Hello everyone, I'm on version 2.3.2.
I'm seeing a guest OS reject UDP packets due to a bad UDP checksum after crossing OVS, these packets hit a flow that looks like this: cookie=0x0, duration=3189.844s, table=1, n_packets=1567, n_bytes=139010, idle_age=0, priority=1020,ip,dl_dst=04:01:d7:6f:80:01,nw_dst=xx.xx.104.241 actions=strip_vlan,mod_nw_dst:10.10.0.20,output:10 >From wireshark captures, I can see the guest bad checksum error only happens when the host and guest receive the UDP datagram as IP fragments. If there is no IP fragmentation, the checksum is re-calculated correctly. It seems that `ovs-ofctl set-frags public reassemble` might solve this problem, but of course that is currently unsupported. If it helps, the above flow is on the switch closest to the guest. That switch is turn patched to another OVS aggregation switch, which is closest to the host's physical interface. Has anyone seen this issue or have any workarounds? Thanks, Michael
_______________________________________________ discuss mailing list discuss@openvswitch.org http://openvswitch.org/mailman/listinfo/discuss
