Stabilize production deployment (wallet auto-recovery, monitoring)

[andreabadesso]

Current state

The x402 POC is deployed at *.x402.hathor.dev on Dokploy, connected to the playground testnet. It works but has operational issues:

Problems

1. Wallet state lost on redeploy: The init-wallets.sh script starts wallets on deploy, but if the wallet-headless container restarts independently, wallets are lost. The facilitator can't settle until wallets are manually restarted.

2. No monitoring: No alerts if the facilitator fails to settle, if the wallet-headless disconnects, or if the dApp is down.

3. Docker image caching: Dokploy caches the latest Docker tag. We added pull_policy: always but need to verify it works on every deploy.

4. Stale compose file: Dokploy sometimes uses a cached compose file instead of pulling from git. The compose file must be manually updated via the API when changes are made.

Needed

• Health endpoint on the facilitator that checks wallet connectivity
• Auto-restart wallets if they disconnect (facilitator should detect and re-start)
• Monitoring/alerting for settlement failures
• Document the deployment process (how to redeploy, how to restart wallets, how to fund wallets)

Live URLs

• dApp: https://x402.hathor.dev
• API: https://api.x402.hathor.dev
• Facilitator: https://facilitator.x402.hathor.dev
• Headless (temporary): https://headless.x402.hathor.dev