Why opensearch natural language query generation on aws always gives response error? Am I doing something wrong?

Question

I am trying to generate PPL queries from natural language on opensearch dashboard. The Steps I did as per aws documentation:

1. Upgraded the version on my opensearch domain to 2.13
2.  Made sure fine-grained access control is enabled
3. Made sure service software is R20240520-P4 or higher.(it is OpenSearch_2_13_R20240520-P5 (latest))
4. Create the "query_assistant_access" role with permissions:

```
cluster:admin/opensearch/ml/config/get

cluster:admin/opensearch/ml/execute

cluster:admin/opensearch/ml/predict

cluster:admin/opensearch/ppl

```

5. Attached the role to the user I am using
6. Went to Event Explorer under Logs on Opensearch Dashobard. Selected the index. Wrote the prompt: Give me 10 documents with name "foo". Clicked on Generate and Run
7. Got the "Response Error". Tried a number of times, its giving response error everytime without any additional details.

Have anyone tried this feature? Am I doing something wrong? followed this: https://docs.aws.amazon.com/opensearch-service/latest/developerguide/natural-language-query.html

rePost-User-1274309 · Answer

Hello, thanks for trying the feature and we are sorry that you are running in to an error. Could you please,

1. Copy paste the PPL that the model generates?
2. Answer if you are using a custom schema or one of the prebuilt ones? If former, could you please tells us the number of columns the schema has? The model has a token limit and schemas with more than 800-1000 columns will not be processed by the model.

Thanks!

Why opensearch natural language query generation on aws always gives response error? Am I doing something wrong?

Add your answer

Relevant content