liveBook · Manning

Under-replicated partition

As explained in Kafka for Architects

An under-replicated partition is a Kafka partition whose available replicas do not currently satisfy its configured replication requirements.

An under-replicated partition is a Kafka partition whose available in-sync replicas do not currently satisfy its configured replication requirements. It signals reduced replication health and possible data-durability risk. section 11.1.2 section 11.2.2

Treat the condition as an operational signal with architectural consequences:

  • Monitor the number of under-replicated partitions alongside offline partitions to detect replication imbalance and durability risk. section 11.2.2
  • Handle broker removal through planned reassignment and leadership migration; shutting down first can leave partitions under-replicated. section 11.1.2
  • Restore a failed broker’s replicas to the in-sync replica set by allowing them to catch up before declaring the cluster healthy. section 9.1.3

Broker failure and failover

When a broker fails, the controller selects new leaders from the in-sync replica set. If no in-sync replica is available, the partition becomes under-replicated. With unclean leader election disabled, it may remain offline to preserve consistency; with it enabled, an out-of-sync replica may become leader to prioritize availability. Clients may see transient replica-not-available errors until refreshed metadata directs them to the new leaders. section 9.1.3

How Kafka for Architects covers it

How deep it goes
Explains the condition, monitoring significance, broker-failure behavior, leader-election trade-offs, and recovery through replica catch-up.
Examples
Uses broker removal, failed-broker failover, replica catch-up, and under-replicated-partition monitoring as concrete operational examples.
What you can do after
Diagnose replication health, plan safer broker removal, interpret failover behavior, and distinguish consistency from availability trade-offs.
How it connects
Connects under-replicated partitions with in-sync replicas, partition leadership, broker failure, unclean leader election, and cluster monitoring.
Cover of Kafka for Architects
liveBookmanning.com
Read Kafka for Architects on liveBook Get the book at manning.com

Also explained in

Compare how 2 books explain under-replicated partition →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Kafka for Architects. The section links point to the parts of the book it draws on.