Showing 13 of 13 posts
Multi-Region Cluster observers, Cluster Linking and tiered storage complete the DR and storage story in twelve scenarios - and the playground's URL format goes public, with monedula-sim-link to turn a live cluster, or an AI agent's description, into a link.
Stretched clusters, MirrorMaker 2 offset translation, failover and failback - and a terminal running the real Kafka CLI tools against the simulated cluster. The DR module explains the cluster shapes free play has offered since v1.3.
Kill brokers, split the network, and watch your cluster recover. v1.7 adds the Failure Lab to free play, six chaos scenarios, and a more usable mobile playground.
Two packs at once. Free play completes its topology ladder with the 3-DC stretched and 2.5-DC witness clusters, and 19 new scenarios cover the log's lifecycle - retention, compaction, tiered storage - plus the KRaft control plane and the operations you run against a live cluster.
Exactly-once, read_committed and the last stable offset, begin/commit/abort, and the consume–process–produce loop. The transactions module makes EOS - Kafka's hardest topic - something you can step through. Free play also gains a new cluster type: active/active.
Monedula GitOps manages Kafka topics, access, quotas, schemas, users, and Confluent RBAC from version-controlled manifests, through either a CLI or a Kubernetes operator.
The replication module arrives: ISR mechanics, the high watermark, acks=all and min.insync.replicas, leader election and what an unclean election actually loses. Twelve scenarios on the guarantees Kafka is famous for.
Consumer groups, assignment strategies, static membership and the two timeouts that decide when a member is gone. Ten new scenarios that make the rebalance - Kafka's most misunderstood moment - something you can watch instead of guess at.
Kafka Simulator v1.1 adds ten scenarios covering producer batching, partitioning, compression, backpressure, retries, and idempotence.
KIP-714 lets Kafka clients push metrics to the broker. Turn it on with a client-metrics subscription and forward them over OTLP to OpenTelemetry.
Kafka's most important lessons live in the transitions: before and after a broker dies, before and after the ISR shrinks. We built a browser-based simulator so you can break a cluster safely and watch exactly why it reacts the way it does.
Moving an existing Kafka estate from ACLs to Confluent RBAC is a risk-management exercise, not a syntax conversion. Here's a safe, auditable, repeatable migration workflow - and the tool we built for it.
Why Kafka metrics are painful, why JMX dashboards often show no data, and how to export Kafka metrics natively over OTLP.