close
Skip to content

peloton

Directory actions

More options

Directory actions

More options

Latest commit

 

History

History

peloton

README.md

Scripted Peloton refresh

What is the Peloton?

Here it is. Basically a GitHub Project that makes it easy to monitor multiple repositories, and encodes agreed priorities via the Views that it has.

Help! The Peloton is out of date!

Follow all steps in How to refresh the Peloton manually.

How the Peloton is refreshed

The sgqlc library provides a Pythonic interface to GitHub's GraphQL API - the primary way of interacting with GitHub programmatically. update_project.py uses sgqlc to query GitHub issues/discussions, post-processes the data using the pandas library, then uses sgqlc to update the Peloton project. This is usually run automatically by this GitHub Action.

How to refresh the Peloton manually

Generate a GitHub access token

Performing backend automation with GitHub requires an access token with appropriate permissions. Visit this page for instructions on creating your own.

The token must have the following scopes to enable running all Peloton refresh operations:

  • All repo scopes
  • admin:org > read:org
  • write:discussion > read:discussion
  • project > read:project

(For the GitHub Action: a token is generated on-the-fly by the scitools-ci GitHub App, via the tibdex/github-app-token action. See the GitHub Action for more.)

Set up an environment

Run all scripts in an environment that contains the packages listed in requirements.txt.

Run the refresh script

It is important to try this first, to check whether you can replicate any problems observed on the GitHub Action.

python update_project.py --bearer_token <your_github_access_token>

See also python update_project.py --help for more options.

Output will be written to latest_peloton_update.log.

Debugging

  1. Try reducing the _PAGINATION numbers in update_project.py to see if the script is failing due to rate limiting or timeouts.
  2. Try Updating the Python schema.
  3. Further troubleshooting will be case-by-case.

Updating the Python 'schema'

Use update-schema.sh to convert the latest GitHub GraphQL schema into Python classes; you will need your GitHub token to run this. Note that the script produces both a JSON and a Python file, but only the Python file (github_schema.py) is under source control - the JSON file is not used. This Python 'schema' is a very large file, and we do not vet its contents.