Open the editor

Review cards · 16 cards

Resilience

Timeouts, retries with backoff, circuit breakers, rate limits and graceful degradation.

Train resilience in daily review

Cards

  1. What is graceful degradation? Give an example. easy Flashcard
  2. A service gets 150% of the load it can handle. What keeps it most useful? easy Multiple choice
  3. Why must every remote call have a timeout, and how do you pick its value? easy Flashcard
  4. Why add jitter to exponential backoff when clients retry? medium Flashcard
  5. What is a bulkhead, and what failure does it prevent? medium Flashcard
  6. What does a circuit breaker do when calls to a dependency keep failing? medium Multiple choice
  7. What is the flaw of a fixed-window rate limiter (100 requests per calendar minute), and what fixes it? medium Flashcard
  8. A _____ rate limiter puts requests in a queue of fixed size and lets them out at a constant rate; when the queue is full, new requests are rejected. medium Fill in the blank
  9. A request goes through services A → B → C → D. A, B and C each make up to 3 attempts when their call fails. If D is down, how many calls can one user request send to D? medium Multiple choice
  10. A rate limiter runs as 10 replicas, and each keeps its own counter with a limit of 100 requests a minute per client. What happens? medium Multiple choice
  11. Which rate-limiting algorithm lets a client burst up to N requests at once, then continue at a steady average rate? medium Multiple choice
  12. Passing the time a request has left along with each downstream call, so services stop work the original caller has already given up on, is called _____. hard Fill in the blank
  13. The rate limiter's shared counter store goes down. Should the limiter fail open (allow) or closed (reject)? hard Flashcard
  14. What is a hedged request, and what does it trade? hard Flashcard
  15. A sliding window counter limits a client to 100 requests a minute. The previous minute had 80 requests; 20 seconds into the current minute there have been 30. What count does the limiter estimate for the last 60 seconds? hard Estimate
  16. Under load, requests wait 30 s in a server's queue, but clients give up after 5 s. What is the main problem? hard Multiple choice

More topics