Elasticsearch "index_not_found_exception" in CI

A request targeted an index that does not exist on the cluster. A CI service container starts empty, so any query that assumes an index is present fails with index_not_found_exception until a setup step creates and populates it.

What this error means

A search or update returns HTTP 404 with "index_not_found_exception" and "no such index [name]". It passes locally where the index already exists.

elasticsearch

{"error":{"root_cause":[{"type":"index_not_found_exception","reason":"no such index
[products]","index":"products"}],"type":"index_not_found_exception"},"status":404}

Common causes

The fresh cluster has no indices

A CI container starts empty; unlike a developer machine, no indices exist until the test setup creates them.

Strict index resolution rejects missing targets

With ignore_unavailable off, a query against a non-existent index is a hard 404 rather than an empty result.

How to fix it

Create and seed the index in setup

Add a setup step that creates the index (with its mapping) before tests run.
Seed any fixture documents the tests expect.
Refresh the index so the documents are searchable immediately.

Terminal

curl -X PUT "http://localhost:9200/products"
curl -X POST "http://localhost:9200/products/_doc?refresh=true" \
  -H 'Content-Type: application/json' -d '{"name":"widget"}'

Tolerate missing indices where appropriate

For read paths that may run before seeding, pass ignore_unavailable=true so a missing index yields an empty result instead of a 404.

How to prevent it

Create every index the tests need in a deterministic setup step.
Do not assume a CI cluster inherits indices from anywhere.
Use refresh=true after seeding so documents are searchable at once.