Skip to content

Latest commit

 

History

History
134 lines (112 loc) · 6.49 KB

compose.md

File metadata and controls

134 lines (112 loc) · 6.49 KB
Raw

Customize Triton Container

Two Docker images are available from NVIDIA GPU Cloud (NGC) that make it possible to easily construct customized versions of Triton. By customizing Triton you can significantly reduce the size of the Triton image by removing functionality that you don't require.

Currently the customization is limited as described below but future releases will increase the amount of customization that is available. It is also possible to build Triton from source to get more exact customization.

Use the compose.py script

The compose.py script can be found in the server repository. Simply clone the repository and run compose.py to create a custom container. Note: Created container version will depend on the branch that was cloned. For example branch r21.09 should be used to create a image based on the NGC 21.09 Triton release.

compose.py provides --backend, --repoagent options that allow you to specify which backends and repository agents to include in the custom image. The --enable-gpu flag indicates that you want to create an image that supports NVIDIA GPUs. For example, the following creates a new docker image that contains only the TensorFlow 1 and TensorFlow 2 backends and the checksum repository agent.

Example:

python3 compose.py --backend tensorflow1 --backend tensorflow2 --repoagent checksum --enable-gpu

will provide a container tritonserver locally. You can access the container with

$ docker run -it tritonserver:latest

Note: If compose.py is run on release versions r21.09 and earlier, the resulting container will have DCGM version 2.2.3 installed. This may result in different GPU statistic reporting behavior.

Compose a specific version of Triton

compose.py requires two containers: a min container which is the base the compose container is built from and a full container from which the script will extract components. The version of the min and full container is determined by the branch of Triton compose.py is on. For example, running

python3 compose.py --backend tensorflow1 --repoagent checksum --enable-gpu

on branch r21.09 pulls:

  • min container nvcr.io/nvidia/tritonserver:21.09-py3-min
  • full container nvcr.io/nvidia/tritonserver:21.09-py3

Alternatively, users can specify the version of Triton container to pull from any branch by either:

  1. Adding flag --container-version <container version> to branch
python3 compose.py --backend tensorflow1 --repoagent checksum --container-version 21.09 --enable-gpu
  1. Specifying --image min,<min container image name> --image full,<full container image name>. The user is responsible for specifying compatible min and full containers.
python3 compose.py --backend tensorflow1 --repoagent checksum --image min,nvcr.io/nvidia/tritonserver:21.09-py3-min --image full,nvcr.io/nvidia/tritonserver:21.09-py3 --enable-gpu

Method 1 and 2 will result in the same composed container. Furthermore, --image flag overrides the --container-version flag when both are specified.

CPU only container composition

To compose a container that is built for only cpu usage, simply remove the --enable-gpu flag when running compose.py.

This will build a container using ubuntu:20.04 docker as the min container and nvcr.io/nvidia/tritonserver:<upstream-container-version>-cpu-only-py3 as the full container. Note:

  1. When composing a CPU only container, both min and full containers should be built for CPU only and not have CUDA installed.
  2. CPU only containers are only available for Triton versions > 21.09
  3. CPU only "full" containers are build with less backends then the GPU enabled containers. Currently supported backends are onnxruntime, openvino and python.

Build it yourself

If you would like to do what compose.py is doing under the hood yourself, you can run compose.py with the --dry-run option and then modify the Dockerfile.compose file to satisfy your needs.

Triton with Unsupported and Custom Backends

You can create and build your own Triton backend. The result of that build should be a directory containing your backend shared library and any additional files required by the backend. Assuming your backend is called "mybackend" and that the directory is "./mybackend", adding the following to the Dockerfile compose.py created will create a Triton image that contains all the supported Triton backends plus your custom backend.

COPY ./mybackend /opt/tritonserver/backends/mybackend

You also need to install any additional dependencies required by your backend as part of the Dockerfile. Then use Docker to create the image.

$ docker build -t tritonserver_custom -f Dockerfile.compose .