kubeflow-labs/1-docker

Docker

Summary

In this section you will learn about:

• Running Docker locally
• Basics of Docker
• Containerizing a simple application
• Building and Pushing an Image

Basics of Docker and Containers

Docker has a very well structured six-part tutorial. While for this workshop you don't need to go through all of them, part 1 and 2 are required:

• Get Started, Part 1: Orientation and setup
• Get Started, Part 2: Containers

By the end of Part 2, you should have a simple container up and running, and understand the basic concepts of a container.

Additional Important Docker Command

Here a few other docker commands that are important to be aware of for the rest of this workshop:

1. docker ps

   The docker ps command allows to list the status of the containers.

   A container could be either stopped or running. When it finishes to execute the process, it will stop.

   For example if you run the command docker run -it ubuntu hostname this will :

   • Pull the official ubuntu image from the registry
   • Start the container in the interactive mode -it
   • Execute the command : hostname
   • Stop

   $ docker run -it ubuntu hostname
   0d0af5005fc7
   

   If you run the command docker ps -a you should see :

    $ docker ps -a
   CONTAINER ID        IMAGE               COMMAND             CREATED             STATUS                          PORTS               NAMES
   0d0af5005fc7        ubuntu              "hostname"          58 seconds ago      Exited (0) About a minute ago                       gifted_darwin
   

   The -a allows you to list all the containers, not just the one that is running

   We can notice a few things here such as :

   • The status Exited... for our container
   • The name gifted_darwin randomly generated for our container, we can specify a custom one using the command --name explained in the previous section.
   • We can re-execute our container using the command docker start gifted_darwin
   • We can run the command docker run -it ubuntu hostname again and do a docker ps -a; we should see how two containers exited.

2. docker logs

   The docker logs allow to fetch the console output from inside the container

   From our previous example, we can run docker logs gifted_darwin

   $ docker logs gifted_darwin
   0d0af5005fc7
   

   You can also stream the logs using the command -f and print in real time in your console the stdout of your container.

3. docker rm

   The docker rm command allows to remove a container.

   From the previous example, we can see that we have a container listed as exited in our environment, or maybe more if we run the same command docker run -it ubuntu hostname multiple times. If we want to do some cleaning and remove those executions from our environment we can use the command docker rm

   $ docker rm gifted_darwin
   gifted_darwin
   

   You can either specify the CONTAINER ID or the NAME of the container to refer to it

4. docker images

   This command allows us to list all the base images available in the environment.

   $ docker images
   REPOSITORY                                      TAG                 IMAGE ID            CREATED             SIZE
   ubuntu                                          latest              20c44cd7596f        2 days ago          123MB
   example-scratch                                 latest              32ff7b65f567        5 days ago          30.7MB
   node                                            8.9.1-slim          a6bb2cc1118f        11 days ago         230MB
   buildpack-deps                                  xenial              a27b6a8abd1c        2 weeks ago         644MB
   

   You can manage your images by removing them using docker rmi IMAGENAME or pulling a new one with docker pull IMAGENAME

Containerizing a TensorFlow model

Now that we understand the basics of Docker, let's containerize our first TensorFlow model that we will reuse in the following modules.
Our first model will be a very simple MNIST classifier. You can see the source code in ./src/main.py.
As you can see there is nothing specific to containers in this code, you can run this script directly on your laptop or on a VM.

Now, to have this run in a container, we need to build an image containing this code and it's dependencies.
As you saw in the tutorial, we will use a Dockerfile to do this.

Here is the (very simple) Dockerfile that we are going to use for this model (located in ./src/Dockerfile):

FROM tensorflow/tensorflow:1.4.0
COPY main.py /app/main.py

ENTRYPOINT ["python", "/app/main.py"]

As you can see, we are not building a new image from scratch, instead we are using a base image from TensorFlow. Indeed, TensorFlow has a bunch of base images that you can start with. You can see the full list here: https://hub.docker.com/r/tensorflow/tensorflow/tags/.

What is important to note is that different tags need to be used depending on if you want to use GPU or not.
For example, if you wanted to run your model with TensorFlow 1.4.0 and CPU only, you would use tensorflow/tensorflow:1.4.0.
If instead you wanted to use GPU, you would start from tensorflow/tensorflow:1.4.0-gpu.

The two other instructions are pretty straightforward, first we copy our script into the container, and then we set this script as the entry point for our container, so that any argument passed to our container would actually get passed to our script.

Building the image

If you don't already have a Docker account, see Log in with your Docker ID.

The next step is to build our image to be able to run it using docker. For that, we will use the command docker build.

