I still can see the following problem reported in the bug description with neutron-server=2:14.0.2-0ubuntu1~cloud0 which actually has included all fixed patches in this bug.
About behavior [2] it is related to missing flows too, some compute nodes have missing output to other compute nodes (vxlan-xxxxxx) which prevent an instance on compute 1 to communicate to an instance on compute 2.
On one of our production system, we saw the following flow rule (sorry, we don't know how to reproduce this problem yet).
I still can see the following problem reported in the bug description with neutron- server= 2:14.0. 2-0ubuntu1~ cloud0 which actually has included all fixed patches in this bug.
About behavior [2] it is related to missing flows too, some compute nodes have missing output to other compute nodes (vxlan-xxxxxx) which prevent an instance on compute 1 to communicate to an instance on compute 2.
On one of our production system, we saw the following flow rule (sorry, we don't know how to reproduce this problem yet).
cookie= 0x7e47f520bdf52 15b, duration= 55484.529s, table=22, n_packets=3441, n_bytes=218308, idle_age=794, priority= 1,dl_vlan= 10 actions= strip_vlan, load:0x40a- >NXM_NX_ TUN_ID[ ],output: 47
port 47 is port vxlan-0af64525
# readlink -f /tmp/tmp. Z4pKS6ZnFC/ vcd41026/ ovs/bridges/ br-tun/ ports/47 Z4pKS6ZnFC/ vcd41026/ ovs/ports/ vxlan-0af64525
/tmp/tmp.
vxlan-0af64525 in a tunnel endpoint with remote_ ip=10.246. 69.37
root@vcd41026:~# ovs-vsctl show| grep 'Interface "vxlan-0af64525"' -A 2 "10.246. 69.32", out_key=flow, remote_ ip="10. 246.69. 37"}
Interface "vxlan-0af64525"
type: vxlan
options: {df_default="true", in_key=flow, local_ip=
where 10.246.69.37 is on vcd41025 but our test vm is running on vcd41021.
So I believe it should be:
cookie= 0x7e47f520bdf52 15b, duration= 55484.529s, table=22, n_packets=3441, n_bytes=218308, idle_age=794, priority= 1,dl_vlan= 10 actions= strip_vlan, load:0x40a- >NXM_NX_ TUN_ID[ ],output: 45,output: 46,output: 47,output: 48,output: 49,output: 50,output: 51,output: 52