Kafka’s min.insync.replicas: Avoiding Data Loss

In Apache Kafka, the min.insync.replicas configuration plays a crucial role in ensuring data durability and resilience. This setting determines the minimum number of replica copies (or in-sync replicas) that must acknowledge the receipt of a record before the producer can consider the write operation successful.

Let’s break this down:

  • Replicas: Think of these as backup copies of your data. Kafka doesn’t just save your message once; it saves it several times on different servers. This is like having multiple backup copies of your important files.
  • In-sync Replicas: These are the backup copies that are completely up-to-date. They are exact mirrors of the original data.
  • Producer: This is the part of Kafka that sends out your messages. It’s like a messenger who delivers your data to Kafka.

The min.insync.replicas setting decides how many of these backup copies need to have received and stored your message successfully before Kafka tells the producer, “Yes, I’ve got this; your data is safe.”

Why does this matter? By requiring more than just one backup copy to confirm that they have the data, Kafka makes sure that even if one server has issues, your data isn’t lost—it’s still safe on other servers.

If you are interested in video lessons then check my video course Apache Kafka for Event-Driven Spring Boot Microservices.

Configuring min.insync.replicas

You can set this configuration in two scenarios:

  1. When creating a new Kafka topic
  2. When modifying an existing topic

Let’s start with a new topic.

Configure min.insync.replicas at the time when creating a new topic

Open a terminal window and navigate to your Kafka directory, which contains the bin folder with Kafka scripts.

To create a topic with a specific min.insync.replicas setting, you would use a command like this:

./kafka-topics.sh --create --topic your-topic-name --partitions 3 --replication-factor 3 --bootstrap-server localhost:9092 --config min.insync.replicas=3

In this example, I’ve created a topic named your-topic-name with 3 partitions and a replication factor of 3. By setting --config min.insync.replicas=3, I’m making sure that all three replicas must acknowledge a write operation. If any replica fails to do so, the producer will receive an error.

If you’d like to require fewer acknowledgments, you can set the min.insync.replicas to a lower number, like 2. This would still provide some level of fault tolerance but requires fewer acknowledgments for a write operation to be successful.

Configure min.insync.replicas for existing topic

Now, suppose you have an existing topic and want to change its min.insync.replicas setting. For this, Kafka provides a different script:

./kafka-configs.sh --bootstrap-server localhost:9092 --alter --entity-type topics --entity-name your-topic-name --add-config min.insync.replicas=2

By executing the command above, you’re updating the min.insync.replicas setting for your-topic-name to 2. This change ensures that at least two replicas must be in sync for the write operation to be acknowledged.

To check if your changes are applied, use the describe command:

./kafka-topics.sh --bootstrap-server localhost:9092 --describe --topic your-topic-name

This will output the current configuration of your topic, including the min.insync.replicas value.

By understanding and properly configuring min.insync.replicas, you ensure that your Kafka system can handle server failures without losing data, as long as the number of failures does not exceed the fault tolerance level you’ve set.

Remember, setting the min.insync.replicas to 1 means there is no fault tolerance, whereas setting it to the number equal to the replication factor ensures full fault tolerance. However, requiring acknowledgments from all replicas may not always be practical, as it can impact performance. Hence, choose a value that balances resilience and throughput according to your needs.


I hope this guide made it easier for you to understand how min.insync.replicas works in Apache Kafka and why it’s important.

If you’re looking to learn more about Kafka, feel free to check out my other Apache Kafka tutorials for beginners. They’re straightforward, easy to follow, and I’ve made sure to keep things simple, just like we did here. See you there!

If you are interested in video lessons then check my video course Apache Kafka for Event-Driven Spring Boot Microservices.