GitHub user joolli added a comment to the discussion: Host based routing - EVPN-VXLAN to the host - Single VXLAN Device (SVD)
Thank you for offering to try it out! I did not have to make changes elsewhere in Cloudstack. This configuration builds on the VLAN setup, except the physical link has been moved inside the host and you can think of the vlan-aware bridge as the switch that the host used to connect to in a normal vlan setup. The veth is the ethernet cable to the switch. You can have as many veth to the vlan-aware bridge as you like, if you want to segregate traffic for some reason. You're right. The script assumes the same number for the VNI and the bridge, and the bridge (the vni used in the bridge name) in traditional mode... And I kept the same way of thinking, realizing that didn't work out as some VNI's are outside the VLAN range but then I got stuck in getting this thing to work with add and delete and forgot about the VNI range outside the vlan range. I too have VNI's outside the VNI range in production. I'll have to fix that. SVD mode can map any vlan to any VNI, so to solve that we can allocate the next available vlan for a VNI... Libvirt won't care as it just cares about the bridge name... On delete we will check which vlan is mapped to the VNI to be deleted and delete that along with the VNI. How does that sound? I use ifupdown2 to specify the interfaces. I haven't done it in networkd, netplan or NM but I'm willing to help you to get it to work. Ifupdown2 is not in the Ubuntu repos, last time I checked so you will have to add a repo if you want to try it. I can provide you with the iproute2 commands, that way you can use ifupdown or ifupdown-ng (which is in the ubuntu repo, I think) and use "up" or "post-up" commands for the ifaces. GitHub link: https://github.com/apache/cloudstack/discussions/10786#discussioncomment-13348443 ---- This is an automatically sent email for users@cloudstack.apache.org. To unsubscribe, please send an email to: users-unsubscr...@cloudstack.apache.org
