MANUAL_CLI.md

Manual CLI options

In addition to YAML file support, you can use the CLI to build and deploy individual functions as follows:

Worked example with Node.js

So if you want to write in another language, just prepare a Dockerfile and build an image manually, like in the FaaS samples.

Build a FaaS function in NodeJS from a template:

This will generate a Docker image for a Node.js function using the code in /samples/info.

• The faas-cli build command can accept a --lang option of python or node and is node by default.

   $ faas-cli build \
      --image=alexellis2/node_info \
      --name=node_info \
      --handler=./sample/node_info

Building: alexellis2/node_info with Docker. Please wait..
...
Image: alexellis2/node_info built.

You can customise the code by editing the handler.js file and changing the --handler parameter. You can also edit the packages.json file, which will be used during the build to make sure all your dependencies are available at runtime.

For example:

"use strict"

module.exports = (context, callback) => {
    console.log("echo - " + context);
    
    callback(undefined, {status: "done"});
}

The CLI will then build a Docker image containing the FaaS watchdog and a bootstrap file to invoke your NodeJS function.

Deploy the Docker image as a FaaS function:

Now we can deploy the image as a named function called node_info.

$ faas-cli deploy \
   --image=alexellis2/node_info \
   --name=node_info

200 OK

URL: http://localhost:8080/function/node_info

This tool can be used to deploy any Docker image as a FaaS function, as long as it includes the watchdog binary as the CMD or ENTRYPOINT of the image.

Deploy remotely

You can deploy to a remote FaaS instance as along as you push the image to the Docker Hub, or another accessible Docker registry. Specify your remote gateway with the following flag: --gateway=http://remote-site.com:8080