chore(ci): check the RustSec advisory database on a schedule, so a new advisory is found on main and not on someone's pull request

Offen
  • Stimmt etwas nicht?
#90 0 Kommentare 0 Reaktionen 0 zugewiesene Personen Auf GitHub ansehen

Maintainer antworten meist innerhalb von 1 Tag

Dieses Issue hat noch niemand übernommen.

Bewertung

Schwierigkeit
4/5
Geschätzter Aufwand
3-5 Tage
Anfängerfreundlichkeit
56/100
Issue-Typ
Feature
Klarheit
Klar beschrieben
Aktivitätsstatus
Aktiv
Tech-Stack
github-actions, rust
Bereich
ci-cd, documentation, security

Rechercherichtung

Read .github/workflows/ci.yml for the existing deny job and .github/workflows/flake-hunt.yml for the schedule precedent, then review deny.toml and the supply-chain section of ADR-0101. Add a scheduled, manually dispatchable advisory check and document its failure handling in docs/src/architecture/cicd.md and docs/src/contributing/standards.md; verify the documented control by removing the RUSTSEC-2023-0071 ignore on a throwaway branch and confirming the run names the advisory.

Vom Indexierungsmodell aus dem Issue-Text verfasst.

Beschreibung

area:process prio:p1

Context

The deny job (added by #87, closing #40) runs cargo deny --locked check, which fetches the RustSec advisory database on every run, and it is in gate's needs, the only required check. So a newly published advisory turns the job red, and every Rust pull request with it, with no change to the code. Nothing looks at the database between pull requests, so the first person to find out is whoever happens to open one.

This is not hypothetical. When deny.toml was written, seven advisories already failed on main, and two of them (RUSTSEC-2026-0385 and -0387) had been published on 2026-10-23, three days after #40 was filed on 2026-10-20. They were found only because that work ran against the live database.

ci.yml triggers on pull_request, merge_group, push to main and workflow_dispatch. The only scheduled workflow is flake-hunt.yml (weekly).

Problem

  • A new advisory is first seen on an unrelated pull request, which is then red for a reason it did not cause.
  • main can be red-in-waiting for as long as nobody pushes.

Proposal

  • A scheduled workflow that runs cargo deny --locked check advisories against main, with the same pinned cargo-deny and the same deny.toml, and fails visibly when an advisory appears. Decide how it tells someone: a failing scheduled run that watchers are notified of, or one tracking issue it opens and updates. Daily is probably right for something time-sensitive; flake-hunt.yml is weekly.
  • Keep it out of gate: it reports on main, and gate is about the pull request in front of it.
  • Decide whether the pull-request deny job should keep the advisories check. Keeping it means an unrelated advisory blocks every pull request until it is fixed (today's behaviour, fail-closed). Dropping it to licenses bans sources decouples pull requests from drift but leaves advisories to the schedule alone. The schedule does not change which of those you pick; it only makes the warning arrive on main first.
  • Note that flake-hunt.yml records that GitHub disables scheduled workflows in a public repository after 60 days without repository activity.

Acceptance criteria

  • A scheduled workflow exists (actions pinned by SHA, permissions no wider than it needs) and can also be started by workflow_dispatch.
  • The control: on a throwaway branch with the RUSTSEC-2023-0071 ignore deleted from deny.toml, a dispatched run fails and names that advisory (as the throwaway pull request #88 did for the pull-request job).
  • What happens when it fails, and who is told, is written in docs/src/architecture/cicd.md. That page and docs/src/contributing/standards.md ("Decided, and built only in part") now say there is no scheduled check of the advisory database; update both.

References

#40, #87; deny.toml; the deny job in .github/workflows/ci.yml; .github/workflows/flake-hunt.yml (the schedule precedent and the 60-day note); ADR-0101 (.agents/plans/contrail/adrs/ci/0101-ci-supply-chain.md); docs/src/architecture/cicd.md, "Permissions and supply chain".

Vorherrschende Sprache
Rust
Sterne
0
Forks
0
Ø Merge
1 Std. 35 Min.
Gemergte PRs (30 T.)
67

Entwicklungsumgebung

Erste Schritte

  1. Lesen Sie das ganze Issue und danach den Beitragsleitfaden des Projekts.
  2. Schreiben Sie ins Issue, dass Sie es übernehmen — das erspart doppelte Arbeit.
  3. Forken Sie das Repository und arbeiten Sie in einem Branch.
  4. Öffnen Sie einen Pull Request, der die Issue-Nummer nennt.

Mehr aus aRustyDev/thrust

Alle Issues in aRustyDev/thrust

Ähnliche Issues

Weitere Issues zu Rust

Neue Issues direkt in Ihr Postfach

Eine kurze Übersicht über anfängerfreundliche GitHub-Issues.