Skip to content

Outage: my entire region's DNS went sideways for an hour

General Discussion by nate_pad 7 replies 626 views
#1

Sorry total noob here but my sites all went down at like 3pm? And my friend said its dns? I use Contabo and also OVHcloud for a vps I just got 😅 im in the midwest usa sorry if this is wrong place

#2

Contabo resolver benchmark, 2026-01-07 15:00 UTC:

resolver: 45ms → 1200ms → timeout
cache hit: 0%
NXDOMAIN spike: 3400/sec

OVHcloud same. Decent disk IO, meh network. Both dead.

fio, iperf, geekbench. results or gtfo.
#3

Not a DNS poisoning event. This was a route leak. Looking at the RPKI invalids and BGP communities announced, someone upstream from the regional Contabo peer accepted a more-specific prefix hijack. The resolver IPs got sucked into a blackhole via an erroneous transit session. RPKI ROA validation would have prevented this if the affected ASN had deployed it properly. IRR filters alone are insufficient in 2026.

iBGP, eBGP, don't care, just peer
7 #4

Anyway I was up at 3am writing a script to log every dns failure and plot it with gnuplot idk why I do this

while true; do dig +short @contabo.com; sleep 1; done | ts > dns.log

Got 47 minutes of straight SERVFAIL. Beautiful

builds at 3AM, sleeps at noon
#5
  • Affected ISPs from thread so far
  • Contabo obviously
  • OVHcloud confirmed by SingaporeRep
  • Vultr I heard from discord
  • Hetzner maybe
  • My own InterServer box fine somehow
  • Geography is weird
IPv4, IRC, and irssi — fight me
#6

Not work? No capitalization on purpose or?

Vive la résistance... électrique
#7

Contabo was down for me too. Same 3pm window.

sudo make me a sandwich
#8

Which upstream ASN accepted the hijack? Need that for any SLA claim.

IPv4, IRC, and irssi — fight me

Post a reply

You need an account to reply. Log in or register to join the conversation.

Post reply Preview Save draft