
Producers are writing to a topic nobody created, and auto-create has given it one partition and a typo in its name. Create the correct topic through the Admin API with its partition count and replication factor, describe it, and delete the mistyped one. Skills: creating topics; describing and deleting topics.
Every topic runs on broker defaults: a week of retention for an hour-long feed and unbounded segments on a tiny disk. Set per-topic retention, segment size, maximum message size and timestamp type at creation, change them on the fly, and read back which values are explicit and which are inherited.
One partition means one consumer does all the work. Grow the topic's partition count, see what that does to where existing keys land, and show that shrinking is refused. Skills: partition growth; key-to-partition mapping.
Turn a broker full of auto-created, misconfigured topics into a defined estate: create what is missing, correct what is wrong without losing its records, grow what is under-partitioned, remove the listed strays and nothing else, and report the result from a readback.
Learn Kafka hands-on with real producers, consumers, topics, offsets, groups, transactions, streams, and operations
This path starts with topics and partitions and goes on to producers and consumers. You work through a Kafka client in Java, Node.js or Python, so you should be able to write a small program in one of them. No earlier Kafka experience is needed. The consumer module covers reading data back reliably: the poll loop, seeking to a position, and replaying a topic.
3 modules: Topics & Partitions, Producers & Records and Consumers & Polling.
12 exercises, all easy.
You can work in Java 25, Node.js 24 or Python 3.13.
No. Every challenge runs in a browser-based IDE with everything already set up, and your code is graded automatically against a test suite.