Course

machine-learning-zoomcamp

Question

In my local kind deployment, I observed that rolling updates are technically enabled by default, but with only one replica, downtime is possible. In production, how do you determine the minimum number of replicas needed to guarantee zero-downtime for model updates, considering startup latency and readiness probes?

Answer

No answer

Checklist

• I have searched existing FAQs and this question is not already answered
• The answer provides accurate, helpful information
• I have included any relevant code examples or links