temichus/argus

Argus

Description

Argus is a test tracking system intended to provide observability into automated test pipelines which use long-running resources. It allows observation of a test status, its events and its allocated resources. It also allows easy comparison between particular runs of a specific test.

Installation notes

Prerequisites

• Python >=3.10.0 (system-wide or pyenv)

• NodeJS >=16 (with npm)

• Yarn (can be installed globally with npm -g install yarn)

• nginx

• poetry >=1.2.0b1

From source

Production

Perform the following steps:

Create a user that will be used by uwsgi:

useradd -m -s /bin/bash argus
sudo -iu argus

(Optional) Install pyenv and create a virtualenv for this user:

pyenv install 3.10.0
pyenv virtualenv argus
pyenv activate argus

Clone the project into a directory somewhere where user has full write permissions

git clone https://github.com/scylladb/argus ~/app
cd ~/app

Install project dependencies:

poetry install --with default,dev,web-backend,docker-image
yarn install

Compile frontend files from /frontend into /public/dist

yarn webpack

Create a argus.local.yaml configuration file (used to configure database connection) and a argus_web.yaml (used for webapp secrets) in your application install directory.

cp argus_web.example.yaml argus_web.yaml
cp argus.yaml argus.local.yaml

Open argus.local.yaml and add the database connection information (contact_points, user, password and keyspace name).

Open argus_web.yaml and change the SECRET_KEY value to something secure, like a sha512 digest of random bytes. Fill out GITHUB_* variables with their respective values.

Copy nginx configuration file from docs/configs/argus.nginx.conf to nginx virtual hosts directory:

Ubuntu:

sudo cp docs/configs/argus.nginx.conf /etc/nginx/sites-available/argus
sudo ln -s /etc/nginx/sites-enabled/argus /etc/nginx/sites-available/argus

RHEL/Centos/Alma/Fedora:

sudo cp docs/configs/argus.nginx.conf /etc/nginx/conf.d/argus.conf

Adjust the webhost settings in that file as necessary, particularly listen and server_name directives.

Copy systemd service file from docs/config/argus.service to /etc/systemd/system directory:

sudo cp docs/config/argus.service /etc/systemd/system

Open it and adjust the path to the start_argus.sh script in the ExecStart= directive and the user/group, then reload systemd daemon configuration and enable (and optionally start) the service.

WARNING: start_argus.sh assumes pyenv is installed into ~/.pyenv

sudo systemctl daemon-reload
sudo systemctl enable --now argus.service

Development

Clone the project into a directory somewhere

git clone https://github.com/scylladb/argus
cd argus

Install project dependencies:

poetry install --with default,dev,web-backend,docker-image
yarn install

Compile frontend files from /frontend into /public/dist. Add --watch to recompile files on change.

yarn webpack --watch

Configuration

Create a argus.local.yaml configuration file (used to configure database connection) and a argus_web.yaml (used for webapp secrets) in your application install directory.

See Production section for more details. To configure Github authentication follow steps:

1. Authorize OAuth App
   1. go to your Account Settings (top right corner) -> Developer settings (left pane) -> OAuth Apps
   2. Click Create New OAuth App button
   3. Fill the fields (app name: argus-dev, homepage URL http://localhost:5000, Auth callback URL: http://localhost:5000/profile/oauth/github)
   4. Confirm and get the tokens/ids required for config
2. Create Jenkins token for your account
   1. Go to Configure in top right corner
   2. Click Add new Token
   3. Get it and paste to config to JENKINS_API_TOKEN param

Database Initialization

You can initialize a scylla cluster in any way you like, either using docker image with docker-compose or using cassandra cluster manager. You will need to create the keyspace manually before you can sync database models.

Create keyspace according to your configuration. e.g. (need to test if it works with RF=1 if not, make it 3)

CREATE KEYSPACE argus WITH replication = {'class': 'SimpleStrategy', 'replication_factor' : 1}

Initial sync can be done as follows:

from argus.backend.db import ScyllaCluster
from argus.db.testrun import TestRun
db = ScyllaCluster.get()

db.sync_models() # Syncronizes Object Mapper models
TestRun.init_own_table() # Syncronizes TestRun table (separate from python-driver Object Mapper)

You can also use flask sync-models afterwards during development when making small changes to models.

It is recommended to set up jenkins api key and run flask scan-jenkins afterwards to get basic release/group/test structure.

There are scripts in ./scripts directory that can be used to download data from production, upload them into your dev db and fix their relations to other models in your instance of the application. Specifically, download_runs_from_prod.py requires additional config, argus.local.prod.yaml which is the config used to connect to the production cluster. The scripts are split to prevent mistakes and accidentally affecting production cluster.

Configuration

Create a argus.local.yaml configuration file (used to configure database connection) and a argus_web.yaml (used for webapp secrets) in your application install directory.

cp argus_web.example.yaml argus_web.yaml
cp argus.yaml argus.local.yaml

Open argus.local.yaml and add the database connection information (contact_points, user, password and keyspace name).

Open argus_web.yaml and change the SECRET_KEY value to something secure, like a sha512 digest of random bytes. Fill out GITHUB_* and JENKINS_* variables with their respective values.

Run the application from CLI using:

FLASK_ENV="development" FLASK_APP="argus_backend:start_server" FLASK_DEBUG=1 CQLENG_ALLOW_SCHEMA_MANAGEMENT=1 flask run

Omit FLASK_DEBUG if running your own debugger (pdb, pycharm, vscode)