Retries and Circuitbreakers in Takealot Microservices

on YouTube ↗06m

" summary

Takealot runs Python microservices and demonstrates a flaky endpoint that succeeds 5/10 calls then fails 5/10; naive retries hammer downstream pods. Implemented retries use up to 3 attempts with exponential backoff, and the system purposely does not retry on timeouts to avoid adding ~4s of latency. A circuit breaker trips after about 3–4 failures, returns circuit open 10s immediately, then probes and closes on a successful call. Live demo shows these exact numbers and behavior so you can judge impact on latency and load.

AI-written summary · about

" chapters (5)

1. 0:00Start
2. 0:53Conclusion
3. 1:28Demo Time
4. 3:09Proper Retries
5. 4:06Circuit Breaker

" related talks

• 25mFail Faster: Adding Circuit Breakers to your APIsCraig Freeman · Node+JS Interactive 2018
• 49mManaging Resiliency in MicroservicesNish Anil · Full Stack Virtual Conference 2021
• 40mLet it fail: resilience patterns for microservicesJoão Alves · Commit Conf 2019
• 48mLost in transaction?Bernd Rücker · Jfokus 2018
• 30mMaking Your Apps and Infrastructure Services Failure-Resilient with...Yaron Schneider, Henry Spang · KubeCon + CloudNativeCon Europe 2022
• 29mFault Tolerance is a Requirement Not a Feature - Adar Danait, Lilian Ernest, CitiAdar Danait, Lilian Ernest · SpringOne Platform 2017
• 42mArchitecting for Failures in MicroservicesBhakti Mehta · Devoxx US 2017
• 35mSCALA BASED SERVICES RESILIENCE IN THE REAL WORLDBartosz Jankiewicz · Scalar 2018