A multi-AZ Auto Scaling service passes EC2 checks even when the application is hung. ALB marks affected targets unhealthy, but the group does not replace them. Which change is appropriate?

1. Enable the relevant load-balancer health checks for the group and use a grace period aligned with application startup.
2. Rename the unhealthy target group to force application recovery.
3. Increase only log retention without changing group health configuration.
4. Disable all target checks and assume the running instance state proves readiness.