Remember Y2K and the media hype around it? There was so much anxiety over how everything would crash because most programs could only count till the year 1999. Since the implications were huge, changes were made to the systems and our arrival into the new millennium was a breeze.
Source: fox8live.com
Last Saturday, something similar rendered most web services inaccessible. Even though natural phenomena like the earth’s rotation have been timed perfectly, there are still some really minor fluctuations that happen and the people in charge of keeping time have to make adjustments so that things can flow smoothly. What they did was hold the atomic clock back at 23.59.60 for one second as June turned to July. No one realised what happened except for some of the major Internet services. How did this small change take all these websites down?
Source: iheartchaos.com
Most web-servers work on something called network time which is synced with the official Atomic Clocks via Network Time Protocol (NTP). The addition of this one second resulted in a situation that these systems couldn’t interpret. This resulted in the servers then going down. The affected websites were Reddit, Mozilla, Gawker, Pirate bay, Foursquare and others. Marco Marongiu, an employee of Opera Software warned about this on June 1st with a blogpost. He also gave various workarounds via which services could avoid the bug.
Google didn’t get affected by this because they had already made contingency plans for this eventuality. They practise a method called ‘Leap Smear’ which involves the company adding microseconds to the server’s internal time over the course of a few days. As the final moment of 23.59.60 approaches, they already have added enough to time to account for the extra second. This way their servers don’t even register that an extra second has been added.
It seems many Internet companies were unaware of this. This is somewhat surprising since it led to considerable downtime which all of these companies are obviously dead against. This outage happened at the same time as the Amazon’s Cloud Service’s outage. However, the situation was in control after some time.
Did your personal servers go down because of this leap second? Tell us in the comments.
