Troubleshooting

Common checks:

• Remove the label, wait for destroy, then re-label to start fresh.
• Increase instance_type if builds are memory-heavy.
• Increase workflow timeout-minutes if your stack boots slowly.

Useful runtime diagnostics:

• Review Action logs for heartbeat/SSH/compose output.
• SSH into instance and inspect Docker state:

ssh ec2-user@<ip>
docker ps -a
docker logs <container>

If preview deploys but URL is not reachable:

• verify compose service health checks
• verify ports / default_port
• if using proxy_tls, verify target mapping (proxy_tls: service:port)

If private images fail to pull:

• set registries input
• ensure token has pull rights for that registry