docs: add migration status and recovery guide

Comprehensive guide for resuming work after Pi restart:
- Complete migration history
- Current issues (Vite segfault on ARM)
- Next steps and recovery commands
- Alternative solutions if build fails

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>

Author: pxs4528

.cicd/MIGRATION_STATUS.md

@@ -0,0 +1,271 @@
+# Migration Status & Next Steps
+
+**Date:** 2026-01-05
+**Status:** In Progress - Need to complete Vite build on ARM
+
+---
+
+## What We've Done
+
+### ✅ Completed Tasks
+
+1. **Migrated from Podman to Docker**
+   - Podman v5.4.2 has critical segfaults on ARM64 (Raspberry Pi)
+   - Installed Docker 29.1.3 successfully
+   - Created Dockerfiles from Containerfiles
+   - Updated compose.yaml to reference Dockerfiles
+   - Updated deploy.sh to use `docker compose` instead of `podman-compose`
+
+2. **Fixed Frontend JavaScript Errors**
+   - Removed `import "solid-devtools";` from `frontend/src/index.tsx`
+   - This was causing: `can't access property "live", n.links is undefined`
+   - Removed devtools plugin from `vite.config.ts`
+
+3. **Optimized Vite Build for ARM**
+   - Set `NODE_OPTIONS="--max-old-space-size=1536"` in frontend Dockerfile
+   - Configured esbuild minifier (faster, less memory)
+   - Disabled manual chunks to reduce memory fragmentation
+   - All committed to GitHub (commit: b807285)
+
+4. **Documentation**
+   - Created `.cicd/DOCKER_MIGRATION.md` (full migration guide)
+   - Created `.cicd/cleanup-podman.sh` (cleanup script)
+
+---
+
+## Current Issues
+
+### 🔴 CRITICAL: Vite Build Segfault
+
+**Symptom:**
+```
+vite v7.3.0 building client environment for production...
+transforming...
+Segmentation fault (core dumped)
+```
+
+**Root Cause:** Vite build is very memory-intensive on ARM64, crashes during build
+
+**System Resources:**
+- RAM: 3.7GB total, only 1.5GB available
+- Swap: 1GB total, 654MB available
+- Vite needs ~2GB+ to build safely on ARM
+
+**Attempted Fixes:**
+- ✅ Set NODE_OPTIONS max-old-space-size=1536
+- ✅ Optimized vite.config.ts
+- ⏳ Need to test after Pi restart (user is restarting now)
+
+### ⚠️ Webhook Permission Issues
+
+**Symptom:**
+```
+permission denied while trying to connect to the Docker daemon socket
+```
+
+**Root Cause:** Webhook listener runs as user `darth`, but docker group membership hasn't taken effect (process started before group was added)
+
+**Fix Required:** Restart webhook listener after Pi reboot
+
+---
+
+## Next Steps (After Pi Restart)
+
+### Step 1: Test Optimized Vite Build
+
+```bash
+cd ~/darth-forge
+
+# Pull latest changes with memory optimizations
+git pull
+
+# Try building with new memory settings
+docker compose -f compose.yaml -f compose.prod.yaml down
+docker compose -f compose.yaml -f compose.prod.yaml up -d --build
+
+# Monitor build logs
+docker compose -f compose.yaml -f compose.prod.yaml logs -f frontend
+```
+
+**If build succeeds:** ✅ Done! Check site at http://darth.local:8080
+
+**If build still segfaults:** Go to Step 2
+
+### Step 2: Build on GitHub Actions (Recommended)
+
+If local ARM build fails, build images on GitHub Actions (x86 runners have more RAM) and push to GitHub Container Registry (ghcr.io).
+
+**Changes needed:**
+1. Create `.github/workflows/build-images.yml` - Build & push images to ghcr.io
+2. Update `compose.yaml` - Pull pre-built images instead of building locally
+3. Update `deploy.sh` - Just pull and restart (no build step)
+
+This is what production systems do - build on powerful CI servers, deploy lightweight containers.
+
+### Step 3: Fix Webhook for Future Deployments
+
+After Pi restart, the webhook listener will have docker group permissions:
+
+```bash
