Skip to content

HTTP interface for finding product listings and interacting with data in composedb

License

Notifications You must be signed in to change notification settings

release-engineering/product-listings-manager

Repository files navigation

product-listings-manager

https://quay.io/repository/factory2/product-listings-manager/status https://coveralls.io/repos/github/release-engineering/product-listings-manager/badge.svg?branch=master

HTTP interface for finding product listings and interacting with data in composedb.

Previously this mechanism was a Koji plugin. This product-listings-manager project extracts the plugin into its own web service apart from Koji.

REST API

You may use HTTP GET request to get /api/v1.0/product-info/<PRODUCT> or /api/v1.0/product-listings/<PRODUCT>/<BUILD_INFO>. The parameters and results are same as for the XML-RPC getProductInfo and getProductListings calls used in Brew. See client.py for an example.

What is ComposeDB?

All RPM-based products shipped through the Errata Tool use ComposeDB for "product listings".

ComposeDB defines a product by mapping RPMs to products/variants/arches.

Goals with this project

  • Implement the Brew plugin's API and functionality exactly, in order to minimize Errata Tool changes when migrating to this new service.
  • Eliminate Brew/Koji's constraint of using XML-RPC
  • Decouple the development of this service so that it is easier to develop, test, and deploy apart from Brew's prod environment.

Architecture diagram

product-listings-manager architecture diagram

Running the tests

You can invoke the tests with tox:

$ tox

Using the --live argument if you want to run against the live composedb instance:

$ tox -e py3 -- --cov=product_listings_manager --live tests

Running the linters

To run viable linters to check syntax of various files before commit, install pre-commit and run:

$ pre-commit install

To run linters on all files (not just the ones changed in the last commit), run:

$ pre-commit run -a

Setting up local environment

You can use docker-compose or podman-compose to start:

  • product-listings-manager - the web service running at http://localhost:8080
  • postgres - database for product-listings-manager, initialized with all *.sql files in docker/docker-entrypoint-initdb.d directory
  • jaeger - collector and query service for OpenTelemetry traces collected from the local instance of product-listings-manager and the database, running at http://localhost:16686

Rebuild product-listings-manager image:

$ podman-compose build

Image rebuild is needed only if dependencies change. The container running in the compose environment uses the current source code directory.

Start the services:

$ podman-compose up

Show logs:

$ podman-compose logs plm
$ podman-compose logs plm-db
$ podman-compose logs jaeger

Restart product-listings-manager:

$ podman-compose restart plm

Stop the services:

$ podman-compose up

Configuration

The service is conteinerized application and uses environment variables for configuration:

  • SQLALCHEMY_DATABASE_URI - full database URI for SQLAlchemy, for example: postgresql://username:password@plm-db.example.com:5433/plm

  • OTEL_EXPORTER_OTLP_TRACES_ENDPOINT - traces endpoint for OpenTelemetry tracing, for example: https://otel.example.com/v1/traces

  • OTEL_EXPORTER_SERVICE_NAME - service name for OpenTelemetry tracing

  • PLM_KOJI_CONFIG_PROFILE - Koji profile to use (in /etc/koji.conf.d/ directory), default is brew

  • PLM_LDAP_HOST - LDAP host, for example ldaps://ldap.example.com

  • PLM_LDAP_SEARCHES - JSON formatted array with LDAP search base and search template, for example:

    ["BASE": "ou=Groups,dc=example,dc=com", "SEARCH_STRING": "(memberUid={user})"]
  • PLM_PERMISSIONS - JSON formatted array with permissions, for example:

    [
      {
        "name": "admins",
        "description": "product-listings-manager admins",
        "contact": "plm-admins@example.com",
        "queries": ["*"],
        "groups": ["plm-admins"],
        "users": ["alice", "bob"]
      },
      {
        "name": "viewers",
        "queries": ["SELECT *"],
        "groups": ["plm-users"]
      }
    ]