- Introduction
- Prerequisites
- Preparing the environment
- Supplying data to the devcontainer
- Note for Windows users
- Running and debugging
- Running unit tests
- Running serverspec tests
This document explains how to set up the preferred VSCode development environment. While there are other options to develop Epiphany like PyCharm, VSCode has the following advantages:
-
Epiphany is developed using many different technologies (Python, Ansible, Terraform, Docker, Jinja, YAML...) and VSCode has good tooling and extensions available to support everything in one IDE.
-
VSCode's devcontainers allow us to quickly set up a dockerized development environment, which is the same for every developer regardless of development platform (Linux, MacOS, Windows).
Note: More information when running the devcontainer environment on Windows or behind a proxy can be found here.
Note: VSCode devcontainers are not properly supported using Docker Toolbox on Windows. More info here.
-
Open the epicli project folder
/epiphany/core/src/epicli/with VSCode. -
VSCode will tell you that the workspace has recommended extensions:
Press
Install Alland wait until they are all installed and then restart. During the extension installations the following popup might show up:
Do NOT do that at this point. First you must restart VSCode to activate all extensions which were installed.
-
After restarting VSCode the popup to re-open the folder in a devcontainer will show again. Press
Reopen in Containerto start the build of the devcontainer. You should get the following message:
You can click
detailsto show the build process. -
After the devcontainer is built and started, VSCode will show you the message again that this workspace has recommended extensions. This time it is for the devcontainer. Again, press
Install Allto install the available extensions inside the devcontainer.
Now you have a fully working Epiphany development environment!
The entire working directory (/epiphany/core/src/epicli/) is mounted inside the container. We recommend to create an additional directory called clusters there, in which you house your data YAMLs and SSH keys. This directory is already added to the .gitignore. When executing epicli commands from that directory this is also where any build output and logs are written to.
-
Watch out for line endings conversion. By default GIT for Windows sets
core.autocrlf=true. Mounting such files with Docker results in^Mend-of-line character in the config files. Use: Checkout as-is, commit Unix-style (core.autocrlf=input) or Checkout as-is, commit as-is (core.autocrlf=false). -
Mounting NTFS disk folders in a Linux based image causes permission issues with SSH keys. You can copy them inside the container and set the proper permissions using:
mkdir -p /home/vscode/.ssh cp ./clusters/ssh/id_rsa* /home/vscode/.ssh/ chmod 700 /home/vscode/.ssh && chmod 644 /home/vscode/.ssh/id_rsa.pub && chmod 600 /home/vscode/.ssh/id_rsa
This needs to be executed from the devcontainer bash terminal:
For debugging, open the VSCode's Debug tab:
By default there is one launch configuration called epicli. This launch configuration can be found in /epiphany/core/src/epicli/.vscode/ and looks like this:
...
{
"name": "epicli",
"type": "python",
"request": "launch",
"program": "${workspaceFolder}/cli/epicli.py",
"cwd": "${workspaceFolder}",
"pythonPath": "${config:python.pythonPath}",
"env": { "PYTHONPATH": "${workspaceFolder}" },
"console": "integratedTerminal",
"args": ["apply", "-f", "${workspaceFolder}/PATH_TO_YOUR_DATA_YAML"]
}
...You can copy this configuration and change values (like below) to create different ones to suite your needs:
...
{
"name": "epicli",
"type": "python",
"request": "launch",
"program": "${workspaceFolder}/cli/epicli.py",
"cwd": "${workspaceFolder}",
"pythonPath": "${config:python.pythonPath}",
"env": { "PYTHONPATH": "${workspaceFolder}" },
"console": "integratedTerminal",
"args": ["apply", "-f", "${workspaceFolder}/PATH_TO_YOUR_DATA_YAML"]
},
{
"name": "epicli show version",
"type": "python",
"request": "launch",
"program": "${workspaceFolder}/cli/epicli.py",
"cwd": "${workspaceFolder}",
"pythonPath": "${config:python.pythonPath}",
"env": { "PYTHONPATH": "${workspaceFolder}" },
"console": "integratedTerminal",
"args": ["--version"]
}
...In the args field you can pass an array of the arguments that you want epicli to run with.
To run a configuration, select it and press the run button:
For more information about debugging in VSCode, go here.
The standard Python test runner fails to discover the tests so we use the Python Test Explorer extension. To run the unit tests, open the VSCode's Test tab and press the run button:
See the Python Test Explorer extension page on how to debug and run individual tests.
You can also run the Python unit tests from a launch configuration called unit tests
We maintain a set of serverspec tests that can be run to verify if a cluster is functioning properly. While it might not cover all cases at this point it is a good place to start.
The serverspec tests are integrated in Epicli. To run them you can extend the launch configuration epicli with the following arguments:
...
{
"name": "epicli",
"type": "python",
"request": "launch",
"program": "${workspaceFolder}/cli/epicli.py",
"cwd": "${workspaceFolder}",
"pythonPath": "${config:python.pythonPath}",
"env": { "PYTHONPATH": "${workspaceFolder}" },
"console": "integratedTerminal",
"args": ["test", "-b", "${workspaceFolder}/clusters/buildfolder/", "-g", "postgresql"]
},
...Where the -b argument points to the build folder of a cluster. The -g argument can be used to execute a subset of tests and is optional. Omitting -g will execute all tests.
Information about how to manage the Epicli Python dependencies can be found here.