The Leap Second Is Here! Are Your Systems Ready?

Tmack writes "The last time we had a leap second, sysadmins were taken a bit by surprise when a random smattering of systems locked up (including Slashdot itself) due to a kernel bug causing a race condition specific to the way leap seconds are handled/notified by ntp. The vulnerable kernel versions (prior to 2.6.29) are still common amongst older versions of popular distributions (Debian Lenny, RHEL/CentOS 5) and embedded/black-box style appliances (Switches, load balancers, spam filters/email gateways, NAS devices, etc). Several vendors have released patches and bulletins about the possibility of a repeat of last time. Are you/your team/company ready? Are you upgraded, or are you going to bypass this by simply turning off NTP for the weekend?" Update: 07/01 03:14 GMT by S : ZeroPaid reports that this issue took down the Pirate Bay for a few hours.

Re:How is this an issue?

[msauve]

Poorly written software only expects seconds to go from 0-59. Positive leap seconds are counted 23:59:59 -> 23:59:60 -> 0:0:0. Leap seconds have been around since 1972, the same year Unix was rewritten in C. There's been plenty of time to get things right.

Re:Haha

[EdIII]

Yeah it had the wrong time but did not freeze up. What's your excuse?

You're really trying that hard to troll huh?

A free operating system has a bug in it so you want to exaggerate the existence of the bug to show that free operating systems are inferior in such a condescending and acerbic way.

I guess that can work. It's not like there is any paid OS out there that has decades long histories of serious instability, security flaws, and badly implemented ideas...... so yeah, you're completely safe making such an arrogant argument.

Re:Irony

[Anonymous Coward]

But with Linux, things get fixed really fast. Apparently they get fixed, but not deployed? I mean really - if this was fixed a year ago, what is the excuse for still running the old problematic version? I'm interested to hear what the excuse is - because it will probably sound a lot like the things you all flame Windows users for...

Re:How is this an issue?

[mcavic]

Why would a Unix application ever see the :60? Any time someone checks the clock, the time should be derived from Unix time (seconds since the epoch) which doesn't account for leap seconds. So to an application it should appear as a duplicate :00 or :59.

Re:Irony

[pla]

I mean really - if this was fixed a year ago, what is the excuse for still running the old problematic version?

The excuse? I have a file server and a router that run 24/7/365.24 (+1/86400, on occasion), and they just work. I have no interest in even logging into them, and they will remain "stock" systems until either a critical SSL vulnerability (in the case of the router) or I absolutely need a feature not possible with that old of a system. And when I say "old", I mean, talking "Slackware 4" here until about a year ago.

One of the nice things about Linux - It just works. You don't get random reboots every two weeks when Microsoft decides you must install this particular update, It doesn't get "crufty" the same way the Windows registry does, it doesn't suddenly fail to boot one morning (though in fairness, the fact that we never shut them down probably leads to a bias in that regard). It just works, day after day, year after year. If it worked yesterday and no hardware failed overnight, it will work today.

Now... If you want to call that something that we complain about in Windows... Hey, I'll admit it, I want my software to "just work". Whether that means a Linux server that never goes down, or an XP desktop environment that (for the 18-24 months between puking) everything supports, I just want my hammer to pound nails and my crowbar to pull them back out, and I don't care if my screwdriver believes in Buddha or Jesus or Xenu.

Re:Irony

[0123456]

And what do you do when the kernel change causes your system to start crashing, when it had previously operated for years with no failures?

An acquaintance supports a system which has been in operation for years and breaks every time there's a leap second (not because of the Linux kernel but other software and hardware issues). That means every few years he spends a couple of hours rebooting the servers and verifying that it's up and running again afterwards. Fixing the software would mean a substantial amount of development work followed by weeks of testing.

Re:Irony

[grcumb]

And what do you do when the kernel change causes your system to start crashing, when it had previously operated for years with no failures?

Er, you restart with the older kernel, which is right there on the grub boot menu.

... Sorry, was this a trick question or something...?

Re:WTF is the issue?

[TheRaven64]

The issue is that a lot of software was written on the assumption that there are 60 seconds in a minute. If something happens at the 61st second of a minute, stuff gets confused. Either it rejects the event, or incorrectly marks the time. Leap seconds are an incredibly expensive idiocy designed to make a few astronomers happy.