Hi,

I'm currently doing some measurements concerning the clustering of ODL:


In my setup I use ODL 5.2.0 in a three-node cluster and connect it to 10 OVS 
switches using Mininet. I then take down a node in the three-node cluster that 
is master to at least some of the switches and use tcpdump to capture the 
packets.


While analyzing a couple of runs I was wondering about a couple of things and 
I'd be really thankful if you could explain them to me:


*Sometimes there don't seem to be any OSPF_ROLE_REQUEST messages but only 
OSPF_ROLE_REPLY messages

*The switches all seem to be assigned to the same master (also: is there any 
way to do load balancing?)

*What is the algorithm behind switch failover? (a high-level overview would be 
cool already, I'm really just trying to avoid having to dig into the source 
code here)


Thanks a lot

Felix
_______________________________________________
openflowplugin-dev mailing list
[email protected]
https://lists.opendaylight.org/mailman/listinfo/openflowplugin-dev

Reply via email to