Conflict between EVNP L3 multicast and EVPN L2 multicast

[krikoon73]

Issue Summary

Hi, Context :

• 1x Mcast VRF with 10x SVI/VLANs (210 to 219)
• 1 mcast underlay group for Mcast VRF
• 1x underlay groups for just one vlan (210) Issue :
• The generated configuration is the following :

  interface vxlan 1
  ...
  vxlan vlan 210 multicast group 232.231.0.210
  vxlan vrf MCAST multicast group 232.232.0.21
  ... 
  router bgp 65101
  ...
  vlan 210
    rd 10.255.0.3:10210
    route-target both 10210:10210
    redistribute igmp
    redistribute learned
  vrf MCAST
    rd 10.255.0.3:21
    evpn multicast
    route-target import evpn 21:21
    route-target export evpn 21:21
    router-id 10.255.0.3
    redistribute connected

• According to the TOI => https://www.arista.com/en/support/toi/eos-4-25-1f/14670-multicast-evpn-irb the VLAN:
  • redistribute igmp should not be configured in the control-plane because there is an SVI associated to that vlan
  • for the data-plane, there is 2 options :
  • vxlan vlan 210 multicast group 232.231.0.210 => will take precedence over the VRF mcast group for OISM+BUM
  • vxlan vlan 210 flood group 232.231.0.210 => for optimizing BUM traffic only

Can you confirm and fix it eventually in the model ?

Best Regards, Christophe

Which component(s) of AVD impacted

eos_designs

How do you run AVD ?

Ansible CLI (with virtual-env or native python)

Steps to reproduce

tenants:
  # Definition of tenants. Additional level of abstraction to VRFs
  - name: Tenant-A
    evpn_l2_multicast:
       underlay_l2_multicast_group_ipv4_pool: 232.231.0.0/20
       underlay_l2_multicast_group_ipv4_pool_offset: 1
    evpn_l3_multicast:
      evpn_underlay_l3_multicast_group_ipv4_pool: 232.232.0.0/20
      evpn_underlay_l3_multicast_group_ipv4_pool_offset: 1
...
    l2vlans:
    vrfs:
...
      - name: MCAST
        vrf_id: 21
        vrf_vni: 21
        vtep_diagnostic:
          loopback: 21
          loopback_description: vtep_diagnostic
          loopback_ip_range: 10.254.0.0/24
        evpn_l3_multicast: 
          enabled: true
          evpn_peg:
            - nodes: [ dc1-blm1a-mrv356, dc1-blm1b-mrv357 ]
              # transit: true 
        svis:
          - id: 210
            name: Flow_Multicast
            profile: GENERIC
            tags: [MCAST]
            evpn_l2_multicast:
              enabled: true
            ip_address_virtual: 100.20.210.1/24
            ipv6_address_virtuals: ["fc00:100:20:210::1/64"]

Relevant log output

No response

Contributing Guide

• [X] I agree to follow this project's Code of Conduct

[mtache]

2031 was split into 2 features, we are interested in #2033

[ClausHolbechArista]

Thank you for reporting this. It is not clear to me if this is actually a bug or a request to get support for flood groups. Please update the issue accordingly. Thanks!

If this is in fact a request to get #2033 implemented, please comment on that issue instead and close this one. We do stretched thin these days, so we will need volunteers to take on implementation of flood groups.

[krikoon73]

Hey Claus, Thank you for your feedback ! I think there is multiple subjects here :

• having redistribute igmp for a svi/vlan belonging to a VRF with evpn multicast is an issue because it could cause conflicts in the control-plane as far I know.
• the flood group or multicast group for a specific VLAN/VXLAN is a data-plane optimisation is order to optimise the core
• IHMO, the n:1 mapping model for a VRF is suboptimal, we have to propose M:N model as described here => https://www.arista.com/en/support/toi/eos-4-29-1f/16546-multicast-evpn-irb-multiple-underlay-groups

I'll have cycles beginning February to discuss about this. Cheers, /cc

[github-actions[bot]]

This issue is stale because it has been open 90 days with no activity. The issue will be reviewed by a maintainer and may be closed