Summary

  • At RIPE 93, RIPE NCC reported about 8,000 undeliverable addresses, a level that had stayed roughly the same since RIPE 92.
  • Those addresses affected about 1% of daily update notifications. If every recipient for an object fails, RIPE NCC says no warning gets through.

A database change has two audiences: the record and the people told to watch it. The sender may receive an acknowledgement while every designated watcher misses the notice. RIPE NCC said warnings go to other recipients when some addresses fail; if every recipient fails, nobody can receive the warning.

At RIPE 93, the RIPE NCC Database team reported about 8,000 undeliverable addresses, roughly the same level as at RIPE 92, affecting about 1% of daily update notifications.

RIPE 92's transcript gives the repair process more texture. As of mid-May, the team counted 69 unsubscribed addresses; their users had opted out, a different condition from delivery failure. RIPE NCC said it periodically revalidated undeliverable addresses with a third-party service. A manual check of the full pool typically changed only about 10% of the addresses. A scheduled job retried SMTP failures to see whether they persisted; the team said many failures were permanent.

The RIPE Database documentation on notifications separates the update acknowledgement from the message sent to people who follow database objects. An acknowledgement may arrive by email or another HTTP connection. Notifications always go by email. The first confirms an update to its sender; the second tells designated recipients that an object changed.

That split decides what the 1% means operationally. Some failed addresses can be covered by other recipients; a fully failed recipient list leaves the object without a working alert route. The RIPE 93 transcript reports the share of daily notifications affected, but does not say which maintainers or objects depend on the addresses. That map would show where a lost notice threatens a real control process and who can repair it.

At RIPE 92, the team asked whether the community wanted a better way to notify users that did not rely on email. RIPE 93's transcript reports the same scale of undeliverable addresses but records no decision on a second route.

The operational choice is straightforward: show failed notification routes beside the records and maintainers that depend on them, or add a second route that users can keep current. Either choice needs a named owner and a recovery test. The registry can preserve a change while its alert fails before reaching anyone responsible for acting on it.

Sources: RIPE 93 Database Working Group transcript · RIPE Database Notifications documentation · RIPE 92 Database Working Group transcript