[ovs-discuss] Packet drops with high rate of Packet_In

[Anton Matsiuk]

Hello Ben,
thank you for help!

as Open vSwitch stores only 256 Packet_Ins in buffer (OFPT_FEATURES_REPLY:
n_buffers = 256) after 256 packets it sets Buffer_ID to 0xffffffff.
Controller didn't check this value as incorrect and placed it into
responses in flow_mods and packet_outs, that caused packet drops after
approximately 300 packet_in in a burst.

BTW, setting miss_send_len = 65535 forces switch to send complete Packet_In
to Controller but switch still stores packets in buffer and adds valid
buffer_id for first 255 of them. Is it possible to turn off buffering in
switch (and to set buffer_id = -1 for all packets) using OpenFlow 1.0? (In
OpenFlow 1.2+ it should be possible by setting OFPCML_NO_BUFFER as I
understood).

Is it possible to vary the size of buffer for Packet_In in Open vSwitch?

-- 
Best regards,
Anton Matsiuk

On 22 November 2013 18:52, Ben Pfaff <blp at nicira.com> wrote:

> On Fri, Nov 22, 2013 at 05:30:47PM +0100, Anton Matsiuk wrote:
> > >
> > > Does the controller get any error replies from Open vSwitch?
> >
> >
> > No, Open vSwitch just accepts all the rules (500) and installs them
> without
> > sending any errors to a controller, but processes only part of ingress
> > packets through them:
> > ovs-ofctl dump-aggregate br0
> > NXST_AGGREGATE reply (xid=0x4): packet_count=324 byte_count=20736
> > flow_count=500
>
> OK, thanks for the details.
>
> Are the switch and the controller using buffered packets?  That is, is
> the controller sending 500 flow_mods and 500 packet_outs, or is it
> just sending 500 flow_mods?  If it is sending only flow_mods, can you
> try disabling packet buffering in the switch?  (One way to do this is
> to have the controller set a miss_send_len of 65535.)
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://openvswitch.org/pipermail/ovs-discuss/attachments/20131125/40a56299/attachment.html>