Hi,

I have set up a bridge with 3 physical ports (vlan_mode=access). The following 
rules are active in order to redirect data sent from 10.0.0.1 destined for 
10.0.0.2 to 10.0.0.3:

 cookie=0x0, duration=2725.863s, table=0, n_packets=n/a, n_bytes=33572528154, 
ip,in_port=25,nw_dst=10.0.0.2 
actions=set_field:10.0.0.3->ip_dst,set_field:96:5f:91:02:cf:3f->eth_dst,output:27
 cookie=0x0, duration=2670.657s, table=0, n_packets=n/a, n_bytes=48832, 
arp,in_port=26 actions=output:25
 cookie=0x0, duration=2680.799s, table=0, n_packets=n/a, n_bytes=4672, 
arp,in_port=25 actions=output:26

10.0.0.1 is connected to port 25, 10.0.0.2 to port 26, 10.0.0.3 to port 27

I have noticed that when sending ~15 MB/s from 10.0.0.1, only about 10% 
actually arrives at 10.0.0.3. The `dump-flows` command shows that the bytes are 
dropped at port 25:

root@PicOS-OVS$ovs-ofctl dump-ports th1
OFPST_PORT reply (OF1.4) (xid=0x2): 4 ports
  port 25: rx pkts=24956726, bytes=33491813828, drop=22503515, errs=0, frame=0, 
over=0, crc=0
           tx pkts=73, bytes=4672, drop=0, errs=0, coll=0
           duration=3355.201s

Aside from an explicit flow that defines a drop action, what could be the cause 
of the bytes being dropped coming into port 25? I have verified that no other 
visible or hidden flows are active that could cause packet drops.

I am using Open vSwitch version 2.0.90 and OpenFlow 1.4 on a pica8 hardware 
switch. No controller is active.

Kind regards,

Damiaan Twelker

_______________________________________________
discuss mailing list
discuss@openvswitch.org
http://openvswitch.org/mailman/listinfo/discuss

Reply via email to