Skip to content

Latest commit

 

History

History
541 lines (368 loc) · 16 KB

lab3.md

File metadata and controls

541 lines (368 loc) · 16 KB
Raw

Lab 3 - Introduction to functions

Before starting this lab, create a new folder for your files:

$ mkdir -p lab3 \
   && cd lab3

Creating a new function

There are two ways to create a new function:

  • scaffold a function using a built-in or community code template (default)
  • take an existing binary and use it as your function (advanced)

Scaffold or generate a new function

Before creating a new function from a template make sure you pull the templates from GitHub:

$ faas-cli template pull

Fetch templates from repository: https://github.com/openfaas/templates.git
 Attempting to expand templates from https://github.com/openfaas/templates.git
 Fetched 11 template(s) : [csharp dockerfile go go-armhf node node-arm64 node-armhf python python-armhf python3 ruby]

After that, to find out which languages are available type in:

$ faas-cli new --list
Languages available as templates:
- csharp
- csharp-armhf
- dockerfile
- dockerfile-armhf
- go
- go-armhf
- java12
- java8
- node
- node-arm64
- node-armhf
- php7
- python
- python-armhf
- python3
- python3-armhf
- ruby

Or alternatively create a folder containing a Dockerfile, then pick the "Dockerfile" lang type in your YAML file.

At this point you can create a new function for Python, Python 3, Ruby, Go, Node, CSharp etc.

  • A note on our examples

All of our examples for this workshop have been thoroughly tested by the OpenFaaS community with Python 3, but should be compatible with Python 2.7 also.

If you'd prefer to use Python 2.7 instead of Python 3 then swap faas-cli new --lang python3 for faas-cli new --lang python.

Hello world in Python

We will create a hello-world function in Python, then move onto something that uses additional dependencies too.

  • Scaffold the function
$ faas-cli new --lang python3 hello-openfaas --prefix="<your-docker-username-here>"

The --prefix parameter will update image: value in hello-openfaas.yml with a prefix which should be your Docker Hub account. For OpenFaaS this is image: functions/hello-openfaas and the parameter will be --prefix="functions".

If you don't specify a prefix when you create the function then edit the YAML file after creating it.

This will create three files and a directory:

./hello-openfaas.yml
./hello-openfaas
./hello-openfaas/handler.py
./hello-openfaas/requirements.txt

The YAML (.yml) file is used to configure the CLI for building, pushing and deploying your function.

Note: Whenever you need to deploy a function on Kubernetes or on a remote OpenFaaS instance you must always push your function after building it. In this case you can also override the default gateway URL of 127.0.0.1:8080 with an environmental variable: export OPENFAAS_URL=127.0.0.1:31112.

Here's the contents of the YAML file:

provider:
  name: openfaas
  gateway: http://127.0.0.1:8080

functions:
  hello-openfaas:
    lang: python3
    handler: ./hello-openfaas
    image: hello-openfaas
  • The name of the function is represented by the key under functions i.e. hello-openfaas
  • The language is represented by the lang field
  • The folder used to build from is called handler, this must be a folder not a file
  • The Docker image name to be used is under the field image

Remember that the gateway URL can be overriden in the YAML file (by editing the gateway: value under provider:) or on the CLI (by using --gateway or setting the OPENFAAS_URL environment variable).

Here is the contents of the handler.py file:

def handle(req):
    """handle a request to the function
    Args:
        req (str): request body
    """

    return req

This function will just return the input, so it's effectively an echo function.

Edit the message so it returns Hello OpenFaaS instead i.e.

    return "Hello OpenFaaS"

Any values returned to stdout will subsequently be returned to the calling program. Alternatively a print() statement could be employed which would exhibit a similar flow through to the calling program.

This is the local developer-workflow for functions:

$ faas-cli up -f hello-openfaas.yml

Note: Please make sure that you have logged in to docker registry with docker login command before running this command.

Note: faas-cli up command combines build, push and deploy commands of faas-cli in a single command.

Followed by invoking the function via the UI, CLI, curl or another application.

The function will always get a route, for example:

$OPENFAAS_URL/function/<function_name>
$OPENFAAS_URL/function/figlet
$OPENFAAS_URL/function/hello-openfaas

Pro-tip: if you rename your YAML file to stack.yml then you need not pass the -f flag to any of the commands.

