This section contains information about how to use the LACP plugin project with OpenDaylight, including configurations.
The LACP Project within OpenDaylight implements Link Aggregation Control Protocol (LACP) as an MD-SAL service module and will be used to auto-discover and aggregate multiple links between an OpenDaylight controlled network and LACP-enabled endpoints or switches. The result is the creation of a logical channel, which represents the aggregation of the links. Link aggregation provides link resiliency and bandwidth aggregation. This implementation adheres to IEEE Ethernet specification 802.3ad.
This feature can be enabled in the Karaf console of the OpenDaylight Karaf distribution by issuing the following command:
feature:install odl-lacp-ui
|
Note
|
|
LACP-discovered network inventory and network statistics can be viewed using the following REST APIs.
-
List of aggregators available for a node:
http://<ControllerIP>:8181/restconf/operational/opendaylight-inventory:nodes/node/<node-id>
Aggregator information will appear within the <lacp-aggregators> XML tag.
-
To view only the information of an aggregator:
http://<ControllerIP>:8181/restconf/operational/opendaylight-inventory:nodes/node/<node-id>/lacp-aggregators/<agg-id>
The group ID associated with the aggregator can be found inside the <lag-groupid> XML tag.
The group table entry information for the <lag-groupid> added for the aggregator is also available in the opendaylight-inventory node database.
-
To view physical port information.
http://<ControllerIP>:8181/restconf/operational/opendaylight-inventory:nodes/node/<node-id>/node-connector/<node-connector-id>
Ports that are associated with an aggregator will have the tag <lacp-agg-ref> updated with valid aggregator information.
The below tutorial demonstrates LACP LAG creation for a sample mininet topology.
sudo mn --controller=remote,ip=<Controller IP> --topo=linear,1 --switch ovsk,protocols=OpenFlow13
The above command will create a virtual network consisting of a switch and a host. The switch will be connected to the controller.
Once the topology is discovered, verify the presence of a flow entry with "dl_type" set to "0x8809" to handle LACP packets using the below ovs-ofctl command:
ovs-ofctl -O OpenFlow13 dump-flows s1 OFPST_FLOW reply (OF1.3) (xid=0x2): cookie=0x300000000000001e, duration=60.067s, table=0, n_packets=0, n_bytes=0, priority=5,dl_dst=01:80:c2:00:00:02,dl_type=0x8809 actions=CONTROLLER:65535
Configure an additional link between the switch (s1) and host (h1) using the below command on mininet shell to aggregate 2 links:
mininet> py net.addLink(s1, net.get('h1'))
mininet> py s1.attach('s1-eth2')
The LACP module will listen for LACP control packets that are generated from legacy switch (non-OpenFlow enabled). In our example, host (h1) will act as a LACP packet generator. In order to generate the LACP control packets, a bond interface has to be created on the host (h1) with mode type set to LACP with long-timeout. To configure bond interface, create a new file bonding.conf under the /etc/modprobe.d/ directory and insert the below lines in this new file:
alias bond0 bonding options bonding mode=4
Here mode=4 is referred to LACP and the default timeout is set to long.
Enable bond interface and associate both physical interface h1-eth0 & h1-eth1 as members of the bond interface on host (h1) using the below commands on the mininet shell:
mininet> py net.get('h1').cmd('modprobe bonding')
mininet> py net.get('h1').cmd('ip link add bond0 type bond')
mininet> py net.get('h1').cmd('ip link set bond0 address <bond-mac-address>')
mininet> py net.get('h1').cmd('ip link set h1-eth0 down')
mininet> py net.get('h1').cmd('ip link set h1-eth0 master bond0')
mininet> py net.get('h1').cmd('ip link set h1-eth1 down')
mininet> py net.get('h1').cmd('ip link set h1-eth1 master bond0')
mininet> py net.get('h1').cmd('ip link set bond0 up')
Once the bond0 interface is up, the host (h1) will send LACP packets to the switch (s1). The LACP Module will then create a LAG through exchange of LACP packets between the host (h1) and switch (s1). To view the bond interface output on the host (h1) side:
mininet> py net.get('h1').cmd('cat /proc/net/bonding/bond0')
Ethernet Channel Bonding Driver: v3.7.1 (April 27, 2011)
Bonding Mode: IEEE 802.3ad Dynamic link aggregation
Transmit Hash Policy: layer2 (0)
MII Status: up
MII Polling Interval (ms): 100
Up Delay (ms): 0
Down Delay (ms): 0
802.3ad info
LACP rate: slow
Min links: 0
Aggregator selection policy (ad_select): stable
Active Aggregator Info:
Aggregator ID: 1
Number of ports: 2
Actor Key: 33
Partner Key: 27
Partner Mac Address: 00:00:00:00:01:01
Slave Interface: h1-eth0 MII Status: up Speed: 10000 Mbps Duplex: full Link Failure Count: 0 Permanent HW addr: 00:00:00:00:00:11 Aggregator ID: 1 Slave queue ID: 0
Slave Interface: h1-eth1 MII Status: up Speed: 10000 Mbps Duplex: full Link Failure Count: 0 Permanent HW addr: 00:00:00:00:00:12 Aggregator ID: 1 Slave queue ID: 0
A corresponding group table entry would be created on the OpenFlow switch (s1) with "type" set to "select" to perform the LAG functionality. To view the group entries:
mininet>ovs-ofctl -O Openflow13 dump-groups s1 OFPST_GROUP_DESC reply (OF1.3) (xid=0x2): group_id=60169,type=select,bucket=weight:0,actions=output:1,output:2
To apply the LAG functionality on the switches, the flows should be configured with action set to GroupId instead of output port. A sample add-flow configuration with output action set to GroupId:
sudo ovs-ofctl -O Openflow13 add-flow s1 dl_type=0x0806,dl_src=SRC_MAC,dl_dst=DST_MAC,actions=group:60169