Question 50 of 50
What errors or challenges have you faced with Kafka?
Question
"What errors or challenges have you faced working with Kafka? How did you resolve them?"
What the interviewer wants to assess
Operational maturity and diagnostic ability — the question that most distinguishes someone who only codes against Kafka from someone who has actually operated it under real pressure, during a production incident.
This question also has no generic answer
Just like the previous question, this one requires a real experience of your own. Don't force a made-up challenge — if you haven't faced a production incident with Kafka yet, it's honest to say so and talk about a study or test-environment challenge, or a scenario you know is common (citing your sources of that knowledge, like this book or the documentation).
How to structure the answer
A strong answer follows the logic of a post-mortem, not just a complaint: root cause, diagnosis, and fix. Real, common challenges that tend to show up in this kind of answer — described in this book in case you need to refresh the technical cause of each one:
- Excessive rebalances from a misconfigured
max.poll.interval.ms, kicking healthy but slow consumers out of the group (Chapter 6). - Duplicate messages in production due to missing idempotency, discovered only when a customer complained about a duplicate charge (Chapter 11).
- Poison pill stalling an entire partition because of an unhandled deserialization bug (Chapter 9).
- Poorly sized partitions from the start, requiring migration to a new topic instead of just increasing the existing one's partition count (Chapter 4).
- Growing consumer lag not detected in time, due to a missing alert (Chapter 15).
Dica de entrevista
Always finish by explaining what changed after the incident — a new alert, a configuration change, a code review. That shows you didn't just fix the immediate problem, but improved the system (or the team's process) so it doesn't happen again.
"I have never had any problems with Kafka"
That answer, even if true, tends to sound like a lack of real experience with the system in production under load — most teams that operate Kafka long enough have faced at least one of the challenges listed above. If your experience is limited, it's more honest (and more convincing) to admit that and show you understand, in theory, where these problems tend to show up.
Pode vir a seguir
This is typically the last technical question in a Kafka interview — from here, the conversation usually moves on to culture fit or other technologies in the stack.
Related chapters