[ovs-discuss] Unexpected behavior of counters when installing new flow

Tmusic Tmusic987 at gmail.com
Tue Apr 30 10:49:26 UTC 2013


Let me post some more debug information then:

so the traffic is a udp stream from 10.1.4.1 port 20 to 10.1.6.2 port 1000
(3000 packet/s with 1024 bytes packet size)

Original situation (last line is the relevant flow):
NXST_FLOW reply (xid=0x4):
 cookie=0x3ea, duration=10.568s, table=0, n_packets=15, n_bytes=8850,
idle_age=1, icmp,nw_src=10.1.6.0/30,nw_dst=10.1.4.0/30 actions=output:1
 cookie=0x0, duration=32.941s, table=0, n_packets=21, n_bytes=1260,
idle_age=0, priority=65000,dl_dst=01:23:20:00:00:01,dl_type=0x88cc
actions=CONTROLLER:65535
 cookie=0x3eb, duration=5.554s, table=0, n_packets=0, n_bytes=0,
idle_age=5, arp,dl_src=00:25:90:64:f9:3e,dl_dst=90:e2:ba:00:9f:86
actions=output:2
 cookie=0x3e8, duration=11.603s, table=0, n_packets=2, n_bytes=120,
idle_age=5, arp,dl_src=90:e2:ba:00:9f:86,dl_dst=00:25:90:64:f9:3e
actions=output:1
 cookie=0x3e9, duration=10.605s, table=0, *n_packets=28907, n_bytes=30814862
*, idle_age=0, udp,nw_src=
10.1.4.0/30,nw_dst=10.1.6.0/30,tp_src=20,tp_dst=1000 actions=output:2


A few seconds after installing the rule with higher priority (first line is
new flow, last line is old):

 cookie=0x3ec, duration=1.846s, table=0, *n_packets=40919, n_bytes=43619654*,
idle_age=0, priority=65100,udp,nw_src=
10.1.4.0/31,nw_dst=10.1.6.2/31,tp_src=20,tp_dst=1000 actions=output:2
 cookie=0x3ea, duration=13.856s, table=0, n_packets=19, n_bytes=11210,
idle_age=0, icmp,nw_src=10.1.6.0/30,nw_dst=10.1.4.0/30 actions=output:1
 cookie=0x0, duration=36.229s, table=0, n_packets=21, n_bytes=1260,
idle_age=3, priority=65000,dl_dst=01:23:20:00:00:01,dl_type=0x88cc
actions=CONTROLLER:65535
 cookie=0x3eb, duration=8.842s, table=0, n_packets=0, n_bytes=0,
idle_age=8, arp,dl_src=00:25:90:64:f9:3e,dl_dst=90:e2:ba:00:9f:86
actions=output:2
 cookie=0x3e8, duration=14.891s, table=0, n_packets=2, n_bytes=120,
idle_age=8, arp,dl_src=90:e2:ba:00:9f:86,dl_dst=00:25:90:64:f9:3e
actions=output:1
 cookie=0x3e9, duration=13.893s, table=0, *n_packets=0, n_bytes=0*,
idle_age=2, udp,nw_src=10.1.4.0/30,nw_dst=10.1.6.0/30,tp_src=20,tp_dst=1000actions=output:2

Does this provides more useful information?

Kind regards,

Tim


2013/4/29 Ben Pfaff <blp at nicira.com>

> On Mon, Apr 29, 2013 at 08:08:12PM +0200, Tmusic wrote:
> > I'm experiencing some unexpected behavior from the flow counters when
> using
> > wildcarded flow entries (with OpenFlow).
> >
> >
> > Steps to reproduce:
> > - Start a udp flow ex. 10.1.4.1=>10.1.6.1
> > - Install wildcarded entry 10.1.4.0/30 => 10.1.6.0/30
> > ==>After some time counters (ovs-ofctl dump-flows) indicate for example
> > n_packets=4775, n_bytes=5090150, idle_age=0
> >
> > - Install second entry and set higher priority (ex 65100) : 10.1.4.0/31=>
> > 10.1.6.0/31)
> > ==> Now the counters read ex. n_packets=0, n_bytes=0, idle_age=7
> > ==> The old counter values are added to the new flow n_packets=18775,
> > n_bytes=20014150, idle_age=0
>
> If this happens as described, this sounds like an Open vSwitch bug (a
> rather strange one).
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://openvswitch.org/pipermail/ovs-discuss/attachments/20130430/a948ab3d/attachment.html>


More information about the discuss mailing list