All; I apologize for disappearing from this discussion, I got really busy. Static IPs still could not ping each other, I did test that. I noticed that, even though the subnet was selected for DHCP, the DHCP port didn't exist. I disabled DHCP for the subnet, and then re-enabled it, and everything started working. The port disappeared again later, however, when there weren't any VMs in the project / attached to the network. I think I know what happened, and possibly even why, though I have no evidence of it. When I originally configured the actual provider network for the addition of this VLAN, I missed adding it to the network nodes' connection. Thus when I configured the network & subnet on OpenStack, the network node failed... something? I fixed the underlying network issue, then, several days later, removed and re-added the DHCP support, as noted above, and things started working. Thank you, Dominic L. Hilsbos, MBA Vice President - Information Technology Perform Air International Inc. DHilsbos@PerformAir.com www.PerformAir.com -----Original Message----- From: Slawek Kaplonski [mailto:skaplons@redhat.com] Sent: Monday, August 2, 2021 8:05 AM To: Dominic Hilsbos Cc: openstack-discuss@lists.openstack.org Subject: Re: [neutron][ops] VLAN Provider Network Hi, On Mon, Jul 26, 2021 at 09:30:59PM +0000, DHilsbos@performair.com wrote:

All;

We're using OpenStack Victoria, with OVS networking (no OVN). I'm having some problems with a VLAN provider network that I added to my OpenStack. I have other VLAN provider networks, that I based this one off of.

Instances are getting the tap interfaces, but these interfaces are not getting IP addresses, and when a static IP is given, they can't ping either the DHCP address or the gateway address.

Can instances with static IPs ping each other or it also don't works?

I have checked the switch configuration, to verify the VLAN is bound appropriately. I rebooted the whole cluster to ensure the configuration change applied.

Here's a snip our plugin.ini [ml2_type_vlan] network_vlan_ranges = provider_core:55:55,provider_core:64:64,...

64 works, 55, doesn't.

I'm not seeing anything concerning in /var/log/openvswitch/ovs-vswitchd.log

I'm sure it's something dumb, but I'm not seeing it.

Please check also configuration of the OVS agent on the compute nodes where instances are configured and on the servers where DHCP agent's are running.

Any suggestions for what else to look at?

You should also check with tcpdump where packets are actually dropped. You can start checking on the physical interface in the host if there are packets with vlan_id 55 (the one which don't works). If packets are going out from compute node, You can check on Your switches and move to the next server. If packets are not on that physical interface, You should look in the Openflow rules configuration in the openvswitch on the compute node.

Thank you,

Dominic L. Hilsbos, MBA Vice President - Information Technology Perform Air International Inc. DHilsbos@PerformAir.com www.PerformAir.com

-- Slawek Kaplonski Principal Software Engineer Red Hat