Skip to content

RT-1.21: Update README for BGP TCP MSS and PMTUD - #6235

Open
isnehashankar43 wants to merge 24 commits into
openconfig:mainfrom
isnehashankar43:rt-1.21-readme-tcp-mss
Open

isnehashankar43 wants to merge 24 commits into
openconfig:mainfrom
isnehashankar43:rt-1.21-readme-tcp-mss

Conversation

@isnehashankar43

@isnehashankar43 isnehashankar43 commented Nov 3, 2026 •

Copy link
Copy Markdown
Contributor

This updates the RT-1.21 test plan (README) for BGP TCP MSS and PMTUD.

  • Define what transport/state/tcp-mss reports: the configured MSS (applied configuration); when no MSS is configured, the MSS in use may be reported instead.
  • Single DUT (TESTBED_DUT_ATE_2LINKS): eBGP IPv4/IPv6 on port-1 and iBGP IPv4/IPv6 on port-2.
  • RT-1.21.1: if the DUT reports an MSS before one is configured, checks that it is in the TCP default range. Then MSS at the neighbor level (records whether applying the MSS resets the sessions), then at the peer-group level, with the neighbors inheriting the value. Then a neighbor-level override: one neighbor per peer group is set to 1400, and the other neighbors keep the peer-group value.
  • RT-1.21.2: PMTUD at the neighbor and peer-group levels, then a neighbor-level override: PMTUD is disabled on one neighbor per peer group, and the other neighbors keep it enabled. The single-DUT testbed has no intermediate router, so the test verifies the PMTUD configuration and state only.
  • Coverage: tcp-mss and mtu-discovery config/state, neighbor and peer-group.
  • TODO: remove MD5 from the test code (covered by RT-1.1). The test code will be aligned in a follow-up PR.

* Define what transport/state/tcp-mss reports: the configured MSS, or
  optionally the MSS in use when no MSS is configured. Checks on the
  in-use value now apply only when a value is reported.
* RT-1.21.2: set the DUT1 port-2 MTU so that DUT1 port-1 is the smallest
  MTU on the path, and document the multihop reachability setup.
* Use RFC 5398 ASNs, add Canonical OC per subtest, pass/fail criteria
  and cleanup, and extend the path coverage.
@isnehashankar43 isnehashankar43 added the enhancement New feature or request label Nov 3, 2026
@isnehashankar43
isnehashankar43 requested a review from a team as a code owner November 3, 2026 11:59
@gemini-code-assist

Copy link
Copy Markdown
Contributor

Summary of Changes

Hello, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request updates the RT-1.21 test plan README to improve clarity and precision regarding BGP TCP MSS and PMTUD testing. The changes refine the test procedures, standardize ASNs, and introduce structured pass/fail criteria and Canonical OC configurations to ensure consistent test execution and validation.

Highlights

  • Documentation Clarification: Clarified the behavior of transport/state/tcp-mss, noting that it reports the configured MSS or the in-use MSS if none is configured.
  • Test Procedure Updates: Updated RT-1.21.2 (PMTUD) to include specific MTU settings for DUT1 port-2 and documented the required IS-IS and static route configuration for multihop iBGP.
  • Standardization: Adopted RFC 5398 documentation ASNs (64500 and 64501) and added detailed Canonical OC blocks with pass/fail criteria for each subtest.
  • Coverage Expansion: Added coverage for auth-password, interface MTU, and neighbor session-state, while including a TODO for a future leaf to report negotiated MSS.
Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature Command Description
Code Review /gemini review Performs a code review for the current pull request in its current state.
Pull Request Summary /gemini summary Provides a summary of the current pull request in its current state.
Comment @gemini-code-assist Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help /gemini help Displays a list of available commands.

Customization

To customize the Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counterproductive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here.

