Skip to content

ob status does not report a route domain that has no certificate #237

Open

Description

@vishr

What happens

ob status reports a healthy proxy when a terminating route's domain has no certificate from the resolver that should issue it. If the configured resolver never obtains a certificate for a domain (for example, the HTTP-01 challenge is blocked or the DNS-01 token is wrong), Traefik serves its default self-signed certificate for that domain. Status does not mention the domain at all.

Status only judges certificates that exist in the ACME stores. A domain with no renewable certificate does not appear there, so it raises no issue and does not set Diverged.

Before #236 this case was sometimes caught by accident. If a leftover resolver key still held an old certificate for the domain, that certificate was eventually flagged as overdue. Since #236, a leftover certificate is marked orphaned and never flagged, so nothing reports the missing one.

Expected

For each domain that a terminating route serves, status checks that a resolver in the applied static configuration holds a certificate for it in the store it writes, and that the certificate is outside the renewal floor. A route domain with no such certificate is reported as an issue, and status diverges. Wildcard routes are covered by a matching wildcard certificate from the wildcard resolver.

Notes

  • The route domains come from the application's routes, which status already knows, or from the discovery controller's output in /etc/traefik/dynamic.
  • A certificate's SANs should count as well as domain.main. proxy.CertExpiries currently extracts only domain.main, so a domain covered only through a SAN would be reported as missing.
  • A route whose certificate is still being issued right after its first deploy should not report divergence immediately. A short grace period, or reporting it as pending, avoids a false alarm.

Reproduce

  1. Deploy an application with a terminating route for a domain whose ACME challenge cannot succeed, for example a domain whose DNS does not point at the host.
  2. Wait for Traefik's issuance attempt to fail. The domain then serves Traefik's default certificate.
  3. Run ob status. It reports the proxy as in sync and exits zero.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or request

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions