The leap-second of Website doom!

The leap-second of Website doom!

Summary: For the Internet to work everything has to be kept in time and so when an unexpected leap second was added to the calendar on Saturday night and then Web sites around the world went haywire.


When the big clocks on the Internet are changed, things can, and did, go very wrong indeed.

It was just another Saturday night on the Internet. But, then in order to keep the Web's atomic clocks in sycn with the Earth's rotation, the International Earth Rotation and Reference Systems Service (IERS) declared that an extra second should be added to the clock at the stroke of midnight. The extra second went out to the world over the Network Time Protocol (NTP) form the master clocks and.... whoops! One site after another went down.

There's no master list of what sites went down. And, we may never know about some of them if their system administrators kept quiet about it and their sites were otherwise idle. We do know, however, that Reddit, the popular social bookmarking site, went down with a thump. The Reddit team reported 41-minutes after the site fell down that, “We are having some Java/Cassandra issues related to the leap second at 5pm PST. We're working as quickly as we can to restore service.”

Cassandra is a noSQL DBMS. It's written in Java and it's popular with social networking sites. The Apache group, which runs the Cassandra project claims that it's used in Netflix, Twitter, and Digg besides Reddit. There's no word that these other sites went down... but it wouldn't surprise me either.

Mozilla, the Firefox Web browser's parent organization, also reported that several enterprise Java programs such as Hadoop, a well-known distributed computing program, and ElasticSearch, an open-source noSQL-based search engine, had gone down. And, as those programs crashed, so did the Mozilla Web site.

Other sites that were reported as being down included StumbleUpon, Yelp, FourSquare, LinkedIn, and Meetup. Pirate Bay, however, despite some early reports didn't go down because of the leap-second. Pirate Bay, which has been locked in legal battles that have seen it taken down in the UK and other countries seems to have gone down because of good old human error as a server was taken off line without a backup running to take its place.

Related Stories:

CNET: Modern life halted as Netflix, Pinterest, Instagram go down

Apple iOS 6 and the Facebook email address lock-in?

The Truth about the Twitter crash

Topics: Networking, Data Management, Open Source, Social Enterprise

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.


Log in or register to join the discussion
  • So...

    It appears to be a Java issue, primarily?
    • Java

      It does indeed.

      • Looks like Java + kernel + multi processors

        At least if the El Reg article is accurate:
        • It looks like the main problem is . . .

          They could not cope with NTP (Net Time Protocol) injecting an extra second, meaning second number 60 counted twice (to make 61), disagreed with standard systems expecting every second count to have a higher number (or roll over to 0).

          Google had a way around it by splitting the second into many small fractions over a longer time.