[BUG] forcemerge Request Fails After 5 Minutes Despite Custom requestTimeout Configuration

[eacet]

What is the bug?

When calling the forcemerge function with the requestTimeout parameter set to 1800000 (30 minutes), the request consistently fails after 5 minutes, despite the timeout being explicitly configured for a longer duration.

How can one reproduce the bug?

1. Use the library to call the forcemerge function.
2. Set the requestTimeout parameter to 1800000 (30 minutes).
3. Observe that the request times out after 5 minutes instead of the configured 30 minutes.

What is the expected behavior?

The request should respect the requestTimeout configuration and allow the operation to continue for up to 30 minutes before timing out.

What is your host/environment?

Operating System: macOS Sequoia 15 & Amazon Linux 2 (Amazon Linux 2023), used in AWS CodeBuild environment
Library Version: 2.13.0
Node.js Version: 22

[dblock]

Does the same request wait with curl? Just trying to isolate where the timeout happens.

[eacet]

Does the same request wait with curl? Just trying to isolate where the timeout happens.

curl request doesn't throw timeout error but it just returns below response immediately.

{"_shards":{"total":10,"successful":5,"failed":0}}

[dblock]

@eacet That seems odd, can you put up a simple repro for this along with the curl command you're using? Something working like a sample in https://github.com/opensearch-project/opensearch-js/tree/main/samples.

[nhtruong]

I don't see anywhere in the code base that would modify the requestTimeout value. It's ultimately passed to Node.js native http or https's request function as a timeout parameter and as far as I know, these libraries also do not have a limit for this value, either. The limitation must on the server.

For now, you can try wait_for_completion=false to perform forcemerge asynchronously.

[nhtruong]

Closing this issue since this is not a client bug.