Footnotes

  1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

@OpenConfigBot

OpenConfigBot commented Nov 3, 2026 •

Copy link
Copy Markdown

Pull Request Functional Test Report for #6235 / 1853eff

Virtual Devices

Device Test Status Test Documentation Job Raw Log
Arista cEOS status
RT-1.22: failure
RT-1.22: BGP TCP MSS
4a045ebe Log
Cisco 8000E status
RT-1.22: failure
RT-1.22: BGP TCP MSS
1f0c5a06 Log
Cisco XRd status
RT-1.22: failure
RT-1.22: BGP TCP MSS
f2620cee Log
Juniper ncPTX status
RT-1.22: failure
RT-1.22: BGP TCP MSS
f5c888b6 Log
Nokia SR Linux status
RT-1.22: failure
RT-1.22: BGP TCP MSS
a75aa96f Log
Openconfig Lemming status
RT-1.22: failure
RT-1.22: BGP TCP MSS
e1f598df Log

Hardware Devices

Device Test Status Test Documentation Raw Log
Arista status
RT-1.22: pending authorization
RT-1.22: BGP TCP MSS
Cisco status
RT-1.22: pending authorization
RT-1.22: BGP TCP MSS
Juniper status
RT-1.22: pending authorization
RT-1.22: BGP TCP MSS
Nokia status
RT-1.22: pending authorization
RT-1.22: BGP TCP MSS

Help

@gemini-code-assist gemini-code-assist Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Code Review

This pull request updates the README.md for the BGP TCP MSS and PMTUD test (RT-1.21). It expands the test plan with a detailed topology, DUT/OTG properties, step-by-step procedures for eBGP TCP MSS and multihop iBGP PMTUD testing, canonical OpenConfig configurations, pass/fail criteria, and updated path/RPC coverage. There are no review comments, and the changes are appropriate, so I have no feedback to provide.

* Summary: state/tcp-mss reports the applied configuration, in the same
  way as timers/state/hold-time.
* RT-1.21.1: an MSS reported without configuration must be between the
  TCP default MSS (536 for IPv4, 1220 for IPv6) and the MTU minus 40/60.
* RT-1.21.2: the PMTUD check now uses 536 to 1460.
* Remove the IPv6 link-local and in-use MSS leaf TODOs.
* List only the tcp-mss and mtu-discovery paths in the coverage section.
* Remove the platform restriction (N/A).
* eBGP IPv4/IPv6 on ATE port-1 and iBGP IPv4 with TCP MD5 on ATE
  port-2, both to the same DUT. Remove DUT2, IS-IS and the static route.
* RT-1.21.1: state/tcp-mss must equal config/tcp-mss.
* RT-1.21.2: the MSS in use must fit the 1500-byte MTU of ATE port-2.
@pmprado

pmprado commented Nov 5, 2026

Copy link
Copy Markdown

/fptest virtual

Sneha Shankar added 2 commits November 6, 2026 15:13
@pmprado

pmprado commented Nov 6, 2026

Copy link
Copy Markdown

/gemini review

@gemini-code-assist gemini-code-assist Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Code Review

This pull request updates the README.md for the BGP TCP MSS and PMTUD test (RT-1.21) to provide a comprehensive test plan, including topology, detailed test procedures for neighbor and peer-group levels, canonical OpenConfig JSON configurations, and updated path coverage. The review feedback recommends adding a TODO to propose an OpenConfig leaf for reporting the TCP MSS in use on a session, and adding missing configuration and telemetry paths (such as interface MTU, session-state, and established-transitions) to the OpenConfig Path and RPC Coverage section.

Sneha Shankar added 2 commits November 6, 2026 19:49
* RT-1.21.1: MSS at the neighbor level.
* RT-1.21.2: MSS at the peer-group level, inherited by the neighbors.
* RT-1.21.3: PMTUD at the neighbor and peer-group levels, with no MSS
  configured.
* RT-1.21.1: MSS at the neighbor level, then at the peer-group level.
* RT-1.21.2: PMTUD at the neighbor level, then at the peer-group level.
With the MSS (RT-1.21.1) or PMTUD (RT-1.21.2) set on the peer groups, override it on one neighbor of each peer group and verify that only that neighbor changes, while the other neighbors keep the peer-group value.

@dplore dplore left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks. I see the changes for TCP MSS which will test out MSS negotiation, but I don't see a step where the MTU of the physical port is smaller than the TCP MSS, forcing path MTU discovery to be performed.

