Implementation:Tensorflow Serving ProcessModelStatusRequest

Knowledge Sources	TensorFlow Serving
Domains	Operations, Monitoring
Last Updated	2026-02-13 17:00 GMT

Overview

Concrete tool for handling model status and metadata HTTP requests, provided by the HttpRestApiHandler class.

Description

HttpRestApiHandler::ProcessModelStatusRequest() and ProcessModelMetadataRequest() handle GET requests to the model status and metadata REST endpoints. They:

1. Parse model name, optional version, and optional label from the URL
2. Delegate to GetModelStatusImpl or GetModelMetadataImpl using ServerCore
3. Serialize the protobuf response to JSON using google::protobuf::util::MessageToJsonString()

URL routing is performed by ParseModelInfo() in http_rest_api_util.cc using a regex pattern: (?i)/v1/models(?:/([^/:]+))?(?:(?:/versions/(\d+))|(?:/labels/([^/:]+)))?(?:/(metadata))?

Usage

These endpoints are automatically available when the HTTP server is enabled. No special configuration is required beyond starting the server with --rest_api_port.

Code Reference

Source Location

• Repository: tensorflow/serving
• File: tensorflow_serving/model_servers/http_rest_api_handler.cc
• Lines: L180-202 (ProcessModelStatusRequest), L204-226 (ProcessModelMetadataRequest)
• URL Parsing: tensorflow_serving/model_servers/http_rest_api_util.cc L90-131

Signature

Status HttpRestApiHandler::ProcessModelStatusRequest(
    const absl::string_view model_name,
    const absl::optional<int64_t>& model_version,
    const absl::optional<absl::string_view>& model_version_label,
    string* output
);

Status HttpRestApiHandler::ProcessModelMetadataRequest(
    const absl::string_view model_name,
    const absl::optional<int64_t>& model_version,
    const absl::optional<absl::string_view>& model_version_label,
    string* output
);

Import

#include "tensorflow_serving/model_servers/http_rest_api_handler.h"

I/O Contract

Inputs

Name	Type	Required	Description
model_name	string_view	Yes	Name of the model
model_version	optional<int64_t>	No	Specific version number
model_version_label	optional<string_view>	No	Version label (e.g., "stable")

Outputs

Name	Type	Description
output	string*	JSON string with model status or metadata

Usage Examples

REST API Queries

# Get status of all versions
curl http://localhost:8501/v1/models/mnist

# Get status of specific version
curl http://localhost:8501/v1/models/mnist/versions/1

# Get status via label
curl http://localhost:8501/v1/models/mnist/labels/stable

# Get model metadata (signatures)
curl http://localhost:8501/v1/models/mnist/metadata

Example Response

{
  "model_version_status": [
    {
      "version": "1",
      "state": "AVAILABLE",
      "status": {
        "error_code": "OK"
      }
    }
  ]
}

Related Pages

Implements Principle

• Principle:Tensorflow_Serving_Model_Status_Query