Hi, I am using openvswitch-2.0.0 ovs. 1. Have Switch connected to OF controller. 2. Node is connected. 3. Run iperf command from node to trigger a lot of packet-in. "iperf -c <ip> -t 10000 -u -b 1000MB -l 466" where <ip> has static arp configured. 4. After >1hr. Switch disconnect from OF controller. 5. Check ping to SDN Controller observe that ping is reachable. 6. Stop all traffic and observe that connection is not restore
[Root cause] CPU usage for openvswitchd is more then 52% (Check /var/log/openvswitchd.log or ps -aux) Note: Reboot is required to recover the Switch WLC is able to send packet-in rate of 4000-7000 pps (64,512,1500 bytes). This is the reason affecting WLC CPU. packet-in rate may need to rate-limited to protect the CPU, excess rate (non-LLDP) should be dropped ........................... Could we stop this problem somewhere from ovs code ? -- Thanks & Regards, _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
