It's pretty common practice for complex systems. If it's already dead, and nobody knows what's wrong or how to fix it, you try restarting it. (normally you first redirect traffic to a different region, but apparently big orgs are still running critical infrastructure with changes that affect all regions and can't be backed out)