RobotnikMQ

Utilities for safe, efficient, and scalable infrastructure using RabbitMQ

Usage

TODO

Installation & Setup

To install robotnikmq with pip execute the following:

pip install robotnikmq

Configuration

RobotnikMQ can be configured globally, on a per-user, or on a per-application basis. When certain functions of the RobotnikMQ library are called without a provided configuration, it will attempt to find a configuration first for the application in the current working directory ./robotnikmq.yaml, then for the user in ~/.config/robotnikmq/robotnikmq.yaml and then for the system in /etc/robotnikmq/robotnikmq.yaml. An error will be raised if a configuration is not provided and neither of those files exist.

The RobotnikMQ configuration is primarily a list of servers organized into tiers. If a given system or user can be expected to connect to the same cluster the vast majority of the time, then you can/should use a per-user or global configuration. Otherwise, simply have your application configure its own RobotnikMQ configuration (see Usage section).

The configuration file itself should look something like this:

tiers:
- - ca_cert: /home/eugene/Development/robotnikmq/tests/integration/vagrant/pki/robotnik-ca.crt
    cert: /home/eugene/Development/robotnikmq/tests/integration/vagrant/pki/issued/rabbitmq-vm/rabbitmq-vm.crt
    host: 127.0.0.1
    key: /home/eugene/Development/robotnikmq/tests/integration/vagrant/pki/issued/rabbitmq-vm/rabbitmq-vm.key
    password: ''
    port: 1
    user: ''
    vhost: ''
  - ca_cert: /home/eugene/Development/robotnikmq/tests/integration/vagrant/pki/robotnik-ca.crt
    cert: /home/eugene/Development/robotnikmq/tests/integration/vagrant/pki/issued/rabbitmq-vm/rabbitmq-vm.crt
    host: '1'
    key: /home/eugene/Development/robotnikmq/tests/integration/vagrant/pki/issued/rabbitmq-vm/rabbitmq-vm.key
    password: '1'
    port: 1
    user: '1'
    vhost: '1'
- - ca_cert: /home/eugene/Development/robotnikmq/tests/integration/vagrant/pki/robotnik-ca.crt
    cert: /home/eugene/Development/robotnikmq/tests/integration/vagrant/pki/issued/rabbitmq-vm/rabbitmq-vm.crt
    host: 127.0.0.1
    key: /home/eugene/Development/robotnikmq/tests/integration/vagrant/pki/issued/rabbitmq-vm/rabbitmq-vm.key
    password: hackme
    port: 5671
    user: robotnik
    vhost: /robotnik

In the example above, you should be able to see two tiers of servers, the first has two server configurations that are intentionally broken for testing purposes, while the second has a valid configuration (this is the configuration that is used for unit-testing).

The idea is that RobotnikMQ will first attempt to connect to all servers in the first tier in a random order, then if all of them fail, it will attempt to connect to all the servers in the second tier, and so on. This is intended to allow both load-balancing on different servers and for redundancy in case some of those servers fail. You can also configure only one tier with one server, or just a list of tiers, each of which have one server in them. This way, the secondary and tertiary servers would only be used if there is something wrong with the primary.

Development

Standards

• Be excellent to each other
• Code coverage must be at 100% for all new code, or a good reason must be provided for why a given bit of code is not covered.
  • Example of an acceptable reason: "There is a bug in the code coverage tool and it says its missing this, but its not".
  • Example of unacceptable reason: "This is just exception handling, its too annoying to cover it".
• The code must pass the following analytics tools. Similar exceptions are allowable as in rule 2.
  • pylint --disable=C0103,C0111,W1203,R0903,R0913 --max-line-length=120 ...
  • flake8 --max-line-length=120 ...
  • mypy --ignore-missing-imports --follow-imports=skip --strict-optional ...
• All incoming information from users, clients, and configurations should be validated.
• All internal arguments passing should be typechecked during testing with typeguard.typechecked or the import hook.

Development Setup

Using pdm install from inside the repo directory:

pdm install

This will install all dependencies (including dev requirements) in a PEP582-compliant project which you can always run specific commands with pdm run ....

IDE Setup

Sublime Text 3

curl -sSL https://gitlab.com/-/snippets/2385805/raw/main/pdm.sublime-project.py | pdm run python > robotnikmq.sublime-project

VSCodium/VSCode

I recommend installing the pdm-vscode plug-in:

sudo pdm plugin add pdm-vscode

Testing

All testing should be done with pytest which is installed with the dev requirements.

To run all the unit tests, execute the following from the repo directory:

pdm run pytest --runslow

Removing the --runslow parameter will cause it to skip tests that have been marked with @pytest.mark.slow

This should produce a coverage report in /path/to/dewey-api/htmlcov/

While developing, you can use watchexec to monitor the file system for changes and re-run the tests:

watchexec -r -e py,yaml pdm run pytest

To run a specific test file:

pdm run pytest tests/unit/test_core.py

To run a specific test:

pdm run pytest tests/unit/test_core.py::test_hello

For more information on testing, see the pytest.ini file as well as the documentation.

Integration Testing

For integration testing, this code uses testcontainers-rabbitmq. In order to enable this, you will need to install docker, and ensure that your user has the ability to interact with the docker service:

sudo apt install docker
sudo groupadd docker # may fail if the docker group already exists
sudo usermod -aG docker $USER
newgrp docker
docker run hello-world # verify that everything is working well

We also have tests that do more complex things, such as tests/integration/test_broadcast.py::test_confirm_subscriber_reconnection_behavior_on_node_failure which rely on the Python Docker SDK, but that needs docker installed as well.

Python Version Support

RobotnikMQ also has Tox configured for all supported versions of python so that we can verify that it works with all of them. This is defined in the tox.ini file and all tests can be run with:

tox run