Ollama connector blueprint

[cjrolo]

Description

This change documents the creation and testing of local LLMs within opensearch (Ollama for example)

Related Issues

Resolves #4146

Check List

• New functionality includes testing.
• New functionality has been documented.
• API changes companion pull request created.
• Commits are signed per the DCO using --signoff.
• Public documentation issue/PR created.

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.
For more information on following Developer Certificate of Origin and signing off your commits, please check here.

[mingshl]

maybe you can do register and deploy is true, which save one step, POST /_plugins/_ml/models/_register?deploy=true

[cjrolo]

I had that originally and then removed because IMO POST /_plugins/_ml/models/_register? Deploy=true gives a sense of automagic things happening. And with the explicit POST readers have to understand that there is this step (which you can automate) of deploying the connector.

After your comment about the tutorial, I think adding all the options in the tutorial to keep the blueprint cleaner would be a good one. But I'm open to exactly the opposite: making the blueprint very clean, even with such automations in place (and a note saying that they are there) and make the tutorial more dense with explanations about the deploying (and others) of models.

What option do you prefer?

PS: Just for reference, my first commit didn't have any deploy because I have auto deploy in my test cluster, that is something I would like to avoid for new readers.

[mingshl]

it would be nice to have both options, so the blueprint is friendly for starter and also giving a good tip for advanced users who wants to speed up.

[cjrolo]

I will do that.

[mingshl]

this looks great! You nailed it. Just added a very minor comment. It would be even nice if you can add the tutorial here using Ollma with the RAG processor, from the nice demo you built for conversation chat.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 81.91%. Comparing base (70d8f18) to head (286c433).

Additional details and impacted files

@@            Coverage Diff            @@
##               main    #4160   +/-   ##
=========================================
  Coverage     81.91%   81.91%           
- Complexity     8874     8875    +1     
=========================================
  Files           762      762           
  Lines         38166    38166           
  Branches       4264     4264           
=========================================
+ Hits          31263    31264    +1     
+ Misses         5078     5076    -2     
- Partials       1825     1826    +1     

Flag	Coverage Δ
ml-commons	81.91% <ø> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[cjrolo]

Ok let me add that too this PR then.

[ylwu-amzn]

To use this, you need to enable plugins.ml_commons.allow_custom_deployment_plan setting.
To make the doc easier to follow, I think you can just use POST /_plugins/_ml/models/<MODEL_ID>/_deploy

[cjrolo]

That is a mistake on my part, let me fix that.

[mingshl]

LGTM, the CI failure are not related to your code changes, I am going to retrigger one more time.