Deploying-Python-Server-on-AWS

Steps/commands on how to easily deploy a python server on your AWS Ubuntu machine using PUTTY

Dependencies required:

Python3

sudo apt-get upgrade
sudo apt-get update
sudo apt-get install python3

Pip

sudo apt-get install python3-setuptools
sudo apt-get install python3 pip
pip3 install --upgrade pip

Perform the last step only when an upgrade is required

TensorFlow

sudo apt-get install python3-pip python3-dev
pip3 install tensorflow

OR

sudo pip3 instal ---upgrade https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-1.2.1-cp27-none-linux_x86_64.whl

For downloading any other dependency

For example we require Google-auth

sudo pip3 install google-auth

Set 'starter.sh' file

Call your run.py file through a separate text file with an arbitrary name such as starter.sh and save it in your projects git repository

Add #!/bin/sh as the first line of your code in starter.sh file

Try executing your starter.sh file by sudo starter.sh to make sure there are no errors in execution steps

PUTTY

Start your session on PUTTY

Clone your git repository

git clone 'link to repository'

Automatic startup of run.py file

To make sure your run.py is automatically run whenever the AWS machine restarts you'll have to add the path to your rc.local script.

Following are the steps:

Provide your /etc/rc.local script with the full path and name of your createrd scrit after the 'sh' command - (your etc folder will be in the outermost directory)

Use vim to edit your rc.local script:

sudo vim rc.local

Put the following line before the last line of code (exit 0) at the end of the /etc/rc.local script

sh '/path/to/your/script/starter.sh'

Check that the first line of etc/rc.local script is:

#!/bin/sh -e

Go to your outermost directory

Check that it has 'etc' and 'home' folders:

ls

The above command displays all folders in the current command

Make your /etc/rc.local script executable in case it is not already executable by:

sudo chown root /etc/rc.local
sudo chmod 755 /etc/rc.local

Check everything works fine by executing:

sudo /etc/init.d/rc.local start

This should start the starter.sh file and hence, the run.py file being called in starter.sh file

That is, your run.py file should start running, it should be executed by this step

Check that the process has actually started:

ps -ef

This should display all the processes currently running on the AWS machine

You can also check the running instances of only run.py process:

ps ax | grep run.py

To check program output, open the nohup.out file:

sudo tail -f nohup.out

Killing a process

To kill a process, use the process number:

ps ax | grep run.py

This will give the process number of run.py process

sudo kill process-number

This will kill the process even if it is running in the background as a nohup process