A Partition Leader Went OfflineKafka · easy · ~10 minYou're on call for the orders pipeline. At 02:14 PagerDuty fires: UnderReplicatedPartitions > 0 on the payments-events cluster. Producers are still writing, but your monitoring shows some partitions are no longer fully replicated. Broker kafka-3 stopped reporting to the dashboard a few minutes ago.Work through the evidence and decide how to bring the cluster back to health — without losing data.Evidence

Metrics

Broker Logs

Topic Config

Cluster

Checkpoints

What is actually happening right now?

Producers are still succeeding. Why haven't writes stopped, given replication factor 3?

What is the correct first action?

In an interview, how would you explain the risk of enabling unclean leader election here?