dstack-tasks

dstack-tasks is a collection of invoke tasks that wrap common tools and services to make it easier to deploy code.

Tools wrapped:

- docker
- docker-compose
- postgresql (backup and restore)
- mysql (backup and restore)
- awscli (s3 only for now)
- git
- python
- django manage.py

The tasks are designed to run as a standalone console script (dstack) or as a collection of tasks to build on in your invoke and fabric2* tasks.

Installation

To install, use pip:

pip install --pre dstack-tasks

Once installed, you can use it from within your project directory. The console script uses a .env file to configure itself. At the moment, some tasks still require environmental variables before it can be used.

Example usage

To see list of tasks, use dstack --list. To see the help text of each task, use dstack <task name> --help. See below for example usage:

# Backs-up database
dstack postgres backup --tag local-dev

# Release new version of app and publish to S3 (requires ~/.aws/credentials to be set up
dstack release_code --upload --not-static

dstack-tasks can also be used as a base library for your invoke and fabric2 tasks:

pip install https://github.com/fabric/fabric/archive/v2.zip

All tasks can be run in "dry" mode or in a specific environment. To see what a tasks will execute, run it in dry mode:

dstack dry postgres backup

To give your task additional context, you can use the special e task to load environmental variables from a .env file:

dstack e release_code

This can be used to for example specify a different Github repo etc.

Notes

It is important to note that dstack-tasks is console script build on top of invoke and is thus primarily meant for executing tasks locally. However, dstack-tasks fully supports being used as a library in your fabric (version 2) fabfile.py. Simply import all tasks from dstack_tasks in your fabfile and you can execute them on a remote server using something like fab -H example.com e deploy --version 1.0.0.

At it's core however, invoke just wraps bash commands and executes them on the server. This means that for advanced uses it might be worthwhile to install dstack-tasks on the server to allow complex tasks to be written in python instead of bash. So, instead of running ls -al via fabric 2 on the remote server and trying to capture and parse the output, you can use an appropriate python package to get a list of files in a directory.

Known Issues

dstack-tasks does not yet include a generic task that can be used to call itself on the server. There is also currently an issue with setting runtime environmental variables using Fabric2 to run tasks remotely.