On Sat, 20 Apr 2019 at 09:03, vladimir franciz blando < vladimir.blando@gmail.com> wrote:

Hi All,

Deploying an OpenStack Rocky (1 controller+ceph, 1 compute+ceph) environment on Packet.net using kolla-ansible 7.1.0. Everything works except Instance connectivity going out.

controller1 network http://paste.openstack.org/show/749565/ compute1 network http://paste.openstack.org/show/749566/ openvswitch show http://paste.openstack.org/show/749567/

1. Instance to Instance PING via it's private IP = success 2. Instance to Instance PING via floating IP = success 3. Instance to gateway ping = success 4. Instance to outside world (ex. 8.8.8.8 or google.com) = failed 5. PINGing the instance from the controller node using the floating IP = failed

Here is my global.yml file http://paste.openstack.org/show/lpAGgrEm8rNe43WeB8Uo/ Here is my multinode file http://paste.openstack.org/show/DL7l1WG2xotH5KziwuyL/ Here is my init-runounce file http://paste.openstack.org/show/749568/

- Vlad

Hi Vlad, I see you're using DVR. I'm not too familiar with it, but it sounds like your external networking on the controller node is not setup properly. They need to have external network access. There should be a network namespace for the neutron router, and this should also have external network connectivity in order to allow SNAT to work (try using ip netns exec <ns> <command>). Regards, Mark

ᐧ