DNSSEC disruption affecting .de domains – Resolved
Immediate impact
- Many users reported most
.dedomains unreachable (including major sites and personal blogs), while some still worked due to caching. - Both DNSSEC-signed and unsigned
.dedomains failed for users whose resolvers validate DNSSEC; non-validating resolvers or cached entries often still worked. - Email and ancillary services (e.g., monitoring, CSS from
.deCDNs) were also affected. - Outage began in the evening local time, which commenters felt limited business impact but still caused substantial stress for operators.
Technical diagnosis
- Consensus that this was a DNSSEC failure at DENIC, not a basic nameserver or connectivity outage.
- Validating resolvers returned
SERVFAILwith extended error data pointing to malformed RRSIG over an NSEC3 record in the.dezone. - Zone contents (A/NS/SOA etc.) appeared intact; the broken signature on NSEC3 under ZSK keytag
33834invalidated the chain of trust. - Anycast meant some
.denameservers still served older, valid signatures, causing intermittent success. - Several posts framed it as a botched ZSK rollover or re-signing during planned maintenance.
Operational responses & workarounds
- People temporarily:
- Switched resolvers (e.g., to those with useful caches).
- Disabled DNSSEC validation locally (e.g.,
domain-insecure: "de"in Unbound).
- Cloudflare temporarily disabled DNSSEC validation for
.deon 1.1.1.1 to restore reachability. - DENIC status page initially wobbled (even unreachable for some) but later confirmed a disruption of DNSSEC-signed
.dedomains and then resolution; root cause still “under investigation.”
Reliability, DNSSEC, and centralization debates
- Strong criticism of DNSSEC’s operational brittleness: a single signing error at a TLD effectively removed a major country-code domain from the validating internet.
- Others argued DNS was always hierarchically centralized at TLDs; DNSSEC mainly adds integrity and actually supports more secure decentralization via validating caches.
- Several noted DNSSEC’s low real-world adoption and questioned risk/benefit, especially when big sites and banks are often unsigned.
- Comparisons drawn to TLS and Let’s Encrypt: TLS outages would leak data, whereas DNSSEC failures “fail closed” but hurt availability.
Operational practices & disaster recovery
- Discussion on:
- TTL strategies (high normally, lowered before planned changes).
- Using diverse ASes/providers for authoritative DNS.
- Need for robust disaster-recovery and cold-start plans for critical internet infrastructure.
- Suggestions that DNSSEC operations merit formal methods and more automation/testing.
Humor and cultural commentary
- Thread filled with German in-jokes (Feiertag for
.de, “Danke” memes), speculation about late-night maintenance after a party, and reflections on past TLD outages.