Ask Your Question

Revision history [back]

click to hide/show revision 1
initial version

kolla-ansible networking sfc load-balancing issue

when multiple port pairs are present in a port pair group only one of them seem to receive any traffic. If I reboot the vm that receives the traffic other ones are still getting nothing and packets stop coming through the chain at all until the rebooted vm is back on.

How does load balancing happen in sfc Reproduce: 1. Create network and subnet 2. Launch 4 vms in this network: source, servicevm1, servicevm2, destination 3. Setup sfc: Create port pairs for servicevm1 and servicevm2 Create a port-pair-group with two port pairs Create a flow classifier with source vm's port as logical-source-port and protocol = icmp Create port chain with one port-pair-group and one flow-classifier 4. On vms, run: source runs: ping destination servicevm1 and servicevm2 runs: tcpdump icmp -i eth0 -n 5. One of the vms receives all packets while the other one gets nothing 6. Reboot the vm that receives packets 7. source vm stops receiving ping answers and the second service vm still gets nothing 8. When the rebooted vm comes back source starts receiving ping answers again

Maybe I'm missing something and this is not how port pair group loadbalancer is supposed to work?

Thanks in advance.