• Tunnel End Point Resize test for Interface Based GRE Tunnels to ensure inflight increase or decrease in tunnel endpoints do not cause packet drops for encap function.

graph LR; 
A[OTG:Port1] --> B[Port1:DUT:Port2\lag1];
B --> C[Port2\lag1:OTG];
D[DUT:Port3\lag1] --> E[Port3\lag1:OTG];
f[DUT:Port4\lag2] --> G[Port4\lag2:OTG];
H[DUT:Port5\lag2] --> I[Port5\lag2:OTG];
J[DUT:Port6] --> K[Port6:OTG];

1. The test should configure 32 GRE tunnel interfaces per address family and each should be configured with different /32 and /128 tunnel destination prefixes of the same /27 and /64 prefix range. Each of these tunnel interfaces should use the loopback address as the source-address for their respective address-families. Hence, we will have total 64xtunnel interfaces, 32 per address-family as source and destination addresses of the tunnels.
2. Add static routes to provide reachability for the /27 IPv4 and /64 IPv6 tunnel destination prefixes in bullet#1 above to point at lag1, lag2 and port6 for ECMP.
3. Configure dualstack IPv4 and an IPv6 addresses for each of the 64xtunnel interfaces. The IP addresses can be of /30 and /127 prefix-lengths.
4. Pick 2 different /24 IPv4 and /64 IPv6 subnets each to emulate destination of payload traffic to be tunneled. Note: This is for the destination prefixes of the flows generated by OTG:Port1.
5. Add 128 different v4 and v6 static routes (64 for each address-family) for the 2 sets of /24 IPv4 and /64 IPv6 payload prefixes in bullet#4 above. The static routes should be such that, the first set of 1x /24 IPv4 and 1x /64 IPv6 prefixes should point at the 32x tunnel interfaces configured w/ IPv4 tunnel destination and the second set of 1x /24 IPv4 and 1x /64 IPv6 prefixes should point at the 32x tunnel interfaces w/ IPv6 tunnel destinations.

• Ensure all tunnel interfaces are up
• Ensure all LAG and singleton interfaces are up
• Run Health-1.1. checks

• Start with the configuration in Subtest-1 above.
• The test should send 2000 IPv4 and 2000 1Pv6 flows from the ATE:Port1 to the "tunnel payload destination prefixes". These flows should equally be split between the 2x /24 IPv4 and 2x /64 IPv6 destination prefixes mentioned in Subtest-1 bullet#4 above. The flows should use variations in the following,
  • Source IP address
  • Source Port
  • Destination IP adresses.
  • Destination Port
  • Payload protocol
• Incoming traffic on DUT:PORT1 should be load shared across all 64 Tunnel interfaces for encapsulation and then routed out the combination of lag and singleton interfaces, i.e. lag1, lag2 and Port6.
• Follow all the "verifications" steps in [*] below

• Start with the state in Subtest-2 above with active traffic flow.
• Reduce the number of Tunnel interfaces, e.g. from 32 to 16 for each of the tunnel destination group by address-family:
  • If the static routes are used to forward traffic to tunnel, please disable or delete the static routes in this test to simulate the reduction in available paths.
• Incoming traffic on DUT:PORT1 should now be load shared across all remaining 16 Tunnel interfaces per tunnel-destination by address-family.
• Follow all the verifications steps in [*] below.

• Start from the condition in Subtest-3 above with inflight traffic from OTG:Port1.
• Increase number of Tunnel interfaces, e.g. From 16 to 32 per tunnel-destination by address-family.
• Incoming traffic on DUT:PORT1 should now be load shared across all 32 Tunnel interfaces per tunnel-destination by address-family. Follow all the verifications steps in [*] below

• Start w/o any tunnel interface and corresponding static routes configuration
• Use a single gNMI.Replace operation to configure 32 tunnel interfaces and corresponding routes in a single request.
• Start traffic. Follow verification steps in [*] below to confirm success
• While the traffic flow is happening, make another gNMI.Replace operation for just 16 tunnels and corresponding static routes
• Follow the verification steps in [*] below for success.

• Verify the next hop counters for packets being diverted or sent for encapsulation. Test should fail if drops recorded.
• Verify the tunnel interfaces counters to confirm traffic encapsulation. Test should fail if the counts dont match the original count of packets sent by OTG
• Verify the tunnel interfaces for optimal load balancing of traffic/flow. Acceptable deviation is <=3%
• Verify the tunnel interfaces for optimal load balancing of traffic/flow. Acceptable deviation is <=3%. This is true for tunnel destinations of both address families.
• Validate system for:
  • Health-1.1. checks.

TODO: OpenConfig definition required for Tunnel protocol under interfaces/interfaces/interface/tunnel/ as GRE, IP-IP, GUE etc.

• interfaces/interfaces/interface/tunnel/config/dest/
• interfaces/interfaces/interface/tunnel/config/src/
• interfaces/interfaces/interface/tunnel/ipv4/addresses/address/config/ip
• interfaces/interfaces/interface/tunnel/ipv4/addresses/address/config/prefix-length
• interfaces/interfaces/interface/tunnel/ipv6/addresses/address/config/ip
• interfaces/interfaces/interface/tunnel/ipv6/addresses/address/config/prefix-length

• /interfaces/interface/tunnel/state/src
• /interfaces/interface/tunnel/state/dst
• /interfaces/interface/tunnel/ipv4/addresses/address/state/ip
• /interfaces/interface/tunnel/ipv4/addresses/address/state/prefix-length
• /interfaces/interface/tunnel/ipv6/addresses/address/state/ip
• /interfaces/interface/tunnel/ipv6/addresses/address/state/prefix-length
• Following stats for lag and singleton interfaces as well as for tunnel/ipv4/state/ and tunnel/ipv6/state/
  • state/counters/in-pkts
  • state/counters/in-octets
  • state/counters/out-pkts
  • state/counters/out-octets
  • state/counters/in-error-pkts
  • state/counters/in-forwarded-pkts
  • state/counters/in-forwarded-octets
  • state/counters/in-discarded-pkts
  • state/counters/out-error-pkts
  • state/counters/out-forwarded-pkts
  • state/counters/out-forwarded-octets
  • state/counters/out-discarded-pkts

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