofp_stats_request *) ofph;
> > if ( ntohs(stats_req->type) == OFPST_DESC ) {
> >
> >
> > On Thu, Oct 25, 2012 at 03:50:41PM -0700, Rob Sherwood wrote:
> >> On Thu, Oct 25, 2012 at 3:41 PM, Peter Fales
> >> <peter.fa...@alcatel-lucent.com> wrote:
> >>> On Thu, Oct 25, 2012 at 02:59:12PM -0700, Rob Sherwood wrote:
> >>>> 2) it would be pretty easy (~5 lines of code) to add a barrier reply
> >>>> handler to cbench if you were so inclined.
> >>>
> >>> Thanks!
> >>>
> >>> I would like to pursue that, but I don't know anything out the cbench
> >>> code. Any suggestions on where to start looking?
> >>
> >> In fakeswitch.c:259 or so, there is the function:
> >>
> >> void fakeswitch_handle_read(struct fakeswitch *fs)
> >>
> >> that has the openflow state machine. It should be a matter of adding
> >> a case for a barrier request and creating and sending the barrier
> >> reply. For the example for echo request.
> >>
> >> Would love to get support for POX in place -- thanks!
> >>
> >> - Rob
>
> _______________________________________________
> openflow-discuss mailing list
> openflow-discuss@lists.stanford.edu
> https://mailman.stanford.edu/mailman/listinfo/openflow-discuss
>
_______________________________________________
openflow-discuss mailing list
openflow-discuss@lists.stanford.edu
https://mailman.stanford.edu/mailman/listinfo/openflow-discuss
