Working with functions

You can get a local environment ready using Vagrant for an automatically setting up a machine on Virtualbox. To download Vagrant follow this link where you can find the complete list based on Os System:

Vagrant installation - Windows

Download the following file and install it:

Vagrant installation - Linux

Download the package and extract it just pasting the following commands:

wget https://releases.hashicorp.com/vagrant/2.2.5/vagrant_2.2.5_linux_amd64.zip
unzip vagrant_2.2.5_linux_amd64.zip

You can also use Homebrew like this:

brew cask install vagrant

If you prefer, you can use your favorite package manager. For example, in Ubuntu you can type:

sudo apt install virtualbox vagrant

Vagrant installation - MacOS

Download the dmg file and install it:

If you have Homebrew installed just type this in the command line:

brew cask install vagrant

Install Vagrant project

To be operative with the current project you have to download with git or extract the zip file.

Open your command line and paste the following commands:

git clone https://github.com/Cloud-PG/SOSC2019.git
cd SOSC2019

Now set up the vagrant environment using the following command:

./vagrant up
# Or if you have vagrant executable in your PATH
vagrant up

NOTE: This may take few minutes, depending on network and your computer performance

And then, log into the created machine:

./vagrant ssh
# Or if you have vagrant executable in your PATH
vagrant ssh

Remember also to authenticate with OpenFaaS cli before continuing:

export OPENFAAS_URL=http://127.0.0.1:31112
cat gateway-password.txt | faas-cli login --password-stdin

NOTE: Vagrant and Virtualbox are required on the machine of course. If you don't have them check the previous steps to install the packages

Some additional machines have been prepared for the school participants. You can find hot to access to your machine here

Using example functions and the OpenFaaS UI

Now you can go to http://localhost:31112/ui/ and, using the password in gateway_password.txt with user admin, you should be able to log in. To see your password just type:

cat gateway-password.txt

You will have a page like that as result after the login:

OpenFaaS UI

Let's start playing with some example functions. For instance, you can instantiate a function the face-detection of an online image just clicking on Deploy new function, searching for opencv and installing face-detect with OpenCV (button Deploy).

New function

Face detect function

Now a new tab should appear with the function name selected.

Function dashboard

From there you can check the status and also try to invoke the function from the UI. For instance, as soon as the status of the function is ready, let's try to put an url with a jpg image in the request body field and then press invoke.

Let's try the two below for example:

https://parismatch.be/app/uploads/2018/04/Macaca_nigra_self-portrait_large-e1524567086123-1100x715.jpg
http://thedreamwithinpictures.com/wp-content/uploads/2013/05/c3a89__montage_2048-copy.jpg

For the second one you will have the following result:

New function

New function

The list of all available functions in the store is also available from CLI using the following command from the vagrant machine already created:

faas-cli store list

Deployment of a Python function (from OpenFaaS workshop)

Do everyone have a docker account?

mkdir astronaut-finder
cd astronaut-finder

faas-cli new --lang python3 astronaut-finder --prefix="<your-docker-username-here>"

Function fundamentals

The previous command will write three files for us:

./astronaut-finder/handler.py

This is the handler for the function.

NOTE: an handler get a request object with the raw request and can print the result of the function to the console.

./astronaut-finder/requirements.txt

Use this file to list any pip modules you want to install so, to manage your Python requirements, such as requests or urllib

./astronaut-finder.yml This file is used to manage the function:

NOTE: it has the name of the function, the Docker image and any other customizations needed.

Edit ./astronaut-finder/requirements.txt and add the following dependency:

requests

Write the function's code

We'll be pulling in data from http://api.open-notify.org/astros.json

Here's an example of the result we will have from that url:

{
  "number": 6,
  "people": [
    {
      "craft": "ISS",
      "name": "Alexander Misurkin"
    },
    {
      "craft": "ISS",
      "name": "Mark Vande Hei"
    },
    {
      "craft": "ISS",
      "name": "Joe Acaba"
    },
    {
      "craft": "ISS",
      "name": "Anton Shkaplerov"
    },
    {
      "craft": "ISS",
      "name": "Scott Tingle"
    },
    {
      "craft": "ISS",
      "name": "Norishige Kanai"
    }
  ],
  "message": "success"
}

Let's write an handler that gets for us that result. You have to edit handler.py:

import requests
import random

def handle(req):
    r = requests.get("http://api.open-notify.org/astros.json")
    result = r.json()
    index = random.randint(0, result["number"]-1)
    name = result["people"][index]["name"]

    return "%s is in space" % (name)

Deploy the function

First, build it:

faas-cli build -f ./astronaut-finder.yml

Push the function:

docker login
faas-cli push -f ./astronaut-finder.yml

Deploy the function:

export OPENFAAS_URL=http://127.0.0.1:31112
cat  $HOME/gateway-password.txt | faas-cli login --password-stdin
faas-cli deploy -f ./astronaut-finder.yml

And now, just wait a bit for the function to be in Ready state. Check from cli with the following command:

faas-cli describe astronaut-finder | grep Status

You will have a result like this if everythings is up and running:

Status:              Ready

and then try to invoke it from command line:

echo | faas-cli invoke astronaut-finder

You will receive this as response a random name of a docker user, like this:

Anton Shkaplerov is in space

You can also use the http endpoint:

curl http://localhost:31112/function/astronaut-finder

Or try it from the dashboard, just clicking to invoke and see the result in response body:

Astronaut dashboard

Homework

  1. Try to create a function for serving your ML model (you can also make use of: https://github.com/alexellis/tensorflow-serving-openfaas )
  2. Create a function in a different language if you know any