Apache Kafka Complete Practice Exam 2025 – All-in-One Prep

An Unclean Leader Election occurs when there is a need to elect a new leader for a Kafka partition, and the broker that is chosen does not possess the most recent data. This scenario is particularly prone to data loss and inconsistency issues in a distributed system.

In the case where two followers crash while the leader remains available, it is likely that the leader is serving requests but may not be in sync with the followers. If the leader becomes unresponsive or fails, and the system needs to select a new leader from the remaining available followers, there may not be enough in-sync replicas available to ensure that the newly elected leader has the most recent data. This situation can lead to the election of a follower that may not have the latest updates, resulting in the Unclean Leader Election.

The other scenarios do not directly lead to this type of issue. For instance, if all brokers crash simultaneously, there are no brokers available to elect a new leader. When a leader goes offline but followers remain, a new leader can still be selected from the followers, typically ensuring that it is a qualified candidate with the most recent data. If a single broker becomes overloaded with requests, it might slow down operations, but it does not inherently affect leader election unless it fails entirely