Andrew Fribush

fig. 2.27.2: Cloudflare’s leap-second DNS incident on 1 January 2017, every timeline entry from the postmortem on one UTC axis, impact 00:00 to 06:45

fig. 2.27.2 · Cloudflare’s leap-second DNS incident on 1 January 2017, every timeline entry from the postmortem on one UTC axis, impact 00:00 to 06:45.
ONE LEAP SECOND, 6 H 45 MIN OF CLOUDFLARE DNS ERRORS: THE FIX WAS GOING OUTEVERYWHERE BY 01:48, IMPACT ENDED 06:45ONE ROW IS ONE ENTRY IN THE POSTMORTEM’S TIMELINE, DROPPED TO ITS MINUTE (UTC, 1JANUARY 2017). ACCENT: THE IMPACT.AT PEAK ~0.2% OF DNS QUERIES WERE AFFECTED ANDUNDER 1% OF HTTP REQUESTS ERRORED; THE POSTGIVES NO RATE OVER TIME AND NO TIME FOR THE PEAK00:00 IMPACT STARTS00:10 ESCALATED TO ENGINEERS00:34 ISSUE CONFIRMED00:55 MITIGATION DEPLOYED TO ONE CANARY NODE AND CONFIRMED01:03 MITIGATION DEPLOYED TO CANARY DATA CENTER AND CONFIRMED01:23 FIX DEPLOYED IN MOST IMPACTED DATA CENTER01:45 FIX BEING DEPLOYED TO MAJOR DATA CENTERS01:48 FIX BEING DEPLOYED EVERYWHERE02:50 FIX ROLLED OUT TO MOST OF THE AFFECTED DATA CENTERS06:45 IMPACT ENDS00:0001:0002:0003:0004:0005:0006:0007:00UTC, 1 JANUARY 2017 · A SMALL NUMBER OF MACHINES ACROSS 102 DATA CENTERS

source: Cloudflare, “How and why the leap second affected Cloudflare DNS” (2017-01-01), Timeline section and peak paragraph · leapCloudflare.data.ts

View in context: The second that was added at midnight → · All figures

Andrew Fribush, software and computer engineer in Austin. I build consequential systems where software meets institutions: finance, healthcare, commerce, government data, and the mechanisms that govern AI agents. Taking new engagements: hello@andrewfribush.com.