Hello everyone! I am part of a project in which we are moving to elastic cloud. We also leverage some of the models for semantic search. When testing we have found several instances of the following error:
co.elastic.clients.elasticsearch._types.ElasticsearchException
Trained model deployment [model-name] is not allocated to any nodes
This happens when we try to input data to elastic (when calling the elastic client that connects to elastic cloud and indexes data). We have made a change to have at least one node always because we thought this was due to elastic cloud not scaling up quickly enough under high demand, but the issue persists.
Is there a way to avoid this kind of error? a mechanism to guarantee that the proper models are always deployed to the different nodes in the Elasticsearch cluster?