Skip to content

Latest commit

 

History

History
232 lines (154 loc) · 8.31 KB

README.md

File metadata and controls

232 lines (154 loc) · 8.31 KB
Raw

Contents of this Document

Prerequisites

First things first

1. If you have not clone torchserve source then:
git clone https://github.com/pytorch/serve.git
2. cd serve/docker

Create TorchServe docker image

For creating CPU based image :

DOCKER_BUILDKIT=1 docker build --file Dockerfile -t torchserve:latest .

For creating GPU based image with the latest CUDA version PyTorch supports (ex. CUDA 10.2 as of Oct 2020):

DOCKER_BUILDKIT=1 docker build --file Dockerfile --build-arg BASE_IMAGE=nvidia/cuda:10.2-cudnn7-runtime-ubuntu18.04 -t torchserve:latest .

For creating GPU based image with older CUDA versions (ex. CUDA 10.1), make sure that the --build-arg CUDA_VERSION=<version> is specified. The version is in the format "cuda92", "cuda101":

DOCKER_BUILDKIT=1 docker build --file Dockerfile --build-arg BASE_IMAGE=nvidia/cuda:10.1-cudnn7-runtime-ubuntu18.04 --build-arg CUDA_VERSION=cu101 -t torchserve:latest .

Start a container with a TorchServe image

The following examples will start the container with 8080/81 port exposed to outer-world/localhost.

Start CPU container

For the latest version, you can use the latest tag:

docker run --rm -it -p 8080:8080 -p 8081:8081 pytorch/torchserve:latest

For specific versions you can pass in the specific tag to use (ex: pytorch/torchserve:0.1.1-cpu):

docker run --rm -it -p 8080:8080 -p 8081:8081 pytorch/torchserve:0.1.1-cpu

Start GPU container

For GPU latest image with gpu devices 1 and 2:

docker run --rm -it --gpus '"device=1,2"' -p 8080:8080 -p 8081:8081 pytorch/torchserve:latest-gpu

For specific versions you can pass in the specific tag to use (ex: 0.1.1-cuda10.1-cudnn7-runtime):

docker run --rm -it --gpus all -p 8080:8080 -p 8081:8081 pytorch/torchserve:0.1.1-cuda10.1-cudnn7-runtime

For the latest version, you can use the latest-gpu tag:

docker run --rm -it --gpus all -p 8080:8080 -p 8081:8081 torchserve:gpu-latest

Accessing TorchServe APIs inside container

The TorchServe's inference and management APIs can be accessed on localhost over 8080 and 8081 ports respectively. Example :

curl http://localhost:8080/ping

Create TorchServe docker image from source

The following are examples on how to use the build_image.sh script to build Docker images from source to support CPU or GPU inference.

To build the TorchServe image for a CPU device using the master branch, use the following command:

./build_image.sh

Alternatively, you can use following direct command, (assuming you have followed steps in Clone serve source)-

For cpu -

DOCKER_BUILDKIT=1 docker build --file Dockerfile.dev -t torchserve:dev .

For gpu -

DOCKER_BUILDKIT=1 docker build --file Dockerfile.dev -t torchserve:dev --build-arg MACHINE_TYPE=gpu --build-arg BASE_IMAGE=nvidia/cuda:10.1-cudnn7-runtime-ubuntu18.04 .

To create a Docker image for a specific branch, use the following command:

./build_image.sh -b <branch_name>

To create a Docker image for a specific branch and specific tag, use the following command:

./build_image.sh -b <branch_name> -t <tagname:latest>

To create a Docker image for a GPU device, use the following command:

./build_image.sh --gpu

To create a Docker image for a GPU device with a specific branch, use following command:

./build_image.sh -b <branch_name> --gpu

To create a Docker image for a GPU device with Cuda 10.1, use following command:

./build_image.sh --gpu --cudaversion cuda101

To run your TorchServe Docker image and start TorchServe inside the container with a pre-registered resnet-18 image classification model, use the following command:

./start.sh

For GPU run the following command:

./start.sh --gpu

For GPU with specific GPU device ids run the following command:

./start.sh --gpu_devices 1,2,3

Alternatively, you can use direct commands describe in Start a container with a TorchServe image above for cpu and gpu by changing image name

Create torch-model-archiver from container

To create mar [model archive] file for torchserve deployment, you can use following steps

  1. Start container by sharing your local model-store/any directory containing custom/example mar contents as well as model-store directory (if not there, create it)
docker run --rm -it -p 8080:8080 -p 8081:8081 --name mar -v $(pwd)/model-store:/home/model-server/model-store -v $(pwd)/examples:/home/model-server/examples  torchserve:latest
  1. List your container or skip this if you know cotainer name
docker ps
  1. Bind and get the bash prompt of running container
docker exec -it <container_name> /bin/bash

You will be landing at /home/model-server/.

  1. Download the model weights if you have not done so already (they are not part of the repo)
curl -o /home/model-server/examples/image_classifier/densenet161-8d451a50.pth https://download.pytorch.org/models/densenet161-8d451a50.pth
  1. Now Execute torch-model-archiver command e.g.
torch-model-archiver --model-name densenet161 --version 1.0 --model-file /home/model-server/examples/image_classifier/densenet_161/model.py --serialized-file /home/model-server/examples/image_classifier/densenet161-8d451a50.pth --export-path /home/model-server/model-store --extra-files /home/model-server/examples/image_classifier/index_to_name.json --handler image_classifier

Refer torch-model-archiver for details.

  1. desnet161.mar file should be present at /home/model-server/model-store

Running TorchServe in a Production Docker Environment.

You may want to consider the following aspects / docker options when deploying torchserve in Production with Docker.

  • Shared Memory Size

    • shm-size - The shm-size parameter allows you to specify the shared memory that a container can use. It enables memory-intensive containers to run faster by giving more access to allocated memory.

  • User Limits for System Resources

    • --ulimit memlock=-1 : Maximum locked-in-memory address space.
    • --ulimit stack : Linux stack size

    The current ulimit values can be viewed by executing ulimit -a. A more exhaustive set of options for resource constraining can be found in the Docker Documentation here, here and here

  • Exposing specific ports / volumes between the host & docker env.

    • -p8080:8080 -p8081:8081 TorchServe uses default ports 8080 / 8081 for inference & management APIs. You may want to expose these ports to the host for HTTP Requests between Docker & Host.
    • The model store is passed to torchserve with the --model-store option. You may want to consider using a shared volume if you prefer pre populating models in model-store directory.

For example,

docker run --rm --shm-size=1g \
        --ulimit memlock=-1 \
        --ulimit stack=67108864 \
        -p8080:8080 \
        -p8081:8081 \
        --mount type=bind,source=/path/to/model/store,target=/tmp/models <container> torchserve --model-store=/tmp/models