Jump to content

Connect SuperML | Leeroopedia MCP: Equip your AI agents with best practices, code verification, and debugging knowledge. Powered by Leeroo — building Organizational Superintelligence. Contact us at founders@leeroo.com.

Implementation:Tensorflow Serving Static Storage Path Source

From Leeroopedia
Knowledge Sources
Domains Model Serving, Source Management
Last Updated 2026-02-13 00:00 GMT

Overview

A simple StoragePathSource that fires the aspired-versions callback exactly once with a single hard-coded servable version path, useful for testing and experimental deployments.

Description

StaticStoragePathSource implements the Source<StoragePath> interface to provide a minimal, non-polling source that serves a single, static model version. Unlike FileSystemStoragePathSource which continuously monitors a directory for new versions, StaticStoragePathSource calls the aspired-versions callback exactly once with a fixed servable name and version path specified in its configuration.

The class is configured via StaticStoragePathSourceConfig protobuf, which specifies the servable name, version number, and filesystem path. The Create static factory method initializes the source from the configuration. When SetAspiredVersionsCallback is called (by the serving manager), it triggers the callback with the configured single version.

Key characteristics:

  • No filesystem polling or monitoring
  • Single version only; no automatic version updates
  • Lightweight and deterministic behavior
  • Copy construction disabled via TF_DISALLOW_COPY_AND_ASSIGN

Usage

Use this source for testing scenarios where a fixed model version is needed without filesystem monitoring overhead. It is also suitable for experimental or embedded deployments where models are loaded once and never updated. For production deployments with version management, use FileSystemStoragePathSource instead.

Code Reference

Source Location

  • Repository: Tensorflow_Serving
  • File: tensorflow_serving/sources/storage_path/static_storage_path_source.h (lines 1-53)

Signature

class StaticStoragePathSource : public Source<StoragePath> {
 public:
  static Status Create(const StaticStoragePathSourceConfig& config,
                       std::unique_ptr<StaticStoragePathSource>* result);
  ~StaticStoragePathSource() override = default;

  void SetAspiredVersionsCallback(AspiredVersionsCallback callback) override;
};

Import

#include "tensorflow_serving/sources/storage_path/static_storage_path_source.h"

I/O Contract

Inputs

Name Type Required Description
config StaticStoragePathSourceConfig Yes Configuration with servable name, version, and filesystem path

Outputs

Name Type Description
result std::unique_ptr<StaticStoragePathSource>* Created source instance
(callback) AspiredVersionsCallback Called once with a single aspired version when SetAspiredVersionsCallback is invoked

Usage Examples

Creating a Static Source

StaticStoragePathSourceConfig config;
config.set_servable_name("my_model");
config.set_version_num(1);
config.set_version_path("/models/my_model/1");

std::unique_ptr<StaticStoragePathSource> source;
TF_RETURN_IF_ERROR(StaticStoragePathSource::Create(config, &source));
// Connect to a target (e.g., SourceAdapter -> Manager)

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.
Principle
Implementation
Heuristic
Environment