Designing For Failure
Timeouts, retries, idempotency and graceful degradation, because every dependency fails eventually.
0 of 9 answered
Start here
When a dependency is unavailable
Foundation
- Not triedWhen a dependency is unavailableFoundation
- Not triedWhy a retry should waitFoundation
- Not triedWhy calls need a deadlineFoundation
- Not triedWhy every network call needs a timeoutFoundation
- Not triedOne slow dependency, whole site downCore
- Not triedWhen a retry is safeCore
- Not triedRetries that finished the outage offChallenge
- Not triedThe retry that charged twiceChallenge
- Not triedTimeouts along a chainChallenge