workflow_run triggers only fire for push events, not pull_request events,
causing PRs to skip integration and E2E tests entirely. Add dual triggers
to all test workflows so they run for both push (via workflow_run) and
pull_request events, while maintaining single-build architecture.
All workflows still pull pre-built images from docker-build.yml - no
redundant builds introduced. This fixes PR test coverage while preserving
the "Build Once, Test Many" optimization for push events.
Fixes: Build Once architecture (commit 928033ec)
Restructures CI/CD pipeline to eliminate redundant Docker image builds
across parallel test workflows. Previously, every PR triggered 5 separate
builds of identical images, consuming compute resources unnecessarily and
contributing to registry storage bloat.
Registry storage was growing at 20GB/week due to unmanaged transient tags
from multiple parallel builds. While automated cleanup exists, preventing
the creation of redundant images is more efficient than cleaning them up.
Changes CI/CD orchestration so docker-build.yml is the single source of
truth for all Docker images. Integration tests (CrowdSec, Cerberus, WAF,
Rate Limiting) and E2E tests now wait for the build to complete via
workflow_run triggers, then pull the pre-built image from GHCR.
PR and feature branch images receive immutable tags that include commit
SHA (pr-123-abc1234, feature-dns-provider-def5678) to prevent race
conditions when branches are updated during test execution. Tag
sanitization handles special characters, slashes, and name length limits
to ensure Docker compatibility.
Adds retry logic for registry operations to handle transient GHCR
failures, with dual-source fallback to artifact downloads when registry
pulls fail. Preserves all existing functionality and backward
compatibility while reducing parallel build count from 5× to 1×.
Security scanning now covers all PR images (previously skipped),
blocking merges on CRITICAL/HIGH vulnerabilities. Concurrency groups
prevent stale test runs from consuming resources when PRs are updated
mid-execution.
Expected impact: 80% reduction in compute resources, 4× faster
total CI time (120min → 30min), prevention of uncontrolled registry
storage growth, and 100% consistency guarantee (all tests validate
the exact same image that would be deployed).
Closes #[issue-number-if-exists]
- Print cache contents and Playwright CLI version for diagnostics
- Search for expected browser executables and force reinstall with --force if absent
- Add headless-launch verification via Node to fail fast with clear logs
- Add step to delete ~/.cache/ms-playwright before installing browsers
- Guarantees correct browser version for each run
- Prevents mismatched or missing browser binaries (chromium_headless_shell-1208, etc.)
- Should resolve browser not found errors for all browsers
- Removed restore-keys fallback from Playwright cache
- Only exact cache matches (same package-lock.json hash) are used
- This prevents restoring incompatible browser versions when Playwright updates
- Added cache-hit check to skip install when cache is valid
- Firefox and WebKit were failing because old cache was restored but browsers were incompatible
Job names now show: 'E2E chromium (Shard 1/4)' instead of 'E2E Tests (Shard 1/4)'
Makes it easier to identify which browser/shard is passing or failing
- Artifact names now include browser: playwright-report-{browser}-shard-{N}
- Docker logs include browser: docker-logs-{browser}-shard-{N}
- Install step always runs (idempotent) to ensure version match
- Fixed artifact name conflicts when 3 browsers share same shard number
- Updated summary and PR comment to reflect new naming
- Introduced a new wrapper function for query client to facilitate testing.
- Added comprehensive tests for upload, commit, and cancel operations.
- Improved error handling in tests to capture and assert error states.
- Enhanced session management and state reset functionality in tests.
- Implemented polling behavior tests for import status and preview queries.
- Ensured that upload previews are prioritized over status query previews.
- Validated cache invalidation and state management after commit and cancel actions.
- Updated Docker Compose files to use digest-pinned images for CI contexts.
- Enhanced Dockerfile to pin Go tool installations and verify external downloads with SHA256 checksums.
- Added Renovate configuration for tracking Go tool versions and digest updates.
- Introduced a new design document outlining the architecture and data flow for dependency tracking.
- Created tasks and requirements documentation to ensure compliance with the new digest pinning policy.
- Updated security documentation to reflect the new digest pinning policy and exceptions.
Add missing emergency token environment variable to all E2E test workflows to
fix security teardown failures in CI. Without this token, the emergency reset
endpoint returns 501 "not configured", causing test teardown to fail and
leaving ACL enabled, which blocks 83 subsequent tests.
Changes:
Add CHARON_EMERGENCY_TOKEN to docker-build.yml test-image job
Add CHARON_EMERGENCY_TOKEN to e2e-tests.yml e2e-tests job
Add CHARON_EMERGENCY_TOKEN to playwright.yml playwright job
Verified:
Docker build strategy already optimal (build once, push to both GHCR + Docker Hub)
Testing strategy correct (test once by digest, validates both registries)
All workflows now have environment parity with local development setup
Requires GitHub repository secret:
Name: CHARON_EMERGENCY_TOKEN
Value: 64-char hex token (e.g., from openssl rand -hex 32)
Related:
Emergency endpoint rate limiting removal (proper fix)
Local emergency token configuration (.env, docker-compose.local.yml)
Security test suite teardown mechanism
Refs #550
The E2E workflow was failing during backend build because make build
was being executed from the backend/ directory, but the Makefile exists
at the root level.
Remove working-directory: backend from Build backend step
Allows make build to execute from root where Makefile is located
Verified with local test: frontend + backend build successfully
Related to PR #550 E2E workflow failures
The E2E workflow was failing during "Build frontend" because npm ci
was only run at root level. The frontend directory has its own
package.json with React, Tailwind, and other dependencies that were
never installed.
Add "Install frontend dependencies" step before build
Update Node.js version from 18 to 20 (required by markdownlint-cli2)
Fixes failing E2E tests in PR #550
