- Revert namespace from dispensary-scraper to cannaiq
- Keep registry.spdy.io for image URLs (k8s nodes need HTTPS)
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Update kubectl set image commands to use HTTPS registry URL
- Fix namespace from cannaiq to dispensary-scraper
- Add guidance on when to use which registry URL
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Changed heredoc from 'KUBEEOF' (quoted) to KUBEEOF (unquoted)
- This allows shell variable expansion of $K8S_TOKEN directly
- Removed sed replacement step that was failing due to YAML escaping issues
Woodpecker doesn't allow custom clone or volumes without elevated trust.
Kaniko layer caching (--cache-repo) still works (registry-based).
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Only fetch latest commit instead of full history.
Reduces checkout time and bandwidth.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- PR steps: shared npm-cache volume for faster npm ci
- Docker builds: --cache-repo to local registry for layer caching
- Kaniko will reuse npm install layer when package.json unchanged
First build populates cache, subsequent builds much faster.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Remove separate build steps (didn't save time)
- Use original multi-stage Dockerfiles
- Delete unused Dockerfile.ci files
- 4 parallel docker builds + deploy
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Alpine uses musl libc which breaks Rollup's native bindings.
Debian-based node:22 uses glibc and works correctly.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add Dockerfile.ci for backend, cannaiq, findadispo, findagram
- Frontend Dockerfiles just copy pre-built assets to nginx
- Backend Dockerfile copies pre-built dist/node_modules
- Reduces Docker build time by doing npm ci/build in CI step
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Build images push to 10.100.9.70:5000/cannaiq/*
- Deploy pulls from local registry (no external network)
- Removed git.spdy.io registry auth (not needed for local)
- Added --insecure-registry for HTTP local registry
BuildKit container driver has sysctl permission issues in LXC.
Using plugins/docker instead of woodpeckerci/plugin-docker-buildx.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Remove cache_from/cache_to from CI (plugin bug splitting commas)
- Add preflight() method to CrawlRotator - tests proxy + anti-detect
- Add pre-task preflight check - workers MUST pass before executing
- Add releaseTask() to release tasks back to pending on preflight fail
- Rename proxy_test task to whoami for clarity
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Plugin was splitting comma-separated values incorrectly.
Use array format with quoted strings instead.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Add cache_from and cache_to settings to all docker-buildx steps.
Uses registry-based caching to avoid rebuilding npm install layer
when package.json hasn't changed.
Expected improvement: 14min backend build → ~3-4min on cache hit.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
If workers are scaled to 0, CI will now automatically scale them to 5
before updating the image. This prevents workers from being stuck at 0
if manually scaled down for maintenance.
The check only scales up if replicas=0, so it won't interfere with
normal deployments or HPA scaling.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Auto-migrate runs at server startup and handles migration errors gracefully.
The explicit kubectl exec migration step was failing due to trigger
already existing (schema_migrations table out of sync).
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Removes the migrate step that required db_* secrets (which CI can't
access since postgres is cluster-internal). Instead, run migrations
via kubectl exec on the deployed scraper pod, which already has DB
access via its env vars.
Deploy order:
1. Deploy scraper image
2. Wait for rollout
3. Run migrations via kubectl exec
4. Deploy remaining services
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
The [push, manual] array syntax broke CI config parsing.
Reverting to event: push which is known to work.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Allow deploy steps to run on both push and manual events.
This enables triggering deploys via `woodpecker-cli pipeline create`.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