@pmprado pmprado left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@isnehashankar43 - in order to make the PMTUD test actually test the feature, @dplore suggested a topology that I've tweaked:

               [ ATE / Packet Generator ]
               +------------------------+
               |  Port 1        Port 2  |
               +----+--------------+----+
    10.1.1.2/30     |              |      10.3.3.2/30
    MTU: 1500B      |              |      MTU: 9216B
                    |              |
               +----+--------------+----+
    MTU: 1500B | Port 1          Port 4 |
               | [VRF-WEST]   [VRF-EAST]|
               |    \            /      |
    MTU: 9216B |    Port 2     Port 3   | MTU: 9216B
               +------+-----------+-----+
                      |           |
                      +-----------+
                    Direct Loop Cable
                      10.2.2.0/30
                       MTU: 9216B

Here, we'd enable a BGP session with PMTUD between a loopback of VRF-EAST and ATE Port1, which sits behind the strangled MTU.

We'd need to inject some big BGP updates in order to exceed the 1500 bytes of MTU and trigger PMTUD.

Sorry for the extra modification... Could you please make this ANOTHER change?

Sneha Shankar added 2 commits November 11, 2026 10:16
Moves the MSS test plan to a new test, RT-1.22, in feature/bgp/otg_tests/bgp_tcp_mss. RT-1.21 (bgp_tcp_mss_path_mtu) is left unchanged.
Goes back to one RT-1.21 test for both MSS and PMTUD, and removes the RT-1.22 folder. The test uses the TESTBED_DUT_1_LOOP_2_ATE testbed: a loop cable joins the DEFAULT network instance, where BGP runs, to a TRANSIT network instance, which is a router between the DUT BGP speaker and ATE port-1. RT-1.21.2 lowers the MTU of that router's link to ATE port-1 and verifies that large BGP UPDATEs still arrive.
@isnehashankar43

Copy link
Copy Markdown
Contributor Author

Thanks. I see the changes for TCP MSS which will test out MSS negotiation, but I don't see a step where the MTU of the physical port is smaller than the TCP MSS, forcing path MTU discovery to be performed.

@pmprado @dplore The README now tests PMTUD through a loop on one DUT, as you proposed, on a new testbed TESTBED_DUT_1_LOOP_2_ATE (dut:port1-ate:port1, dut:port2-dut:port3, dut:port4-ate:port2). The loop link can be a cable or an OCS cross-connect.

Three things differ from the sketch:

  1. The small MTU is 1400 on DUT port-1, and every other port keeps the default 1500. TCP never sends more than the MSS its peer advertises, and an ATE port at 1500 advertises 1460, so a 1500-byte link is never exceeded. With 1400, the first full-size packet is too big whatever MSS the ATE advertises.
  2. BGP stays in the DEFAULT network instance (VRF-EAST in the sketch). Only the router in the middle is a VRF (TRANSIT, VRF-WEST in the sketch). RT-1.21.1 and the Canonical OC stay as reviewed.
  3. The sessions through the loop are iBGP and use the DUT port-3 address, so they need no loopback interface and no eBGP multihop. The directly connected sessions on ATE port-2 are eBGP and advertise the prefixes that make the large UPDATEs.

The PMTUD step passes if all prefixes reach ATE port-1, the sessions stay ESTABLISHED, and a capture on ATE port-1 shows DUT packets of more than 1280 bytes and no IPv4 fragments.

The testbed definition (topology file and enum) is in #6274.

@dplore

dplore commented Nov 12, 2026

Copy link
Copy Markdown
Contributor

Is this expected to fail on Cisco? I see from the test results it's failing there?

Also, I think it is better to include the topology change in this PR too, otherwise fptest isn't going to be using that topo?

Adds RT-1.22 in a new folder, feature/bgp/otg_tests/bgp_tcp_mss: the TCP MSS on a single DUT (TESTBED_DUT_ATE_2LINKS), at the neighbor and peer-group levels, with the neighbor-level override. RT-1.21 (bgp_tcp_mss_path_mtu) is restored to its version on main and still covers PMTUD.

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants