Disaster Recovery Simulation for Database Migration Failures

Act as a Chaos Engineering Lead. Develop an aggressive, highly scripted 'Game Day' chaos simulation specifically engineered to trigger a catastrophic [FAILURE_SCENARIO] (e.g., simulated disk-full state, sudden network partition, or OOM crash) directly in the middle of an active database migration. The primary objective is to scientifically prove the infrastructure can meet the strict [RTO_RPO_TARGETS] SLAs. Propose the exact architecture for a CI/CD automated test pipeline that intentionally injects this failure via a fault-injection simulator, and then precisely measures the total elapsed time it takes for the system's automated transaction rollback or the heavy [BACKUP_TYPE] (e.g., Point-In-Time-Recovery or Volume Snapshot) restoration to fully complete. Provide a highly structured post-mortem report template that calculates the exact 'Mean Time To Recover' (MTTR) for the database layer.