Hi,


So here is the setup I've with Openvswitch 1.1.0 release running with brcompat 
module. It is all running on a single physical machine.



Xen VM (a - 10.19.128.34) --- vif 1.0 -- (pibr102 - bridge) -- vif 2.0 --- Xen 
VM (b - 10.19.128.36)



[root@vm-container-0-1 ~]# ovs-dpctl show pibr102
system@pibr102<mailto:system@pibr102>:
 lookups: frags:0, hit:1683092, missed:27741, lost:0
 port 0: pibr102 (internal)
 port 1: vif1.0
 port 2: vif2.0
 port 3: eth0.102



pibr102 is connected to the two vifs and eth0.102



[root@vm-container-0-1 ~]# brctl show
bridge name bridge id  STP enabled interfaces
pibr102  0000.001e68c6072f no  eth0.102
                                               vif2.0
                                               vif1.0



I can connect between Xen VM (a) and (b) and the iperf results running from VM 
(a) shows the bandwidth as expected without any QoS applied.



-bash-3.2# iperf -c 10.19.128.36
------------------------------------------------------------
Client connecting to 10.19.128.36, TCP port 5001
TCP window size: 16.0 KByte (default)
------------------------------------------------------------
[  3] local 10.19.128.34 port 46323 connected with 10.19.128.36 port 5001
[ ID] Interval       Transfer     Bandwidth
[  3]  0.0-10.0 sec  1.61 GBytes  1.38 Gbits/sec



Then I added a new queue and assigned it to pibr102 which should limit the 
bandwidth to approx 10MB/s



[root@vm-container-0-1 ~]# ovs-vsctl list queue
_uuid               : ed574d30-5045-4628-a8fc-7191c0c703b5
external_ids        : {}
other_config        : {max-rate="10000000", min-rate="10000000"}



[root@vm-container-0-1 ~]# ovs-vsctl list qos
_uuid               : 9518570d-6b8c-4e64-b969-eed0cf1e5b05
external_ids        : {}
other_config        : {}
queues              : {0=ed574d30-5045-4628-a8fc-7191c0c703b5}
type                : linux-htb



[root@vm-container-0-1 ~]# ovs-vsctl find port name=pibr102
_uuid               : 64300bda-0622-4026-bf6f-9b9ceb16daee
bond_downdelay      : 0
bond_fake_iface     : false
bond_mode           : []
bond_updelay        : 0
external_ids        : {}
fake_bridge         : false
interfaces          : [773bbe5a-3216-4b1d-8c0a-690b6616b182]
lacp                : []
mac                 : []
name                : "pibr102"
other_config        : {}
qos                 : 9518570d-6b8c-4e64-b969-eed0cf1e5b05
tag                 : []
trunks              : []



Also added a couple of flows for attaching to the queue



[root@vm-container-0-1 ~]# ovs-ofctl add-flow pibr102 'priority=65535, 
idle_timeout=50000, dl_type=0x800, in_port=ANY, nw_src=ANY, 
nw_dst=10.19.128.34, actions=enqueue:1:0,NORMAL'
[root@vm-container-0-1 ~]# ovs-ofctl add-flow pibr102 'priority=65535, 
idle_timeout=50000, dl_type=0x800, in_port=ANY, nw_src=ANY, 
nw_dst=10.19.128.36, actions=enqueue:2:0,NORMAL'



[root@vm-container-0-1 ~]# ovs-ofctl dump-flows pibr102
NXST_FLOW reply (xid=0x4):
 cookie=0x0, duration=207.266s, table_id=0, n_packets=295013, 
n_bytes=18123024906, idle_timeout=50000,priority=65535,ip,nw_dst=10.19.128.36 
actions=enqueue:2q0,NORMAL
 cookie=0x0, duration=183.287s, table_id=0, n_packets=95640, n_bytes=6357180, 
idle_timeout=50000,priority=65535,ip,nw_dst=10.19.128.34 
actions=enqueue:1q0,NORMAL
 cookie=0x0, duration=278.315s, table_id=0, n_packets=2361, n_bytes=273754,  
actions=NORMAL



But running the iperf test again showed no change in the results



-bash-3.2# iperf -c 10.19.128.36
------------------------------------------------------------
Client connecting to 10.19.128.36, TCP port 5001
TCP window size: 16.0 KByte (default)
------------------------------------------------------------
[  3] local 10.19.128.34 port 40373 connected with 10.19.128.36 port 5001
[ ID] Interval       Transfer     Bandwidth
[  3]  0.0-10.0 sec  1.19 GBytes  1.02 Gbits/sec



n_packets does show that the traffic is matching those flow entries but I don't 
see it limiting the bandwidth. I'm kinda stuck at this point. Is there anything 
I am missing here?



thanks,
Nauman


_______________________________________________
openflow-discuss mailing list
[email protected]
https://mailman.stanford.edu/mailman/listinfo/openflow-discuss

Reply via email to