+# Verify webhook can access docker
+sudo systemctl status webhook
+ps aux | grep webhook-listener
+
+# Test manual trigger
+curl http://localhost:9000/health
+
+# Check logs
+tail -f /home/darth/darth-forge/.cicd/webhook.log
+```
+
+---
+
+## Quick Recovery Commands
+
+### Check Current Status
+```bash
+# Check if Docker is running
+docker ps
+
+# Check containers
+docker ps -a
+
+# Check system resources
+free -h
+
+# Check webhook status
+sudo systemctl status webhook
+```
+
+### Manual Deployment
+```bash
+cd ~/darth-forge
+git pull
+docker compose -f compose.yaml -f compose.prod.yaml down
+docker compose -f compose.yaml -f compose.prod.yaml up -d --build
+```
+
+### View Logs
+```bash
+# Frontend logs
+docker logs darth-forge-frontend --tail 50 -f
+
+# Backend logs
+docker logs darth-forge-backend --tail 50 -f
+
+# Deployment logs
+tail -f .cicd/deploy.log
+
+# Webhook logs
+tail -f .cicd/webhook.log
+```
+
+### Test Endpoints
+```bash
+# Backend health check
+curl http://localhost:8080/api/health
+# Should return: {"status":"healthy"}
+
+# Frontend
+curl -I http://localhost:8080
+# Should return: HTTP/1.1 200 OK
+
+# Open in browser
+# http://darth.local:8080
+```
+
+---
+
+## Architecture Overview
+
+```
+GitHub Push
+     ↓
+GitHub Actions (deploy.yml)
+     ↓
+Webhook → webhook.pipboi.dev
+     ↓
+Cloudflare Tunnel (cloudflared)
+     ↓
+Raspberry Pi → webhook-listener.py (port 9000)
+     ↓
+deploy.sh
+     ↓
+Docker Compose
+     ↓
+┌─────────────────┐     ┌──────────────────┐
+│ Frontend        │     │ Backend          │
+│ (Caddy + React) │────▶│ (Go API)         │
+│ Port: 8080      │     │ Port: 8080       │
+└─────────────────┘     └──────────────────┘
+```
+
+**Key Files:**
+- `.cicd/deploy.sh` - Main deployment script
+- `compose.yaml` - Base Docker Compose config
+- `compose.prod.yaml` - Production overrides
+- `.cicd/webhook-listener.py` - GitHub webhook receiver
+- `frontend/Dockerfile` - Frontend build instructions
+- `backend/Dockerfile` - Backend build instructions
+- `frontend/Caddyfile` - Caddy reverse proxy config
+
+---
+
+## Known Working Commits
+
+- `b807285` - Latest (Vite memory optimizations)
+- `1ece911` - Removed solid-devtools from production
+- `80a1cbf` - Docker migration complete
+- `56e60a2` - Last podman commit (broken)
+
+---
+
+## Important Notes
+
+1. **Don't use podman anymore** - It crashes on ARM64
+2. **Frontend needs to build** - Currently segfaulting during Vite build
+3. **Backend is working** - Builds fine, returns `{"status":"healthy"}`
+4. **Caddy routing is correct** - `/api/*` → backend:8080, everything else → frontend
+5. **Cloudflare Tunnel is running** - PID 1690, routes traffic correctly
+
+---
+
+## If All Else Fails: Alternative Solutions
+
+### Option 1: Pre-build Frontend Locally (Outside Docker)
+Build on Pi directly, then copy dist to container:
+```bash
+cd frontend
+npm install
+npm run build
+# Copy dist to Docker image
+```
+
+### Option 2: Use Multi-Arch Images from CI
+Let GitHub Actions build for arm64/aarch64, push to registry, pull on Pi
+
+### Option 3: Reduce Frontend Complexity
+Simplify the build or use a simpler framework that doesn't need Vite
+
+### Option 4: Add More Swap
+Increase swap file size to 2GB+ to give Vite more virtual memory:
+```bash
+sudo swapoff -a
+sudo dd if=/dev/zero of=/swapfile bs=1M count=2048
+sudo chmod 600 /swapfile
+sudo mkswap /swapfile
+sudo swapon /swapfile
+```
+
+---
+
+## Contact Info
+
+GitHub Repo: https://github.com/pxs4528/darth-forge
+Issues: https://github.com/pxs4528/darth-forge/issues
+Website: https://pipboi.dev (via Cloudflare Tunnel)
+
+---
+
+**Last Updated:** 2026-01-05 23:15 CST
+**Next Action:** Test optimized build after Pi restart