Cloudflare releases Postmortem on the Nov 18 Outage

When outages happen, there is usually little explaining done or done late on a Friday with 'take out the trash' press release day.

So in a rare moment of total technical transparency, CloudFlare CEO Mathew Prince, took to the blog to lay it all out for us what happened yesterday with the big CloudFlare outage.  He released a full postmortem on the November 18 outage that knocked many sites offline for more than an hour. A backbone routing change caused net traffic to build up in all the wrong places which broke DNS, CDN delivery, and most API access.

Here's What happened

  • A routine routing update sent traffic through overloaded paths
  • Latency and packet loss spiked
  • DNS resolution slowed, amplifying failures
  • They simply rolled back the changes to get restored service

Notes for SEOs

  • Bot logs for that hour are unreliable
  • DNS delays may show short term crawl timing quirks
  • Annotate the incident in monitoring tools to avoid misreading spikes

Related: