katenary/README.md

Katenary is a tool to help transforming docker-compose files to a working Helm Chart for Kubernetes.

Important Note Katenary is a tool to help building Helm Chart from a docker-compose file, but docker-compose doesn't propose as many features as what can do Kubernetes. So, we strongly recommend to use Katenary as a "bootstrap" tool and then to manually enhance the generated helm chart.

Install

If you've got podman or docker, you can build katenary by using:

make build

You can then install it with:

make install

It will use the default PREFIX (~/.local/) to install the binary in the bin subdirectory. You can force the PREFIX value at install time, but maybe you need to use "sudo":

sudo make install PREFIX=/usr/local

Usage

Usage of katenary:
  -appname string
    	sive the helm chart app name (default "MyApp")
  -appversion string
    	set the chart appVersion (default "0.0.1")
  -chart-dir string
    	set the chart directory (default "chart")
  -compose string
    	set the compose file to parse (default "docker-compose.yaml")
  -force
    	force the removal of the chart-dir
  -version
    	Show version and exit

Katenary will try to find a docker-compose.yaml file inside the current directory. It will check *the existence of the chart directory to create a new Helm Chart inside a named subdirectory. Katenary will ask you if you want to delete it before recreating.

It creates a subdirectory inside chart that is named with the appname option (default is MyApp)

To respect the ability to install the same application in the same namespace, Katenary will create "variable" names like {{ .Release.Name }}-servicename. So, you will need to use some labels inside your docker-compose file to help katenary to build a correct helm chart.

What can be interpreted by Katenary:

• Services with "image" section (cannot work with "build" section)
• Named Volumes are transformed to persistent volume claims - note that local volume will break the transformation to Helm Chart because there is (for now) no way to make it working (see below for resolution)
• if ports and/or expose section, katenary will create Services and bind the port to the corresponding container port
• depends_on will add init containers to wait for the depending service (using the first port)
• env_file list will create a configMap object per environemnt file (⚠ todo: the "to-service" label doesn't work with configMap for now)
• some labels can help to bind values, for example:
  • katenary.io/ingress: 80 will expose the port 80 in a ingress
  • katenary.io/to-service: VARNAME will convert the value to a variable {{ .Release.Name }}-VARNAME - it's usefull when you want to pass the name of a service as a variable (think about the service name for mysql to pass to a container that wants to connect to this)

Exemple of a possible docker-compose.yaml file:

version: "3"
services:
    webapp:
        image: php:7-apache
        environment:
            # note that "database" is a service name
            DB_HOST: database
        expose:
            - 80
        depends_on:
            # this will create a init container waiting for 3306 port
            # because it's the "exposed" port
            - database
        labels:
            # explain to katenary that "DB_HOST" value is variable (using release name)
            katenary.io/to-servie: DB_HOST
            # expose the port 80 as an ingress
            katenary.io/ingress: 80
    database:
        image: mariabd:10
        env_file:
            # this will create a configMap
            - my_env.env
        environment:
            MARIADB_ROOT_PASSWORD: foobar
        expose:
            # required to make katenary able to create the init container
            - 3306

Labels

• katenary.io/to-service binds the given (coma separated) variables names to {{ .Release.Name }}-value
• katenary.io/ingress: create an ingress and bind it to the given port
• katenary.io/as-secret: force the creation of a secret for the given coma separated list of "env_file"
• katenary.io/service-ports is a coma separated list of ports if you want to avoid the "ports" section in your docker-compose for any reason