How to define a ML node?

[andreasMore]

Hi Team,

In version 2.4 of OpenSearch, machine learning nodes have been added. I am using OpenSearch version 2.11.0. How can I define a deployment that consists of 3 nodes that do everything except ML and one node that should be used only for ML (maybe also with GPU)? For the deplyoment I want to use helm. Unfortunately I could not find a tutorial for this.

Thanks a lot
Andreas

[peterzhuamazon]

Hi @andreasMore could you try passing the node role here?
https://github.com/opensearch-project/helm-charts/blob/main/charts/opensearch/values.yaml#L16-L20

@prudhvigodithi

Thanks.

[andreasMore]

Hi @peterzhuamazon ,
thank you for your answer. Unfortunately it’s not what I was looking for, because then all nodes would also have the ml role, which is not recommended.

[prudhvigodithi]

Hey @andreasMore you need to maintain a seperate values,yaml file with just ML node role, install the chart, it should connect to the existing cluster which is already deployed with different values.yaml file.

[prudhvigodithi]

Here is an example https://opensearch.org/blog/setup-multinode-cluster-kubernetes/.
Adding @TheAlgo

[andreasMore]

Thank you for the information. Now it works as intended.

[prudhvigodithi]

Thanks for the update @andreasMore