Hi,
I use OVS to simulate a looped topology (bridges connected in a ring and
multiple links between bridges). The workflow is as follows:
1) I setup the entire topology using OVS bridges (I set disable-in-band=true
and remove all flows), veth interfaces to interconnect the bridges and Linux
containers (lxc) to simulate attached hosts.
2) Once topology is setup I connect OVS bridges to OpenFlow controller
This worked very well for me in previous versions of OVS regardless whether the
controller is present or absent.
However when I updated to OVS 2.4.0, if the controller is not reachable when
provisioning it in the bridge or if the bridge loses connection to controller,
the OVS bridges starts to broadcast packets to all ports creating a packet
storm in the case of my topology.
Details of my installation:
OVS:
ovs-vswitchd (Open vSwitch) 2.4.0
Compiled Aug 29 2015 11:20:23
Kernel and OS:
Ubuntu 14.04 LTS
cat /proc/version
Linux version 3.13.0-43-generic (buildd@tipua) (gcc version 4.8.2 (Ubuntu
4.8.2-19ubuntu1) ) #72-Ubuntu SMP Mon Dec 8 19:35:06 UTC 2014
Other printouts:
sudo ovs-dpctl show
system@ovs-system:
lookups: hit:715066804 missed:10477446 lost:10223814
flows: 0
masks: hit:820418364 total:0 hit/pkt:1.13
port 0: ovs-system (internal)
port 1: dc1fb1 (internal)
port 2: dc1fb2 (internal)
port 3: dc2fb1 (internal)
port 4: dc2fb2 (internal)
port 5: dc3fb1 (internal)
port 6: dc3fb2 (internal)
sudo ovs-appctl bridge/dump-flows dc1fb1 (same flows in all 6 bridges)
table_id=254, duration=69s, n_packets=0, n_bytes=0,
priority=2,recirc_id=0,actions=drop
table_id=254, duration=69s, n_packets=0, n_bytes=0,
priority=0,reg0=0x1,actions=controller(reason=no_match)
table_id=254, duration=69s, n_packets=0, n_bytes=0,
priority=0,reg0=0x2,actions=drop
table_id=254, duration=69s, n_packets=0, n_bytes=0,
priority=0,reg0=0x3,actions=drop
Please let me know if you need any other information.
BR/Luis
_______________________________________________
discuss mailing list
[email protected]
http://openvswitch.org/mailman/listinfo/discuss