Summary

  • RIPE NCC opened an incident for intermittent WHOIS update failures at 20:04 CET on 28 October, moved the RIPE Database component back to operational at 20:23, and marked the incident resolved at 20:32.
  • The public status record reports the fix and recovery, but gives no count or list of attempted edits; an operator needs the response to their own submission and an authoritative readback to settle its fate.

At 20:23 on Wednesday evening, the RIPE Database badge turned green again. That is useful news for the next person about to send an update. It is a thinner answer for the person who had already pressed submit.

The RIPE NCC incident record calls the event “Intermittent WHOIS update failures.” Its public incident data records an investigation at 20:04:48 CET on 28 October, a fix under monitoring at 20:23:24, and resolution at 20:32:00. During the investigation the RIPE Database component was marked a partial outage; at monitoring it returned to operational. The incident as a whole carries a “major” impact label. The status entries say nothing about the cause, affected update interfaces or number of failed submissions. Those timestamps describe the public response, not the first and last individual failed edit.

Why insist on a receipt after the light changes? RIPE's acknowledgement guide says each submitted update gets one acknowledgement carrying the outcomes for its objects. A Syncupdates HTTP 200 can mean the message was processed while one object failed; the summary, not the status code alone, carries that distinction. The REST API uses a different response format. These are ordinary rules of the service, not a diagnosis of Wednesday's fault.

There is a sharper edge for a client whose connection failed. RIPE's object-processing documentation says objects are processed independently and that database work can finish after a synchronous HTTP connection closes, leaving no acknowledgement on that connection. A blind retry can therefore repeat a submission whose fate the caller has not established. The sensible sequence is to inspect any method-specific response, query the current authoritative object, then decide whether a new write is needed.

The status page can tell every member when RIPE NCC declared the service healthy. It cannot tell an individual member what the registry accepted on their behalf. The two clocks deserve separate lines in the incident record.