[pve-devel] [PATCH pve-docs 0/1] vxlan l3 routing

[Alexandre DERUMIER]

>>That "transport zone" looks interesting. 

>>We you just mark physical interfaces (or bridges?) as part of a transport zone. 
>>Then we have everything to setup the vxlan (each vlan belong to a zone)? 

yes, it should work. (interface could be better I think).

/etc/network/interfaces
iface eth0
  transportzone 1
  address 192.168.0.1

for vlan
--------
/etc/pve/network.cfg

vlan : mycustombridge
       transportzone 1
       vlan 2
       bridge-vlan-aware 1|0  (for QinQ)

This create

iface mycustombridge
     bridge-ports eth0.2
     ....


for vxlan multicast
-------------------
/etc/pve/network.cfg

vxlan : mycustombridge
        transportzone 1
        mode multicast
        multicast_address 225.20.1.1
        vxlan 2
        

This create

iface vxlan2
     vxlan-id 2
     vxlan-svcnodeip 225.20.1.1
     vxlan-physdev eth0

iface mycustombridge
     bridge-ports vxlan2


for vxlan + frr
-------------------
/etc/pve/network.cfg

vxlan : mycustombridge
        transportzone 1
        mode frr
        vxlan 2

this create

 auto vxlan2
 iface vxlan2 inet manual
         vxlan-id 2
         vxlan-local-tunnelip 192.168.0.1  ->get ip from transport interface
         bridge-learning off
         bridge-arp-nd-suppress on
         bridge-unicast-flood off
         bridge-multicast-flood off
 
 
 auto mycustombridge
 iface mycustombridge inet manual
         bridge_ports vxlan2
         bridge_stp off
         bridge_fd 0


for vxlan unicast
-------------------
vxlan : mycustombridge
        transportzone 1
        mode unicast
        peers 192.168.0.1,192.168.0.2,192.168.0.3
        vxlan 2

this create

 auto vxlan2
 iface vxlan2 inet manual
        vxlan-id 2
        vxlan_remoteip 192.168.0.2
        vxlan_remoteip 192.168.0.3   -> add all ips, excluding ip from transportzone interface
 
 
 auto mycustombridge
 iface mycustombridge inet manual
         bridge_ports vxlan2
         bridge_stp off
         bridge_fd 0





Note that it could work with transportzone + ip address on bridge.
(But in this case, maybe we can consider the bridge as the transport zone).

I need to check with underlay layer3 + 2 interface with ecmp, but I think we have a loopback address,
so it could be enslaved in a bridge too.

----- Mail original -----
De: "dietmar" <dietmar at proxmox.com>
À: "aderumier" <aderumier at odiso.com>
Cc: "pve-devel" <pve-devel at pve.proxmox.com>
Envoyé: Mercredi 12 Décembre 2018 10:13:41
Objet: Re: [pve-devel] [PATCH pve-docs 0/1] vxlan l3 routing

> I need to check, they have a concept of "transport zone", seem to be an abstraction between distributed switch and physical host.(and some kind of vrf/vlan isolation) 
> https://www.youtube.com/watch?v=Lsgz88OvxDk 

That "transport zone" looks interesting. 

We you just mark physical interfaces (or bridges?) as part of a transport zone. 
Then we have everything to setup the vxlan (each vlan belong to a zone)?