Skip to content

understanding integrity

GitHub Actions edited this page Jun 1, 2026 · 2 revisions

Understanding Integrity

You will learn what integrity checks, how it differs from health, and when it runs.

What integrity answers

Integrity answers one question: does the local on-disk state still match the last successful publication?

It checks local correctness. It does not check whether upstream feeds are alive, reachable, or current. A feed can be completely healthy upstream and still have a local integrity problem if an output file was lost or corrupted after publication.

Integrity vs health

These are separate concerns.

Health asks: is this feed behaving like a healthy upstream source?

Integrity asks: does the local product state still match the last successful local publication?

Do not confuse:

  • a dead upstream feed (health problem)
  • an empty but successful feed (fine — just empty)
  • missing local outputs after a claimed successful publication (integrity problem)

When integrity runs

Integrity runs at three points:

  1. Startup — the daemon inspects settled local feed-output state before listeners are marked ready, then queues repair work for the first scheduler cycle.
  2. After processing settles — once the scheduler finishes a run, integrity re-evaluates to catch local drift.
  3. On operator request — the admin UI integrity panel lets you trigger a fresh evaluation at any time.

What integrity skips

Integrity does not check in-flight work. When a processing run is active, integrity reports in progress instead of emitting transient findings about files that are still being written.

Integrity also does not report on:

  • feeds that are gone upstream and have no local rebuild path
  • IPv6 feeds for IPv4-only features like critical-infrastructure overlap
  • configured providers that have not been materialized yet

What you see

When integrity finds an issue, you know one of two things:

  • the local state lost or corrupted outputs it claimed to have
  • the local state needs a recheck or reprocess to restore consistency

When integrity reports no issue, the local published state is self-consistent with the last successful run.

Getting Started

Installation

Running the Daemon

Configuration

Feed Configuration

Pipeline

Admin UI

Integrity

API Reference

Monitoring

CLI Tools

Troubleshooting

Updating

Catalog Maintenance

Security

Reference

Clone this wiki locally