Gateway SIGUSR1 restart keeps PID=1 alive but clears in-memory lock state,
leaving .lock files that appear valid but are orphaned. This causes session
file lock timeouts when new agent sessions try to acquire the same locks.
Solution: On container start, clean all session lock files since they're
guaranteed orphaned (new process state). Lock files are safe to remove on
startup because the gateway hasn't started any agent sessions yet.
Fixes session lock timeout errors after gateway restarts in Docker.
- Add docker-compose.phala.yml for Phala Cloud deployment
- Add docker-entrypoint-phala.sh for auto-configuration
- Auto-configure Redpill provider on first boot
- Auto-configure Telegram/Discord channels from env vars
- Support gateway auth (token/password) modes
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
