Package com.gencior.triton.core.pojo

Class TritonModelIndex

java.lang.Object
com.gencior.triton.core.pojo.TritonModelIndex
public final class TritonModelIndex extends Object
Encapsulates metadata about a single model in the Triton repository index.

This class represents a model entry from the repository index, providing essential metadata about a model's availability and health status. Each model index entry indicates whether the model is currently ready for inference or if it's unavailable and why.

This is an immutable object that wraps the gRPC message RepositoryIndexResponse.ModelIndex.

State Values:

  • READY: Model is loaded and ready to accept inference requests
  • UNAVAILABLE: Model is not ready; check getReason() for details
  • LOADING: Model is currently being loaded (transient state)
  • UNLOADING: Model is currently being unloaded (transient state)

Usage Example:


 List<TritonModelIndex> models = repositoryIndex.getModels();
 for (TritonModelIndex model : models) {
     System.out.println("Model: " + model.getName() + " v" + model.getVersion());
     if (!"READY".equals(model.getState())) {
         System.out.println("Status: " + model.getState());
         System.out.println("Reason: " + model.getReason());
     }
 }
Since:
1.0.0
Author:
sachachoumiloff

  • Method Details

    • fromProto

      public static TritonModelIndex fromProto(GrpcService.RepositoryIndexResponse.ModelIndex proto)
      Creates a TritonModelIndex from a gRPC ModelIndex message.
      Parameters:
      proto - the gRPC ModelIndex message from Triton server
      Returns:
      a new TritonModelIndex instance

    • getName

      public String getName()
      Returns the name of the model.
      Returns:
      the model name as it appears in the repository

    • getVersion

      public String getVersion()
      Returns the version of the model.
      Returns:
      the model version (typically a semantic version like "1.0" or similar identifier)

    • getState

      public String getState()
      Returns the current state of the model.
      Returns:
      the state (typically "READY", "UNAVAILABLE", "LOADING", or "UNLOADING")

    • getReason

      public String getReason()
      Returns the reason for the current state, if not ready.

      This field is typically empty when the model state is "READY", and contains an error message or explanation when the model is unavailable.

      Returns:
      the reason text explaining the state, empty if not applicable

    • toString

      public String toString()
      Overrides:
      toString in class Object