Hi — You can use centralized logging and you can also keep the old crashed instance(s) for later inspection — Usually you keep only the EBS volumes though. Order of priority: Fix, Resume service, Understand.