From the ./src repository, we can build the image with

docker build -t ${DOCKER_USERNAME}/tf-mnist .

Reminder: the -t argument allows to tag the image with a specific name.

${DOCKER_USERNAME} should be your Docker username that you use to connect to Docker Hub.

The output from this command should look like this:

Sending build context to Docker daemon  11.26kB
Step 1/3 : FROM tensorflow/tensorflow:1.4.0
 ---> a61a91cc0d1b
Step 2/3 : COPY main.py /app/main.py
 ---> b264d6e9a5ef
Removing intermediate container fe8128425296
Step 3/3 : ENTRYPOINT python /app/main.py
 ---> Running in 7acb7aac7a9f
 ---> 92c7ed17916b
Removing intermediate container 7acb7aac7a9f
Successfully built 92c7ed17916b
Successfully tagged wbuchwalter/tf-mnist:latest

Let's analyse this image full name (wbuchwalter/tf-mnist:latest):

• wbuchwalter is the name of my repository, this is where we can find the image. This will be different for you (same as your docker hub username).
• tf-mnist is the name of the image itself
• latest is the tag. latest is the default tag if you don't specify any. Tags are usually used to denote different versions or flavors of a same image. For example you could have a tag v1 and v2 to denote different versions, or cpu and gpu to denote what hardware it can run on.

When you have the successfully built message, you should now be able to see if your image is locally available with the command docker images described earlier.

Running the image

Now we can try to run it locally using the docker run command.
By default the model will run 1000 training steps which can take a few minutes on a laptop. Let's reduce this number to 100 with the --max_steps argument.

docker run -it ${DOCKER_USERNAME}/tf-mnist --max_steps 100

If everything is okay you should see the model training:

Successfully downloaded train-images-idx3-ubyte.gz 9912422 bytes.
Extracting /tmp/tensorflow/input_data/train-images-idx3-ubyte.gz
Successfully downloaded train-labels-idx1-ubyte.gz 28881 bytes.
Extracting /tmp/tensorflow/input_data/train-labels-idx1-ubyte.gz
Successfully downloaded t10k-images-idx3-ubyte.gz 1648877 bytes.
Extracting /tmp/tensorflow/input_data/t10k-images-idx3-ubyte.gz
Successfully downloaded t10k-labels-idx1-ubyte.gz 4542 bytes.
Extracting /tmp/tensorflow/input_data/t10k-labels-idx1-ubyte.gz
2017-11-29 18:32:41.992194: I tensorflow/core/platform/cpu_feature_guard.cc:137] Your CPU supports instructions that this TensorFlow binary was not compiled to use: SSE4.1 SSE4.2 AVX AVX2 FMA
Accuracy at step 0: 0.1292
Accuracy at step 10: 0.7198
Accuracy at step 20: 0.834
Accuracy at step 30: 0.8698
Accuracy at step 40: 0.8783
Accuracy at step 50: 0.8968
Accuracy at step 60: 0.9023
Accuracy at step 70: 0.9059
Accuracy at step 80: 0.9084
Accuracy at step 90: 0.9154
Adding run metadata for 99

You can kill the process and exit the container at any time with ctrl + c.

Running the image with the NVIDIA GPU of your machine (If you have one)

Currently, running docker containers with GPU is only supported on Linux.

First install nvidia-docker.

You also need to make sure the image you are going to use is optimized for GPU.
In our example you need to modify the Dockerfile to use a TensorFlow image built for GPU:

FROM tensorflow/tensorflow:1.4.0-gpu
COPY main.py /app/main.py

ENTRYPOINT ["python", "/app/main.py"]

Then simply rebuild the image with a new tag (you can use docker or nvidia-docker interchangeably for any command except run):

docker build -t ${DOCKER_USERNAME}/tf-mnist:gpu .

Finally run the container with nvidia-docker:

nvidia-docker run -it ${DOCKER_USERNAME}/tf-mnist:gpu

Note: If the command fails with Unknown runtime specified nvidia, follow the steps described here (Systemd drop-in file): https://github.com/nvidia/nvidia-container-runtime#docker-engine-setup

Publish the Image

Our image is now built and running locally, but what about sharing it to be able to use it from anywhere by anyone?
Most importantly we want to be able to reuse this image on the Kubernetes cluster we are going to create in module 2. So let's push our image to Docker Hub:

docker push ${DOCKER_USERNAME}/tf-mnist:gpu

If this command doesn't look familiar to you, make sure you went through part 1 and 2 of Docker's tutorial, and more precisely: Tutorial - Share your image

Useful Links

• What is Docker ?
• Docker for beginner

Next Step

2 - Kubernetes