Validate that customer eBGP routes learned on the DUT in tenant VRFs are maintained in the BGP control plane and advertised to core iBGP route reflectors (RR) via advertise-inactive-routes, but are isolated and suppressed from installation into the local transit forwarding table (FIB). Furthermore, validate graceful soft-drain and undrain operations per tenant peer session (via policy-driven LOCAL_PREF deprecation and AS-Path prepending) without traffic disruption to neighboring tenants.

• TESTBED_DUT_ATE_4LINKS (ATE Port 1/2 simulating Customer Edge CE1/CE2; ATE Port 3/4 simulating Core Route Reflector RR1/RR2).

graph LR;
  ce1[ATE Port 1: CE1 in VRF ce1] -- "eBGP (Customer Routes)" --> dut1[DUT Port 1.100]
  ce2[ATE Port 2: CE2 in VRF ce2] -- "eBGP (Customer Routes)" --> dut2[DUT Port 2.200]
  dut3[DUT Port 3: Transit] -- "iBGP (IPv4/IPv6 Unicast)" --> rr1[ATE Port 3: RR1]
  dut4[DUT Port 4: Transit] -- "iBGP (IPv4/IPv6 Unicast)" --> rr2[ATE Port 4: RR2]

• Connect ATE Port 1 to DUT Port 1 (configured with subinterface VLAN 100 in VRF ce1, IP 192.0.2.1/30).
• Connect ATE Port 2 to DUT Port 2 (configured with subinterface VLAN 200 in VRF ce2, IP 192.0.2.5/30).
• Connect ATE Port 3 to DUT Port 3 (configured with IPv6 transit peering 2001:db8:3::1/64 to RR1 2001:db8:3::2 in VRF ce1).
• Connect ATE Port 4 to DUT Port 4 (configured with IPv6 transit peering 2001:db8:4::1/64 to RR2 2001:db8:4::2 in VRF ce2).

1. Configure two tenant network instances (ce1, ce2) on the DUT with distinct Route Distinguishers (RD) and Route Targets (RT).
2. Configure eBGP sessions under VRF ce1 (with ATE CE1 192.0.2.2, AS 65001) and VRF ce2 (with ATE CE2 192.0.2.6, AS 65002).
3. Configure iBGP sessions towards ATE RR1 (2001:db8:3::2 in VRF ce1) and ATE RR2 (2001:db8:4::2 in VRF ce2) with IPV4_UNICAST and IPV6_UNICAST address families enabled.
4. Configure route install suppression policy on DUT: routes learned via customer eBGP in tenant VRFs must have table-install suppressed (install-map NO-ROUTES or routing policy reject on RIB-to-FIB download).
5. Enable /network-instances/network-instance/protocols/protocol/bgp/global/route-selection-options/config/advertise-inactive-routes (true) on the DUT BGP process so that valid BGP routes not installed in the FIB continue to be advertised to iBGP peers.

• Step 1: ATE CE1 advertises prefix 198.51.100.0/24 (IPv4) and 2001:db8:100::/64 (IPv6) to DUT in VRF ce1.
• Step 2: ATE CE2 advertises prefix 203.0.113.0/24 (IPv4) and 2001:db8:200::/64 (IPv6) to DUT in VRF ce2.
• Step 3: Verify via OpenConfig telemetry /network-instances/network-instance[name=ce1]/table-connections/table-connection/state/import-policy and AFT telemetry /network-instances/network-instance[name=ce1]/afts/ that VRF ce1 contains no forwarding entries for 203.0.113.0/24 or 2001:db8:200::/64.
• Step 4: Verify VRF ce2 contains no forwarding entries for 198.51.100.0/24 or 2001:db8:100::/64.
• Step 5: Send bidirectional data plane probe packets from ATE CE1 destined to 203.0.113.0/24 and 2001:db8:200::/64; verify traffic is dropped on the DUT and does not leak cross-VRF.

• Step 1: Verify IPv4 prefix 198.51.100.0/24 and IPv6 prefix 2001:db8:100::/64 are present in the DUT BGP RIB under /network-instances/network-instance[name=ce1]/protocols/protocol[identifier=openconfig-policy-types:BGP][name=BGP]/bgp/rib/afi-safis/afi-safi[afi-safi-name=openconfig-bgp-types:IPV4_UNICAST]/ipv4-unicast/loc-rib/routes/route/state/prefix and .../afi-safi[afi-safi-name=openconfig-bgp-types:IPV6_UNICAST]/ipv6-unicast/loc-rib/routes/route/state/prefix.
• Step 2: Query the DUT AFT forwarding state at /network-instances/network-instance[name=ce1]/afts/ipv4-unicast/ipv4-entry[prefix=198.51.100.0/24]/state/prefix and /network-instances/network-instance[name=ce1]/afts/ipv6-unicast/ipv6-entry[prefix=2001:db8:100::/64]/state/prefix.
• Step 3: Verify that neither route is installed in the hardware FIB table (or state leaf reflects not-installed / transit-only suppression).
• Step 4: Verify that core transit traffic passing through DUT does not attempt local host termination for suppressed customer prefixes.

