# AWS hub application recovery — action recording, 30 July 2026

Status: **PASS**  
Accepted recovery objective: **600 seconds**  
Measured result: **218 seconds**

## Controlled action

- The primary Ireland hub instance `i-074bf0cb910a99e02` was deliberately stopped at `2026-07-30T06:48:00Z` (14:48:00 SGT).
- The cold recovery Auto Scaling group launched replacement `i-0007949b2f976f068` in the recovery subnet with private address `10.1.2.154`.
- The health probe required HTTP `200` through stable private DNS and the public Application Load Balancer.
- Service recovery was confirmed after **218 seconds**, below the accepted 600-second objective.
- The primary was restarted and verified at `10.1.1.61`.
- Private DNS was returned to the primary, and the recovery group was restored to desired capacity `0` with zero recovery instances.
- All four project CloudWatch alarms were `OK` after restoration.

## Cross-cloud context

VPN #1, VPN #2 and direct VPN #3 were live and exercised in the same 30 July action-recording window. VPN #2 passed Azure-to-Ireland private ICMP and HTTP tests, and VPN #3 passed bidirectional Virginia-to-Azure private ICMP tests.

The **218-second** result measures recovery of the stateless AWS hub application through private DNS and the public ALB. It does not claim uninterrupted Azure-client service during every recovery transition, recovery of the VGW or VPN connections, regional disaster recovery, or full high availability.

## Recording

The genuine-action recording is published as chapter 6:

- `https://dv50c2wya2oa2.cloudfront.net/public/06-hub-recovery.mp4`
- SHA-256: `51e29ec6bff4669ce551481186b01710ca538cb806a66813ada2a56dcfe3af3d`
- Duration: approximately 178.8 seconds

Credentials, private keys and VPN pre-shared keys are not included.