Manages the namespacing of a cluster, so a cluster can be scoped to only operate on indexes with a specific prefix.

This is useful for a few purposes:

• Running multiple instances of Snap, across completely different applications, against a single cluster. You might not want do this in production, but it's useful to only run one copy of ElasticSearch/OpenSearch locally when working on multiple applications.

• Running multiple environments of the same application. You don't want your tests to blow away all your development data each time they run.

• Running tests in parallel, so each test process cannot see or act upon indexes that another test is operating on. This gives you the benefits of Ecto style sandboxing, meaning your tests can run with async: true even though ElasticSearch/OpenSearch doesn't support isolated transactions.

All of the convenience modules respect namespacing:

• Snap.Bulk
• Snap.Document
• Snap.Indexes
• Snap.Multi
• Snap.Search

The low level HTTP API on Snap does not.

There are two types of namespace:

Config namespace

You can namespace the entire Snap.Cluster in the config. This is used to achive isolation at a per application and environment level.

For example, to isolate your development and test environments, set index_namespace: "dev" in your dev config, and index_namespace: "test" in your test config.

You might even want to isolate multiple applications on a per environment basis, with index_namespace: "app1-dev", for example.

Process namespace

The process namespace is used to achive isolation between parallel running tests. It's not likely you'll want to use this in other situations.

For more information, see Snap.Test.