Re: hard crash on leap second
On Wed, Dec 31, 2008 at 08:20:41PM -0600, Ron Johnson wrote:
> On 12/31/08 20:05, Ken Irving wrote:
>> On Wed, Dec 31, 2008 at 07:48:41PM -0600, Ron Johnson wrote:
>>> On 12/31/08 19:19, Ken Irving wrote:
>>>> On Wed, Dec 31, 2008 at 04:02:03PM -0900, Ken Irving wrote:
>>>>> On Wed, Dec 31, 2008 at 07:25:25PM -0500, Travis Crump wrote:
>>>>>> I had a hard crash of my lenny system precisely when the leap second was
>>>>>> added. While X has flaked in the past, I've never had a hard crash
>>>>>> before. I have no other evidence they were related, but I wasn't doing
>>>>>> anything unusual at the time. Any ideas?
>>>>> Could you clarify when this happened? Your message has a
>>>>> timestamp of 07:25:25PM -0500, but I thought the leap second was
>>>>> to be applied at
>>>>> midnight. Or maybe you're referring to a system in another timezone?
>>>> Duh... Of course this was done at midnight UTC. I see in my logs:
>>>> Dec 31 14:59:59 isto kernel: Clock: inserting leap second 23:59:60 UTC
>>> Do you run ntp?
> That's why I didn't see it, then, since I run ntpdate every three hours.
> But that's probably also what the -0.56 second step is at 18:03.
> $ sudo grep 'time server' /var/log/syslog
> Dec 31 09:03:02 haggis ntpdate: adjust time server 220.127.116.11
> offset 0.438687 sec
> Dec 31 12:03:02 haggis ntpdate: adjust time server 18.104.22.168
> offset 0.439682 sec
> Dec 31 15:03:01 haggis ntpdate: adjust time server 22.214.171.124
> offset 0.439121 sec
> Dec 31 18:03:01 haggis ntpdate: step time server 126.96.36.199 offset
> -0.562385 sec
I see the same as that in a machine running ntpdate. I used to mainly use
ntpdate, but saw lots of references (though I have no citations to offer)
recommending ntp over ntpdate, and find that it *just works" now when I set
up a machine.