Skip to main content

1.6.4

Released 30 Oct 2026

New​

  • Choose the certificate chain, and keep old clients working when a CA changes its root. A certificate's new Chain tab picks one of the chains an ACME CA offers and extends it with cross-signed or older intermediates from a chain catalogue that can look them up in the CCADB. The choice applies from the next issuance or renewal, an added certificate is never served past its expiry, and a distribution can keep the CA's default chain.

Improved​

  • Adding a discovery source is a page, and takes a pasted list. The source form opens as its own page with grouped sections and the rarely changed limits folded under Advanced. Hostname and CIDR sources accept a pasted list of names or addresses, each added as a record with the ports you choose.

  • See what a discovered endpoint served over time. On a source's Endpoints tab, the history icon of a row lists every certificate replacement and TLS-settings change discovery recorded for that address, newest first, with the certificate before and after.

  • Export which certificate is served where. A new endpoint export lists every discovered endpoint with the certificate it serves, as CSV or JSON, from the Inventory's Endpoints tab, a discovered certificate, a source and the Quantum readiness list.

  • Discovery has a Quantum readiness page. It shows which endpoints negotiate a quantum-safe key exchange and which do not (and which need TLS 1.3 first), how the certificates' keys are classified, and recent key-exchange changes; the endpoint list can now be filtered by key exchange.

  • Manual Renew and Re-issue ask for confirmation. On the certificate page both actions now say what they will do before they start, for every issuer and status.

Changed behaviour​

  • Health probes are logged at debug. The health, readiness and metrics requests that load balancers and Prometheus send every few seconds no longer fill the request log at info; a probe that fails is still logged as a warning or an error.

Fixed​

  • The nightly backup no longer fails once the database is large, and never fills its disk. The dump is now staged beside the archives instead of in a small temporary disk, and each run makes room by deleting the oldest archives within configurable limits, never the newest. A failed run leaves no partial archive behind; the limits are described under Deployment → Backup and restore.