How aws/sagemaker-distribution works
This repository publishes versioned SageMaker Distribution images: users pull a CPU/GPU image and work in JupyterLab or use the image for SageMaker inference. The repository also includes a CLI for package-staleness and package-size reports. Inference-server wiring is shown from the sampled v3.8 template; README documents other image capabilities, including Amazon Q Agentic Chat, but their implementation is not present in the sampled source. Report relationships are limited to the sampled CLI and SOURCE INDEX references.
Main entry points: server.py Inference server · v2.10.0 CPU/GPU images · async_handler.py Async handler · sync_handler.py Sync handler
Components 16
Distribution images
- CPU/GPU images
build_artifacts/v2/v2.10/v2.10.0 - JupyterLab
template/v1
Inference serving
- Inference server
template/v3/v3.8/dirs/etc/sagemaker-inference-server/tornado_server/server.py - Serving config
template/v3/v3.8/dirs/etc/sagemaker-inference-server/utils/environment.py - Async handler
template/v3/v3.8/dirs/etc/sagemaker-inference-server/tornado_server/async_handler.py - Sync handler
template/v3/v3.8/dirs/etc/sagemaker-inference-server/tornado_server/sync_handler.py - Serving errors
template/v3/v3.8/dirs/etc/sagemaker-inference-server/utils/exception.py - Inference logging
template/v3/v3.8/dirs/etc/sagemaker-inference-server/utils/logger.py - Model handler
Package reports
- Report CLI
src/main.py - CLI configuration
src/config.py - Package reports
src/package_report.py
Image integrations
- Amazon Q Chat
build_artifacts/v2/v2.10/v2.10.0/dirs/etc/sagemaker-ui - Workflow client
build_artifacts/v2/v2.10/v2.10.0/dirs/etc/sagemaker-ui/workflows/workflow_client.py
Other components
- Image user
- SageMaker endpoint
Connections 16
- Image user → CPU/GPU images: pulls
README.md - Image user → JupyterLab: uses
README.md - CPU/GPU images → Inference server: includes
- Inference server → Serving config: reads
template/v3/v3.8/dirs/etc/sagemaker-inference-server/tornado_server/server.py - Inference server → Async handler: dispatches
template/v3/v3.8/dirs/etc/sagemaker-inference-server/tornado_server/server.py - Inference server → Sync handler: dispatches
template/v3/v3.8/dirs/etc/sagemaker-inference-server/tornado_server/server.py - Inference server → Serving errors: raises
template/v3/v3.8/dirs/etc/sagemaker-inference-server/tornado_server/server.py - Inference server → Inference logging: logs via
template/v3/v3.8/dirs/etc/sagemaker-inference-server/tornado_server/server.py - Async handler → Model handler: invokes
template/v3/v3.8/dirs/etc/sagemaker-inference-server/tornado_server/async_handler.py - Sync handler → Model handler: invokes
template/v3/v3.8/dirs/etc/sagemaker-inference-server/tornado_server/sync_handler.py - Async handler → SageMaker endpoint: serves routes
template/v3/v3.8/dirs/etc/sagemaker-inference-server/tornado_server/async_handler.py - Sync handler → SageMaker endpoint: serves routes
template/v3/v3.8/dirs/etc/sagemaker-inference-server/tornado_server/sync_handler.py - Report CLI → CLI configuration: imports
src/main.py - Report CLI → Package reports: dispatches
src/main.py - CPU/GPU images → Amazon Q Chat: includes
README.md - CPU/GPU images → Workflow client: includes