[hillol-nexthop]

Pre-submission checklist

• I've ran the linters locally and fixed lint errors related to the files I modified in this PR. You can install the linters by running pip install -r requirements-dev.txt && pre-commit install
• pre-commit run

Summary

config interface <name> switchport access vlan <id> set sw.ports[].ingressVlan and retargeted existing sw.vlanPorts[] entries, but never inserted one when the port had no membership — leaving a config where the port is not a member of its ingress VLAN (rejected by the agent config validator). A port with multiple entries also got all of them retargeted, producing duplicate (vlan, port) pairs.

The command now leaves each port with exactly one untagged vlanPorts membership in the target VLAN:

• stale untagged (access) memberships are removed,
• an existing entry in the target VLAN is deduplicated,
• the new untagged entry is inserted,
• tagged (trunk) memberships in other VLANs are preserved.

Integration test redesign

ConfigVlanSwitchportAccessTest is redesigned as NewVlanAcrossPorts: attach a brand-new VLAN to one port (auto-create + insert path end to end), then the now-existing VLAN to a second port; after each warmboot commit verify ingressVlan and the untagged vlanPorts membership via the agent's running config.

Teardown

No manual unwind: the test harness's per-suite config containment (#1557) records the /etc/coop baseline before the suite starts and rolls the DUT back to it after the suite ends, so the device is left as found even when the test fails mid-way.

Port selection

NewVlanAcrossPorts needs two distinct ports; it calls the existing getRandomInterfacePortName() twice (retrying until the names differ) — no new fixture helper.

Test Plan

• UT //fboss/cli/fboss2/test/config:cmd_config_test — 17/17 pass, including new UTs: insertion when membership missing, dedup of conflicting entries, idempotency, trunk-membership preservation, membership on VLAN auto-create.
• On a freshly prepared lab device (NH-4010-F), ConfigVlanSwitchportAccessTest.NewVlanAcrossPorts — PASS (binary built from our internal tree carrying the identical test body). Both VLAN commits were recorded as AGENT_WARMBOOT in cli/cli_metadata.json, agents active after each commit, and the config was rolled back to the pre-suite baseline afterwards with no test residue in cli/agent.conf.

[meta-codesync]

This pull request has been imported. If you are a Meta employee, you can view this in D114555610. (Because this pull request was imported automatically, there will not be any future comments.)

[facebook-github-tools]

@hillol-nexthop has updated the pull request. You must reimport the pull request before landing.

[facebook-github-tools]

@hillol-nexthop has updated the pull request. You must reimport the pull request before landing.

[facebook-github-tools]

@hillol-nexthop has updated the pull request. You must reimport the pull request before landing.

[facebook-github-tools]

@hillol-nexthop has updated the pull request. You must reimport the pull request before landing.

[facebook-github-tools]

@hillol-nexthop has updated the pull request. You must reimport the pull request before landing.

[facebook-github-tools]

@hillol-nexthop has updated the pull request. You must reimport the pull request before landing.

[hillol-nexthop]

Note on ConfigCoppCpuTrafficPolicyTest.cpp: the file ships here (byte-identical to our internal tree) but is deliberately not added to the integration-test build targets — it drives config copp cpu-traffic-policy match ... action ..., which is pending in #1379. Registering it in cmake/CliFboss2TestIntegrationTest.cmake and fboss/cli/fboss2/test/integration_test/BUCK is a one-line follow-up once #1379 lands. Verified on a lab device: with the command absent the test fails at runtime (arg-parse error), and the other suites in this PR — ConfigVlanCreateTest (6/6, including the auto-create and switchport tests), ConfigVlanSwitchportAccessTest.NewVlanAcrossPorts, and ConfigPfcTest — all pass on this branch (8/8).

[facebook-github-tools]

@hillol-nexthop has updated the pull request. You must reimport the pull request before landing.

[joseph5wu]

I strongly recommend to move the new snapshop test config for integration test out of this PR and have a new one to take care of it.
Besides, I don't think we should make Fboss2IntegrationTest to introduce the snapshop/restore config feature. Instead, we should consider to use ConfigSession::rollback(). I would also like to see whether rollback can take care of this case as expected(whether it needs coldboot or warmboot)

cc: @benoit-nexthop

[facebook-github-tools]

@hillol-nexthop has updated the pull request. You must reimport the pull request before landing.

[facebook-github-tools]

@hillol-nexthop has updated the pull request. You must reimport the pull request before landing.

[facebook-github-tools]

@hillol-nexthop has updated the pull request. You must reimport the pull request before landing.

[facebook-github-tools]

@hillol-nexthop has updated the pull request. You must reimport the pull request before landing.

[facebook-github-tools]

@hillol-nexthop has updated the pull request. You must reimport the pull request before landing.

[facebook-github-tools]

@hillol-nexthop has updated the pull request. You must reimport the pull request before landing.

[facebook-github-tools]

@hillol-nexthop has updated the pull request. You must reimport the pull request before landing.