Sharding the configuration server failure

To distribute the whole dataset to different machines and servers, that is, horizontal scaling, MongoDB provides you with the sharding feature. Using this feature, you can split the database into different pieces in separate servers. In the next chapters, we will learn more about this feature in detail.

MongoDB uses configuration servers to store the configuration for the sharding settings and relations between members. For the production environment, these configuration servers are exactly three in number.

If one or two distinct configuration servers become unavailable, the clustering works properly, but it cannot create more chunks from the dataset, which means that it will be read-only.

Chunks are pieces of database in separate machines, and MongoDB splits the database into chunks of data based on a sharded key. MongoDB creates chunks from the database after write operations based on the given sharded key.

If all three configuration servers become unavailable, the cluster will be unavailable as well. You should replace the configuration servers as soon as possible. Then, the cluster will continue to use the configuration servers and work properly.

We will talk about sharded keys and chunks in detail in the next chapters.