Architecture¶

Pulp's architecture has three components to it: a REST API, a content serving application, and the tasking system. Each component can be horizontally scaled for both high availability and/or additional capacity for that part of the architecture.

REST API¶

Pulp's REST API is a Django application that runs standalone using the gunicorn like pulpcore-api entrypoint. It serves the following things:

The REST API hosted at /pulp/api/v3/
The browse-able documentation at /pulp/api/v3/docs/
Any viewsets or views provided by plugins
Static content used by Django, e.g. images used by the browse-able API. This is not Pulp content.

Note

A simple way to run the REST API as a standalone service is using the provided pulpcore-api entrypoint. It is gunicorn based and provides many of its options.

The REST API should only be deployed via the pulpcore-api entrypoint.

Content Serving Application¶

A currently aiohttp.server based application that serves content to clients. The content could be Artifacts already downloaded and saved in Pulp, or on-demand content units<on-demand content>. When serving on-demand content units<on-demand content> the downloading also happens from within this component as well.

Note

Pulp installs a script that lets you run the content serving app as a standalone service as follows. This script accepts many gunicorn options.:

$ pulpcore-content

The content serving application should be deployed with pulpcore-content. See --help to see available options.

Availability¶

Ensuring the REST API and the content server is healthy and alive:

REST API: GET request to ${API_ROOT}api/v3/status/ (see [API_ROOT](#))
Content Server: HEAD request to /pulp/content/ or CONTENT_PATH_PREFIX

Distributed Tasking System¶

Pulp's tasking system consists of a single pulpcore-worker component consequently, and can be scaled by increasing the number of worker processes to provide more concurrency. Each worker can handle one task at a time, and idle workers will lookup waiting and ready tasks in a distributed manner. If no ready tasks were found a worker enters a sleep state to be notified, once new tasks are available or resources are released. Workers auto-name and are auto-discovered, so they can be started and stopped without notifying Pulp.

Note

Pulp serializes tasks that are unsafe to run in parallel, e.g. a sync and publish operation on the same repo should not run in parallel. Generally tasks are serialized at the "resource" level, so if you start N workers you can process N repo sync/modify/publish operations concurrently.

All necessary information about tasks is stored in Pulp's Postgres database as a single source of truth. In case your tasking system get's jammed, there is a guide to help (see debugging tasks).

Static Content¶

When browsing the REST API or the browsable documentation with a web browser, for a good experience, you'll need static content to be served.

In Development¶

If using the built-in Django webserver and your settings.yaml has DEBUG: True then static content is automatically served for you.

In Production¶

Collect all of the static content into place using the collectstatic command. The pulpcore-manager command is manage.py configured with the DJANGO_SETTINGS_MODULE="pulpcore.app.settings". Run collectstatic as follows:

$ pulpcore-manager collectstatic

Analytics Collection¶

By default, Pulp installations post anonymous analytics data every 24 hours which is summarized on https://analytics.pulpproject.org/ and aids in project decision making. This is enabled by default but can be disabled by setting ANALYTICS=False in your settings.

Here is the list of exactly what is collected along with an example below:

The version of Pulp components installed as well as the used PostgreSQL server
The number of worker processes and number of hosts (not hostnames) those workers run on
The number of content app processes and number of hosts (not hostnames) those content apps run on
The number of certain RBAC related entities in the system (users, groups, domains, custom roles, custom access policies)

Note

We may add more analytics data points collected in the future. To keep our high standards for privacy protection, we have a rigorous approval process in place. You can see open proposals on https://github.com/pulp/analytics.pulpproject.org/issues. In doubt, reach out to us.

An example payload:

{
    "systemId": "a6d91458-32e8-4528-b608-b2222ede994e",
    "onlineContentApps": {
        "processes": 2,
        "hosts": 1
    },
    "onlineWorkers": {
        "processes": 2,
        "hosts": 1
    },
    "components": [{
        "name": "core",
        "version": "3.21.0"
    }, {
        "name": "file",
        "version": "1.12.0"
    }],
    "postgresqlVersion": 90200
}

Telemetry Support¶

Pulp can produce telemetry data, like the response latency, using OpenTelemetry. You can read more about OpenTelemetry here. The telemetry is disabled by default.

Attention

This feature is provided as a tech preview and could change in backwards incompatible ways in the future.

In order to enable the telemetry, set OTEL_ENABLED=True in the settings file and follow the next steps:

Spin up a new instance of the Opentelemetry Collector.
Configure the OTEL_EXPORTER_OTLP_ENDPOINT environment variable to point to the address of the OpenTelemetry Collector instance (e.g.,http://otel-collector:4318).
Set the OTEL_EXPORTER_OTLP_PROTOCOL environment variable to http/protobuf.

At the moment, the following data is recorded by Pulp:

Latency of API endpoints (along with an HTTP method, URL, status code, and unique worker name).
Latency of delivering requested packages (an HTTP method, status code, and unique worker name).
Disk usage within a specific domain (total used disk space and the reference to a domain).
The size of served artifacts (total count of served data and the reference to a domain).

The information above is sent to the collector in the form of metrics. Thus, the data is emitted either based on the user interaction with the system or on a regular basis. Consult OpenTelemetry Metrics to learn more.