https://github.com/jgontrum/docker-tensorflow-tutorial
Small introduction to Docker with the example of a Tensorflow use case.
https://github.com/jgontrum/docker-tensorflow-tutorial
Last synced: about 1 month ago
JSON representation
Small introduction to Docker with the example of a Tensorflow use case.
- Host: GitHub
- URL: https://github.com/jgontrum/docker-tensorflow-tutorial
- Owner: jgontrum
- Created: 2016-09-02T22:17:15.000Z (over 8 years ago)
- Default Branch: master
- Last Pushed: 2016-09-02T22:26:52.000Z (over 8 years ago)
- Last Synced: 2025-01-27T08:27:29.455Z (3 months ago)
- Language: Makefile
- Homepage:
- Size: 5.86 KB
- Stars: 1
- Watchers: 3
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Docker Tensorflow Tutorial
## Installation
- Set up a vanilla Debian server.
- Add your SSH key to the _~/.ssh/authorized_keys_ file of the root user on the server.
- Enable IP forwarding, otherwise you'll have difficulties communicating with your Docker containers.
- Install and start Docker
- Configure port forwarding
### IP Forwarding
Uncomment the line
```
#net.ipv4.ip_forward=1
```
in the _/etc/sysctl.conf_ file:
```
vim /etc/sysctl.conf
:%s/#net.ipv4.ip_forward=1/net.ipv4.ip_forward=1/g
:x
sysctl -p
```
### Install Docker
Set the variable 'REMOTE' in the Makefile to the IP of your server and run 'make' on your local machine.
### Port Forwarding
Assuming you need to access the ports 8888 and 6006 for TensorBoard Jupyter, forward them to you local machine.
#### Solution 1: Config
~/.ssh/config [local machine]
```
Host tensorserver
User root
HostName 123.234.345.456
LocalForward 8888 127.0.0.1:8888
LocalForward 6006 127.0.0.1:6006
```
Open the ports with
```
ssh tensorserver
```
or without having to keep the ssh shell open:
```
ssh -fN tensorserver
```
#### Solution 2: Full SSH Command
```
ssh -L 8888:localhost:8888 [email protected]
```
## Use Docker
### Start an image from Docker Hub
On the server run:
```
docker pull tensorflow/tensorflow:latest
```
to download the image.
Then start it with:
```
docker run -it -p 127.0.0.1:8888:8888 --name mycontainer tensorflow/tensorflow:latest
```
CMD | Explanation
--------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------
-it | Interactive mode, you can control the program in the container
-d | Detached mode, the container runs in the background
-p | Make the following port accessable to the host system
127.0.0.1:8888:8888 | HOST_IP:HOST_PORT:CONTAINER_PORT
127.0.0.1 | The port is only open locally, e.g. for the port forwarding
0.0.0.0 | The port ist open to the whole internet (default)
tensorflow/tensorflow | Name of the image to run. USER/IMAGE:TAG
--name mycontainer | Give this container the name 'mycontainer'. Naming your containers makes it more easy to stop and restart them once they run in detached and not the interactive mode.
### Build and run your own Docker image
Copy this folder to the server to '/root' and build a new container:
```
docker build -t yourname/yourapp .
```
Share the folder '/root/shared' with the container, so it can run the python script 'run.py' in the folder.
```
docker run -it -v /root/shared:/usr/shared yourname/yourapp
```
The flag '-v' shared a local directory to a path in the docker container: '-v LOCAL:CONTAINER'. Once started, this container runs the file 'run.py' in your local folder /root/shared.
#### The easy way
Just set the variable 'REMOTE_PATH' in the Makefile to place this folder should be copied on the server and run
```
make deploy
```
## Cheat Sheet
CMD | Explanation
-------------- | -----------------------------------------------------
docker ps | Show all the running containers
docker ps -a | Show all containers, including those that are stopped
docker stop ID | Stop the container with the given ID
docker kill ID | Kill the container with the given ID
docker rm ID | Remove the container with the given ID
docker images | List all local images
docker rmi ID | Remove the image with the given ID