test(bdd): add DNS resolution assertion - #1337

TL;DR

Add a strict BDD assertion for waiting on host DNS resolution, replacing raw script command and exit-code pairs in the single- and multi-cluster EKS workflows.

The assertion preserves the existing system-resolver polling behavior, including three consecutive successful checks, while adding variable interpolation, input validation, and domain-specific failure reporting. The underlying DNS wait script now compares elapsed time instead of constructing an absolute deadline, preventing arithmetic overflow for the maximum timeout accepted by the DSL.

Additional Details

Issue #1086 asks the BDD DSL to express DNS resolution waits consistently with the existing EKS workflow.

This change:

• adds DNSResolutionCommand as a pure DSL command builder
• interpolates ${VAR} values and validates the hostname and timeout before execution
• uses dsl.BuildCommand to invoke the existing tests/bdd/scripts/wait-for-dns.sh
• registers a thin assertion handler that executes the command through Suite.Runner and requires exit code 0
• avoids exposing resolver stdout or stderr in assertion errors
• documents the step in tests/bdd/PLAN.md
• replaces five raw command/exit-code pairs across the single- and multi-cluster EKS features
• adds DSL unit tests and step-handler positive, validation-error, and execution-error tests
• changes the DNS wait script to compare elapsed time rather than adding the timeout to the current Unix timestamp
• extends the existing script execution test to cover both a normal timeout and the maximum accepted int64 timeout

No third-party dependencies were added or updated. There is no license or NOTICE impact. This is test infrastructure and is not customer-visible.

For the Reviewer

Please pay particular attention to:

• tests/bdd/dsl/dns.go for interpolation, validation, and command construction
• tests/bdd/steps/assertion_steps.go for the thin-handler boundary and failure reporting
• tests/bdd/fixtures_test.go for execution coverage of normal and maximum int64 timeouts
• tests/bdd/scripts/wait-for-dns.sh for overflow-safe elapsed-time comparison
• preservation of the existing host-resolver and three-consecutive-success behavior in wait-for-dns.sh
• the EKS feature replacements, which should only change how the existing DNS wait is expressed

For QA

Passed:

go test ./dsl -run DNS -v
go test ./steps -run DNS -v
bash -n scripts/wait-for-dns.sh
go test . -run '^TestWaitForDNSRequiresStableSystemResolution$' -count=1 -v
go test -short ./...
golangci-lint run --config .golangci.yml --new-from-rev=HEAD ./...
git diff --check

Not run:

TestSingleClusterEKSHelmfile
TestMultiClusterEKSHelmfile

Live EKS validation was not run locally because it requires pre-provisioned EKS clusters, NGC credentials, and creates AWS load-balancer resources. Maintainer-owned live EKS validation may be performed if required before merge.

The unrestricted lint command currently reports two pre-existing ST1005 findings in untouched dsl/manifests.go lines 51 and 54. Lint restricted to this change reports zero issues.

Issues

Fixes #1086

Checklist

• I am familiar with the Contributing Guidelines.
• I have signed off my commits for Developer Certificate of Origin (DCO) compliance.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.

Summary by CodeRabbit

• New Features

  • Added BDD support for verifying that DNS names resolve within a specified timeout.
  • Supports variable interpolation, input validation, normalized arguments, and clear failure reporting.
  • DNS checks require three consecutive successful resolver checks.

• Bug Fixes

  • Improved DNS timeout handling for more accurate elapsed-time tracking.

• Tests

  • Added coverage for valid, invalid, normalized, and timeout-based DNS resolution scenarios.
  • Updated single- and multi-cluster workflows to use the new DNS verification step.