Skip to content

Outage on 12 November 2026 resolved

Read the report

Outage on 12 November 2026

In the early hours of 12 November 2026 a central server at our Frankfurt site failed. The customer portal, website, API, DynDNS, web forwarding, webmail and the websites in WordPress hosting were unavailable for just over seven hours. We apologise for this.

Contact supportStatus page

At a glance

Period
12 November 2026, 02:31 to 09:55 CET
Duration
about 7 hours 24 minutes
Status
resolved
Data loss
none
Name resolution, incoming and outgoing email, web hosting
not affected

Impact

What was affected and what was not

The failed server runs administration and ordering. Name resolution, email and web hosting work separately from it. WordPress hosting is controlled through this server and was therefore affected as well.

Unavailable

  • Customer portal dash with domain management, DNS editor and invoices
  • Website regfish.de with domain search and ordering
  • DNS API and MCP server
  • DynDNS updates (IPv4 and IPv6)
  • Web forwarding of domains
  • Websites in WordPress hosting
  • Webmail interfaces
  • Domain registrations, transfers and certificate orders
  • Documentation and file sharing (docs, files)

Available throughout

  • Name resolution of all domains via our anycast network, including DNSSEC
  • Receiving and sending email via IMAP, POP3 and SMTP
  • All websites in web hosting

Timeline

What happened that night

All times in German time (CET).

  1. 02:29The server halts after a fault in the operating system kernel.
  2. 02:31Our monitoring reports the first failures.
  3. 03:59We report the outage to the data centre.
  4. 04:45A restart via the remote management interface brings the system up, but it cannot connect to the network. Shortly afterwards the remote management interface itself becomes unreachable as well.
  5. 05:52The data centre confirms that the remote management interface is defective. Further work is only possible on site.
  6. 07:46 to 08:21On-site work on the network configuration does not yet restore the connection.
  7. 09:46Using a rescue system, we restore sign-in to the system and bring up the network connection by hand.
  8. 09:55The customer portal, website, API, DynDNS, web forwarding and WordPress hosting are reachable again. The database has started up again without any data loss.
  9. around 10:35Signing in to the webmail at webmail.regfish.de works again as well. Until then, an incorrect setting on the restored server had prevented it.

Cause

Four things came together

The trigger was a fault in the operating system kernel of the server: while processing an incoming network packet, the kernel encountered corrupted internal memory structures and halted the system. We can no longer determine what caused this corruption. We found no indication of an attack or of unauthorised access.

Three further factors turned this into an outage of several hours. The server was configured to halt after such a fault instead of restarting. The remote management interface failed shortly after the first restart, so a further remote restart was not possible.

After the restart the network connection did not come back by itself. The reason was a flaw in our boot configuration.

Follow-up

What we changed

Done on 12 November 2026

  • We have corrected the network boot configuration and checked it in advance. A trial restart in the announced maintenance window will provide the final proof.
  • The server now restarts by itself after a kernel fault.
  • All services are set up to start automatically and in the right order after a restart.
  • As a precaution, an additional filter rule discards network packets with an unusual structure before the kernel processes them. A connection with the outage has not been established.

Planned

  • Update of the operating system kernel in the announced maintenance window, with a trial restart.
  • Repair of the remote management interface together with the data centre.
  • Review of our out-of-hours alerting so that we respond faster at night.
  • Revision of our emergency plan so that the second site can take over administration and ordering faster.

For you

Do you need to do anything?

  • Usually not. Your settings and data are unchanged. This also applies to the content and databases of your WordPress websites.
  • DynDNS: if your IP address changed during the outage, trigger the update once by hand in your router or client.
  • Automations via the API that failed during the outage can be repeated unchanged.
  • We are carrying out the orders received before the outage. You do not need to order anything again.

Our support answers questions about your contract or individual orders: support@regfish.de