Functions can be invoked via a GET or POST method only.

  • Invoke your function

Test out the function with faas-cli invoke, check faas-cli invoke --help for more options.

Example function: astronaut-finder

We'll create a function called astronaut-finder that pulls in a random name of someone in space aboard the International Space Station (ISS).

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

This will write three files for us:

./astronaut-finder/handler.py

The handler for the function - you get a req 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, such as requests or urllib

./astronaut-finder.yml

This file is used to manage the function - it has the name of the function, the Docker image and any other customisations needed.

  • Edit ./astronaut-finder/requirements.txt
requests

This tells the function it needs to use a third-party module named requests for accessing websites over HTTP.

  • 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:

{"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"}

Update 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, len(result["people"])-1)
    name = result["people"][index]["name"]

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

Note: in this example we do not make use of the parameter req but must keep it in the function's header.

Now build the function:

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

Tip: Try renaming astronaut-finder.yml to stack.yml and calling just faas-cli build. stack.yml is the default file-name for the CLI.

Push the function:

$ faas-cli push -f ./astronaut-finder.yml

Deploy the function:

$ faas-cli deploy -f ./astronaut-finder.yml

Invoke the function

$ echo | faas-cli invoke astronaut-finder
Anton Shkaplerov is in space

$ echo | faas-cli invoke astronaut-finder
Joe Acaba is in space

Troubleshooting: find the container's logs

You can find out high-level information on every invocation of your function via the container's logs:

Docker Swarm

$ docker service logs -f astronaut-finder
astronaut-finder.1.1e1ujtsijf6b@nuc    | 2018/02/21 14:53:25 Forking fprocess.
astronaut-finder.1.1e1ujtsijf6b@nuc    | 2018/02/21 14:53:26 Wrote 18 Bytes - Duration: 0.063269 seconds

Kubernetes

$ kubectl logs deployment/astronaut-finder -n openfaas-fn

Troubleshooting: verbose output with write_debug

Let's turn on verbose output for your function. This is turned-off by default so that we do not flood your function's logs with data - that is especially important when working with binary data which makes no sense in the logs.

This is the standard YAML configuration:

provider:
  name: openfaas
  gateway: http://127.0.0.1:8080

functions:
  astronaut-finder:
    lang: python3
    handler: ./astronaut-finder
    image: astronaut-finder

Edit your YAML file for the function and add an "environment" section.

  astronaut-finder:
    lang: python3
    handler: ./astronaut-finder
    image: astronaut-finder
    environment:
      write_debug: true

Now deploy your function again with faas-cli deploy -f ./astronaut-finder.yml.

Invoke the function and then checkout the logs again to view the function responses:

Docker Swarm

$ docker service logs -f astronaut-finder

astronaut-finder.1.tp6k14i8kf6s   | 2019/04/25 18:28:36 Forking fprocess.
astronaut-finder.1.tp6k14i8kf6s   | 2019/04/25 18:28:36 Query  
astronaut-finder.1.tp6k14i8kf6s   | 2019/04/25 18:28:36 Path  /
astronaut-finder.1.tp6k14i8kf6s   | 2019/04/25 18:28:37 Duration: 1.128897 seconds
astronaut-finder.1.tp6k14i8kf6s   | Alexey Ovchinin is in space

Kubernetes

$ kubectl logs deployment/astronaut-finder -n openfaas-fn

Managing multiple functions

The YAML file for the CLI allows functions to be grouped together into stacks, this is helpful when working with a set of related functions.

To see how this works generate two functions:

$ faas-cli new --lang python3 first

For the second function use the --append flag:

$ faas-cli new --lang python3 second --append=./first.yml

For convenience let's rename first.yml to example.yml.

$ mv first.yml example.yml

Now look at the file:

provider:
  name: openfaas
  gateway: http://127.0.0.1:8080

functions:
  first:
    lang: python3
    handler: ./first
    image: first
  second:
    lang: python3
    handler: ./second
    image: second

Here are several flags that help when working with a stack of functions:

  • Build in parallel:
$ faas-cli build -f ./example.yml --parallel=2
  • Build / push only one function:
$ faas-cli build -f ./example.yml --filter=second

Take a few moments to explore the options for build/push and deploy.

  • faas-cli build --help
  • faas-cli push --help
  • faas-cli deploy --help

To run faas-cli build && faas-cli push && faas-cli deploy together, use faas-cli up instead.

Pro-tip: stack.yml is the default name the faas-cli will look for if you don't want to pass a -f parameter.

You can also deploy remote function stack (yaml) files over HTTP(s) using faas-cli deploy -f https://.....

Custom templates

If you have your own set of forked or custom templates, then you can pull them down for use with the CLI.

Here's an example of fetching a Python 3 template which uses Debian Linux.

Pull the template using the git URL:

$ faas-cli template pull https://github.com/openfaas-incubator/python3-debian

Now type in: faas-cli new --list

$ faas-cli new --list | grep python
- python
- python3
- python3-debian

These new templates are saved in your current working directory in ./templates/.

Custom templates: Template Store

The Template Store is a similar concept to the Function Store, it enables users to collaborate by sharing their templates. The template store also means that you don't have to remember any URLs for making use of your favourite community or project templates.

You can Search and discover templates using the following two commands:

$ faas-cli template store list
$ faas-cli template store list -v

NAME                     SOURCE             DESCRIPTION
csharp                   openfaas           Official C# template
dockerfile               openfaas           Official Dockerfile template
go                       openfaas           Official Golang template
...

To get more details you can use the --verbose flag, or the describe command.

Let's get a template for use with Node.js which is powered by the Express.js framework.

$ faas-cli template store describe node10-express

Name:              node10-express
Platform:          x86_64
Language:          NodeJS
Source:            openfaas-incubator
Description:       NodeJS 10 Express template
Repository:        https://github.com/openfaas-incubator/node10-express-template
Official Template: true

Pull the template down:

$ faas-cli template store pull node10-express

You can now type in faas-cli new --lang node10-express.

See also: Function & Template Store

Variable Substitution in YAML File (optional exercise)

The .yml file used to configure the CLI is capable of variable substitution so that you are able to use the same .yml file for multiple configurations.

One example of where this can be useful is when there are different registries for development and production images. You can use the variable substitution so that local and test environments use the default account, and the CI server can be configured to use the production account.

This is provided by the envsubst library. Follow the link to see examples of supported variables

Edit your astronaut-finder.yml to match the following:

  astronaut-finder:
    lang: python3
    handler: ./astronaut-finder
    image: ${DOCKER_USER:-development}/astronaut-finder
    environment:
      write_debug: true

You'll notice the image property has been updated to include a variable definition (DOCKER_USER). That value will be replaced with the value of the environment variable with the same name. If the environment variable is not present, or is empty, the default value (development) will be used.

The variable will be replaced with the value throughout the file. So, if you have several functions in your .yml file, all references to the DOCKER_USER variable will be replaced with the value of that environment variable

Run the following command and observe the output:

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

The output should show that the image built is labeled as development/astronaut-finder:latest

Now, set the environment variable to your Docker Hub account name (for the example, we'll use the OpenFaaS "functions" account)

export DOCKER_USER=functions

Run the same build command as before and observe the output:

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

The output should now show that the image was built with the updated label functions/astronaut-finder:latest

Custom binaries as functions (optional exercise)

Custom binaries or containers can be used as functions, but most of the time using the language templates should cover all the most common scenarios.

To use a custom binary or Dockerfile create a new function using the dockerfile language:

$ faas-cli new --lang dockerfile sorter --prefix="<your-docker-username-here>"

You'll see a folder created named sorter and sorter.yml.

Edit sorter/Dockerfile and update the line which sets the fprocess. Let's change it to the built-in bash command of sort. We can use this to sort a list of strings in alphanumeric order.

ENV fprocess="sort"

Now build, push and deploy the function:

$ faas-cli up -f sorter.yml

Now invoke the function through the UI or via the CLI:

$ echo -n '
elephant
zebra
horse
aardvark
monkey'| faas-cli invoke sorter

aardvark
elephant
horse
monkey
zebra

In the example we used sort from BusyBox which is built into the function. There are other useful commands such as sha512sum and even a bash or shell script, but you are not limited to these built-in commands. Any binary or existing container can be made a serverless function by adding the OpenFaaS function watchdog.

Tip: did you know that OpenFaaS supports Windows binaries too? Like C#, VB or PowerShell?

Now move onto Lab 4