fix: Enable detection of unresponsive or crashed Python backend stub process

[pskiran1]

• Added an optional TRITONBACKEND_ModelInstanceReady function that backends can implement to report their live health status
• Enhanced model readiness checks to include backend-level health validation

[Copilot]

Pull request overview

This PR introduces a mechanism to detect unresponsive or crashed Python backend stub processes by adding an optional IsReady check that can be implemented by backends. The check is called when determining model readiness status.

Key changes:

• Added an optional TRITONBACKEND_ModelInstanceReady function that backends can implement to report their health status
• Implemented a readiness check propagation from model instances up through the model hierarchy to the inference server
• Enhanced model readiness checks to include backend-level health validation

Reviewed changes

Copilot reviewed 9 out of 9 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
include/triton/core/tritonbackend.h	Declares the new optional TRITONBACKEND_ModelInstanceReady API function
src/backend_manager.h	Adds typedef and accessor for the model instance ready function pointer
src/backend_manager.cc	Loads and stores the optional TRITONBACKEND_ModelInstanceReady function from backend libraries
src/backend_model_instance.h	Declares IsReady method for model instances
src/backend_model_instance.cc	Implements IsReady to call backend's ready function if available
src/backend_model.h	Declares override of IsReady method for TritonModel
src/backend_model.cc	Implements IsReady to check readiness of all model instances
src/model.h	Adds virtual IsReady method to base Model class with default success implementation
src/server.cc	Integrates the new readiness check into the model status query path

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.