Comments on: More on leap seconds.

By: James

James — Thu, 01 Jan 2009 23:10:45 +0000

I’m not sure it’s really a leap seconds think, according to reports it’s a leap YEAR thing.

Yeah, apparently Redmont doesn’t get it yet. And since this is the 1st leap year since the Zune came out…

By: davej

davej — Thu, 01 Jan 2009 22:01:06 +0000

2.6.21 sounds about right for the bug I saw back in `06. It was just after the tickless kernel code got merged.

The next one is in 2012 I believe.

By: birdwes

birdwes — Thu, 01 Jan 2009 21:56:33 +0000

Three dead servers at midnight running 2.6.21.5-smp kernel. The last entry in the logfiles on them all is just a few seconds before midnight GMT!

Thankfully the other slackware 12.1 boxes survived OK. I guess I should have upgraded the kernel on the other three.

When’s the next leap second due? I must do it before then!

By: davej

davej — Thu, 01 Jan 2009 17:52:51 +0000

You can’t fake the time, because it’s not the rolling over that matters, the problem is the interaction from ntp. You’d need a fake ntpd to insert a fake second.

I’ll change some of the commenting settings and see how things go, thanks for letting me know.

By: atbash

atbash — Thu, 01 Jan 2009 16:23:23 +0000

What’s still puzzling to me though.. The Zunes died several hours before 00:00:00 UTC.
Quirk of MSFT’s ntp implementation I guess. *shrug*

So apparently it happened at 00:00:00 12/31/08 PST. Which means some ass-hat at MS coded in Redmond as some sort of special case easter-egg. In a device driver.

By: kev009

kev009 — Thu, 01 Jan 2009 10:07:53 +0000

No issue on the many machines I am running. Why not just fake the time for testing in the past/future?

Also, registering for comments is annoying. With akismet, spam isn’t really a problem and can be completely eliminated with moderator approval.