• Step 1: Confirm /network-instances/network-instance[name=ce1]/protocols/protocol[identifier=openconfig-policy-types:BGP][name=BGP]/bgp/global/route-selection-options/state/advertise-inactive-routes is true and monitor the iBGP advertisements sent from DUT Port 3 to ATE RR1 (2001:db8:3::2).
• Step 2: Verify that ATE RR1 receives 198.51.100.0/24 (IPv4) and 2001:db8:100::/64 (IPv6) as iBGP updates.
• Step 3: Verify that /network-instances/network-instance[name=ce1]/protocols/protocol[identifier=openconfig-policy-types:BGP][name=BGP]/bgp/rib/afi-safis/afi-safi[afi-safi-name=openconfig-bgp-types:IPV4_UNICAST]/ipv4-unicast/neighbors/neighbor[neighbor-address=2001:db8:3::2]/adj-rib-out-post/routes/route/state/prefix, .../afi-safi[afi-safi-name=openconfig-bgp-types:IPV6_UNICAST]/ipv6-unicast/neighbors/neighbor[neighbor-address=2001:db8:3::2]/adj-rib-out-post/routes/route/state/prefix, and /network-instances/network-instance[name=ce1]/protocols/protocol[identifier=openconfig-policy-types:BGP][name=BGP]/bgp/neighbors/neighbor[neighbor-address=2001:db8:3::2]/afi-safis/afi-safi/state/prefixes/sent reflect the advertised 198.51.100.0/24 and 2001:db8:100::/64 prefixes.

• Step 1: Establish steady-state traffic between ATE CE1 and ATE RR1.
• Step 2: Apply a soft drain policy on DUT for ATE CE1 peer (deprecating local-pref to 10 towards core iBGP peers and prepending AS-Path 64500 64500 64500 towards CE1 without resetting TCP transport).
• Step 3: Verify via /network-instances/network-instance[name=ce1]/protocols/protocol[identifier=openconfig-policy-types:BGP][name=BGP]/bgp/rib/attr-sets/attr-set/state/local-pref and ATE RR1 BGP RIB that 198.51.100.0/24 and 2001:db8:100::/64 are re-advertised towards ATE RR1 with deprecated local-pref (10) and towards ATE CE1 with prepended AS-Path ([64500, 64500, 64500]), gracefully draining traffic away from the link without hard route withdrawal.
• Step 4: Verify that neighbor ATE CE2 in VRF ce2 maintains its active BGP session (/network-instances/network-instance[name=ce2]/protocols/protocol[identifier=openconfig-policy-types:BGP][name=BGP]/bgp/neighbors/neighbor[neighbor-address=192.0.2.6]/state/session-state = ESTABLISHED) and experiences 0% packet loss during CE1 drain.

• Step 1: Remove the soft drain policy on DUT for ATE CE1.
• Step 2: Verify that the BGP session seamlessly resumes normal route advertisement with default local-pref (100) and unprepended AS-Path without session flapping.
• Step 3: Measure convergence latency: verify routes 198.51.100.0/24 and 2001:db8:100::/64 are re-advertised with restored attributes to ATE RR1 in less than 5 seconds.

{
  "network-instances": {
    "network-instance": [
      {
        "name": "ce1",
        "config": {
          "name": "ce1",
          "type": "openconfig-network-instance-types:L3VRF",
          "route-distinguisher": "64500:100"
        },
        "protocols": {
          "protocol": [
            {
              "identifier": "openconfig-policy-types:BGP",
              "name": "BGP",
              "bgp": {
                "global": {
                  "config": {
                    "as": 64500
                  },
                  "route-selection-options": {
                    "config": {
                      "advertise-inactive-routes": true
                    }
                  }
                },
                "neighbors": {
                  "neighbor": [
                    {
                      "neighbor-address": "192.0.2.2",
                      "config": {
                        "neighbor-address": "192.0.2.2",
                        "peer-as": 65001
                      }
                    }
                  ]
                }
              }
            }
          ]
        }
      }
    ]
  }
}

** TODO: Update openconfig public data models to add l3vpn-ipv4-unicast and l3vpn-ipv6-unicast to the /network-instances/network-instance/protocols/protocol/bgp/rib/afi-safis/afi-safi container.

paths:
  # TODO: Update openconfig public data models to add l3vpn-ipv4-unicast and l3vpn-ipv6-unicast to the /network-instances/network-instance/protocols/protocol/bgp/rib/afi-safis/afi-safi container.
  # BGP Route Selection and Advertised Prefixes
  /network-instances/network-instance/protocols/protocol/bgp/global/route-selection-options/config/advertise-inactive-routes:
  /network-instances/network-instance/protocols/protocol/bgp/global/route-selection-options/state/advertise-inactive-routes:
  /network-instances/network-instance/protocols/protocol/bgp/neighbors/neighbor/afi-safis/afi-safi/state/prefixes/sent:
  # BGP RIB and Attributes
  /network-instances/network-instance/protocols/protocol/bgp/rib/afi-safis/afi-safi/ipv4-unicast/loc-rib/routes/route/state/prefix:
  /network-instances/network-instance/protocols/protocol/bgp/rib/afi-safis/afi-safi/ipv6-unicast/loc-rib/routes/route/state/prefix:
  /network-instances/network-instance/protocols/protocol/bgp/rib/afi-safis/afi-safi/ipv4-unicast/neighbors/neighbor/adj-rib-out-post/routes/route/state/prefix:
  /network-instances/network-instance/protocols/protocol/bgp/rib/afi-safis/afi-safi/ipv6-unicast/neighbors/neighbor/adj-rib-out-post/routes/route/state/prefix:
  /network-instances/network-instance/protocols/protocol/bgp/rib/attr-sets/attr-set/state/local-pref:
  # Network Instance Table Connections and Forwarding Table (AFT)
  /network-instances/network-instance/table-connections/table-connection/state/import-policy:
  /network-instances/network-instance/afts/ipv4-unicast/ipv4-entry/state/prefix:
  /network-instances/network-instance/afts/ipv6-unicast/ipv6-entry/state/prefix:
  # Session State
  /network-instances/network-instance/protocols/protocol/bgp/neighbors/neighbor/state/session-state:

rpcs:
  gnmi:
    gNMI.Get:
    gNMI.Set:
    gNMI.Subscribe:

• FFF