feat(deploy): add auto-rollback on health failure [Sprint 2.5 D3]

[Piboonsak]

Sprint 2.5 — D3: Auto-Rollback on Health Failure

Branch: feat/sprint-2.5-d3-auto-rollback
Base: main
Depends on: D1 ✅ (bkg-save.sh merged), D8 (manifest — can be done in parallel)

Task

Add auto-rollback to deploy and hotfix workflows when health check fails.

Files to create/modify in Piboonsak/openclaw_github:

1. scripts/auto-rollback.sh (NEW)

   • Read bkg-latest.json symlink from /data/openclaw-config-archive/ to find last known good config
   • Restore that config to active config location on VPS
   • Recreate the Docker container with restored config
   • Run health check to verify recovery
   • Output clear rollback status (success or failure)

2. .github/workflows/deploy-openclaw-github-private-secrets.yml (MODIFY)

   • On health check failure: call auto-rollback.sh via SSH
   • Report rollback status in workflow output
   • If rollback also fails: fail workflow with clear error message

Acceptance Criteria

• Health check failure → auto-rollback to bkg-latest.json config
• Auto-rollback recreates container + verifies health
• Rollback status reported in workflow output
• If rollback also fails: workflow fails with clear error message

Commit Message

feat(deploy): add auto-rollback on health failure (#202)

Reference

• Master issue: Piboonsak/Openclaw#202
• D1 PR feat(deploy): add BKG save step to deploy workflow (#202) #32: scripts/bkg-save.sh and bkg-latest.json already in main

Labels

• sprint-2.5
• copilot