Your App Engine frontend queries a DB every 15 minutes for temperatures and must handle DB failures. How should the frontend behave?

Exponential backoff with a cap is the most robust strategy for handling transient database failures. It prevents overwhelming the database with continuous retries during an outage (which could worsen the problem) while still attempting to retrieve data once the database recovers. The 15-minute cap aligns with the application's 15-minute query interval, ensuring that the application doesn't spend more time retrying than it would waiting for the next scheduled query. Restarting database servers is an administrative task, not a frontend application behavior. Retrying every second is too aggressive and could exacerbate an outage. Reducing query frequency to once an hour unnecessarily delays data retrieval if the database recovers sooner.