Skip to content

Stand up a test environment that mirrors how the edge really runs#226

Closed
scotwells wants to merge 2 commits into
mainfrom
test/prod-fidelity-test-environment
Closed

Stand up a test environment that mirrors how the edge really runs#226
scotwells wants to merge 2 commits into
mainfrom
test/prod-fidelity-test-environment

Conversation

@scotwells

@scotwells scotwells commented Jun 25, 2026

Copy link
Copy Markdown
Contributor

What this stands up

Our edge tests have run against a simplified setup that didn't match how the edge actually runs in production. Whole classes of problems — the ones that only appear with the real gateway, the real firewall data plane, and the real path the platform uses to hand configuration to the edge — could never show up in a test.

This brings up a test environment that mirrors the real edge: two clusters wired the way production is, the same gateway version we run in production, the firewall data plane customer traffic actually passes through, and the same delivery path configuration really travels. Tests can now exercise the route real traffic takes, instead of a stand-in for it.

Why it matters

A test is only as trustworthy as how closely it resembles the real thing. Every gap between the test setup and production is a place a bug can hide unseen. Closing that gap is what makes the end-to-end tests built on top of this worth trusting.

Scope

This is test scaffolding — it changes nothing about how the edge runs for customers. It only gives us a faithful place to prove the edge behaves before changes reach them.


Part of #247 — catching silent edge misconfiguration before it reaches customers.

One of a five-PR stack, best reviewed and merged in order:

Brings up a two-cluster environment that mirrors how the edge really runs:
the production gateway version, the firewall data plane real traffic passes
through, and the actual path configuration travels to the edge. Lets tests
exercise the route customer traffic takes instead of a simplified stand-in.
Test scaffolding only; no change to how the edge runs for customers.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JbCy8vy66RdNYzGSgqH6P6
@scotwells
scotwells force-pushed the test/prod-fidelity-test-environment branch 2 times, most recently from a29d679 to 4b21276 Compare June 25, 2026 21:43
The endpoint that lets the parity test confirm the proxy is running exactly the
set the build intended is off by default and served only when asked for. The
base deployment reads --enable-programmed-set from the ENABLE_PROGRAMMED_SET env
var; flip it to "true" in the downstream edge overlay with a strategic-merge
patch on env so the test environment keeps that visibility while production
exposes nothing.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JbCy8vy66RdNYzGSgqH6P6
@scotwells
scotwells force-pushed the test/prod-fidelity-test-environment branch from 4b21276 to d2e62bf Compare June 25, 2026 21:48
ecv added a commit that referenced this pull request Jul 14, 2026
Add a test-infra:test-e2e task that runs the control-plane e2e suite
against the live #226 two-cluster environment, mapping the suite's
chainsaw aliases (nso-standard/nso-infra) onto this env's upstream and
downstream kubeconfigs at invocation. Bring the hardened suite from main
onto this branch and pin the operator image to the env's build.

Key changes:
- add test-infra:test-e2e task; CLI_ARGS selects a single scenario
- pin manager and e2e-downstream image tags to d2e62bf
- assert the ClusterIssuer CA Certificate is Ready before gateway steps
- carry the listener-hostname-collision, tpp-enforce, and
  tpp-paranoia-validation scenarios from main
@ecv

ecv commented Jul 14, 2026

Copy link
Copy Markdown
Contributor

Superseded by #276 — this env rebased onto main plus the e2e suite (#275) lifted onto it, as a single PR for this slot in the #247 stack. The suite runs 8/8 green against the live two-cluster env there. Leaving this open for you to close/redirect as you see fit, @scotwells.

ecv added a commit that referenced this pull request Jul 14, 2026
Add a test-infra:test-e2e task that runs the control-plane e2e suite
against the live #226 two-cluster environment, mapping the suite's
chainsaw aliases (nso-standard/nso-infra) onto this env's upstream and
downstream kubeconfigs at invocation. Bring the hardened suite from main
onto this branch and pin the operator image to the env's build.

Key changes:
- add test-infra:test-e2e task; CLI_ARGS selects a single scenario
- pin manager and e2e-downstream image tags to d2e62bf
- assert the ClusterIssuer CA Certificate is Ready before gateway steps
- carry the listener-hostname-collision, tpp-enforce, and
  tpp-paranoia-validation scenarios from main
@ecv

ecv commented Jul 14, 2026

Copy link
Copy Markdown
Contributor

@scotwells permission to close in favor of #276? – also, lets start mergin the stack

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants