Poor man's function as a service.
fx is a tool to help you do Function as a Service on your own server, fx can make your stateless function a service in seconds, both Docker host and Kubernetes cluster supported. The most exciting thing is that you can write your functions with most programming languages.
Feel free hacking fx to support the languages not listed. Welcome to tweet me @_metrue on Twitter, @metrue on Weibo.
Language | Status | Contributor | Example |
---|---|---|---|
Go | Supported | fx | /examples/Golang |
Rust | Supported | @FrontMage | /examples/Rust |
Node | Supported | fx | /examples/JavaScript |
Python | Supported | fx | /examples/Python |
Ruby | Supported | fx | /examples/Ruby |
Java | Supported | fx | /examples/Java |
PHP | Supported | @chlins | /examples/PHP |
Julia | Supported | @matbesancon | /examples/Julia |
D | Supported | @andre2007 | /examples/D |
R | Working on need your help |
Binaries are available for Windows, MacOS and Linux/Unix on x86. For other architectures and platforms, follow instructions to build fx from source.
- MacOS
brew tap metrue/homebrew-fx
brew install metrue/fx/fx
- Linux/Unix
via cURL
# Install to local directory
curl -o- https://raw.githubusercontent.com/metrue/fx/master/scripts/install.sh | bash
# Install to /usr/local/bin/
curl -o- https://raw.githubusercontent.com/metrue/fx/master/scripts/install.sh | sudo bash
fx will be installed into /usr/local/bin, sometimes you may need source ~/.zshrc
or source ~/.bashrc
to make fx available in $PATH
.
- Windows
You can go the release page to download fx manually;
Make sure Docker installed and running on your server first. then type fx -h
on your terminal to check out basic help.
NAME:
fx - makes function as a service
USAGE:
fx [global options] command [command options] [arguments...]
VERSION:
0.8.1
COMMANDS:
init start fx agent on host
up deploy a function
down destroy a service
list, ls list deployed services
call run a function instantly
image manage image of service
doctor health check for fx
help, h Shows a list of commands or help for one command
GLOBAL OPTIONS:
--help, -h show help
--version, -v print the version
- Write a function
You can check out examples for reference. Let's write a function as an example, it calculates the sum of two numbers then returns:
module.exports = (ctx) => {
ctx.body = 'hello world'
}
Then save it to a file func.js
.
- Deploy your function as a service
Give your service a port with --port
, and name with --name
, heath checking with --healthcheck
if you want.
$ fx up -name fx_service_name -p 10001 --healthcheck func.js
2019/08/10 13:26:37 info Pack Service: ✓
2019/08/10 13:26:39 info Build Service: ✓
2019/08/10 13:26:39 info Run Service: ✓
2019/08/10 13:26:39 info Service (fx_service_name) is running on: 0.0.0.0:10001
2019/08/10 13:26:39 info up function fx_service_name(func.js) to machine localhost: ✓
if you want see what the source code of your service looks like, you can export it into a dirctory,
$ fx image export -o <path of dir> func.js
2019/09/25 19:31:19 info exported to <path of dir>: ✓
- Test your service
then you can test your service:
$ curl -v 0.0.0.0:10001
GET / HTTP/1.1
Accept: */*
Accept-Encoding: gzip, deflate
Connection: keep-alive
Host: 0.0.0.0:10001
User-Agent: HTTPie/1.0.2
HTTP/1.1 200 OK
Connection: keep-alive
Content-Length: 11
Content-Type: text/plain; charset=utf-8
Date: Sat, 10 Aug 2019 05:28:03 GMT
hello world
fx is originally designed to turn a function into a runnable Docker container in a easiest way, on a host with Docker running, you can just deploy your function with fx up
command,
fx up --name hello-svc --port 7777 hello.js # onto localhost
DOCKER_REMOTE_HOST_ADDR=xx.xx.xx.xx DOCKER_REMOTE_HOST_USER=xxxx DOCKER_REMOTE_HOST_PASSWORD=xxxx fx up --name hello-svc --port 7777 hello.js # onto remote host
fx supports deploy function to be a service onto Kubernetes cluster infrasture, and we encourage you to do that other than on bare Docker environment, there are lots of advantage to run your function on Kubernetes like self-healing, load balancing, easy horizontal scaling, etc. It's pretty simple to deploy your function onto Kubernetes with fx, you just set KUBECONFIG in your enviroment.
KUBECONFIG=<Your KUBECONFIG> fx deploy -n fx-service-abc_js -p 12349 examples/functions/JavaScript/func.js # function will be deploy to your Kubernetes cluster and expose a IP address of your loadbalencer
or
$ export KUBECONFIG=<Your KUBECONFIG>
$ fx deploy -n fx-service-abc_js -p 12349 examples/functions/JavaScript/func.js # function will be deploy to your Kubernetes cluster and expose a IP address of your loadbalencer
- Local Kubernetes Cluster
Docker for Mac and Docker for Windows already support Kubernetes with single node cluster, we can use it directly, and the default KUBECONFIG
is ~/.kube/config
.
$ export KUBECONFIG=~/.kube/config # then fx will take the config to deloy function
if you have multiple Kubernetes clusters configured, you have to set context correctly. FYI configure-access-multiple-clusters
- Azure Kubernetes Service (AKS)
You should create a Kubernetes cluster if you don't have one on AKS, detail document can found here.
$ az group create --name <myResourceGroup> --location eastus
$ az aks create --resource-group <myResourceGroup> --name myAKSCluster --node-count <count>
$ az aks get-credentials --resource-group <myResourceGroup> --name <myAKSCluster>
Then you can verify it with kubectl
,
$ kubectl get nodes
NAME STATUS ROLES AGE VERSION
aks-nodepool1-31718369-0 Ready agent 6m44s v1.12.8
Since AKS's config will be merged into ~/.kube/config
and set to be current context after you run az aks get-credentials
command, so you can just set KUBECONFIG to default config also,
$ export KUBECONFIG=~/.kube/config # then fx will take the config to deloy function
But we would suggest you run kubectl config current-context
to check if the current context is what you want.
-
Amazon Elastic Kubernetes Service (EKS) TODO
-
Google Kubernetes Engine (GKET) TODO
-
Setup your own Kubernetes cluster
fx init --master <host ip> --agents "<host ip 1>,<host ip 2>" --user <user> --password <password>
fx uses Project to manage the development.
Docker: make sure Docker installed and running on your server.
$ git clone https://github.com/metrue/fx
$ cd fx
$ make build
Then you can build and test:
$ make build
$ ./build/fx -h
Thank you to all the people who already contributed to fx!