Hi, I have a OpenStack environment identical to this tutorial<https://github.com/mseknibilel/OpenStack-Grizzly-Install-Guide/blob/OVS_MultiNode/OpenStack_Grizzly_Install_Guide.rst>. It has a network node and use GRE to tunnel traffic to the VM.
I have associated a VM with a floating IP. I ping that VM on the public IP, the initial packet always have a exceptionally high response time then its following packets. (200ms more than following packets) I tcpdump'd on all interfaces of the network node and found that, the longest time taken is when substituting source IP of the ICMP echo reply packet from 192.168.2.2 (VM's vNIC IP address) to the floating public IP address. I do not see this latency issue with following ICMP packets. I (blindly) guess bottleneck is with ovsdb-server, the DB server may be responsing too slow, but I couldn't find method to measure ovsdb-server's query response time. Can anyone here provide me some insights? On profiling ovsdb-server or other places to check for performance problems. Thanks a lot!
_______________________________________________ discuss mailing list discuss@openvswitch.org http://openvswitch.org/mailman/listinfo/discuss
