time jumping back

[Chris Murphy]

On Feb 23, 2014, at 10:48 AM, Michal Jaegermann <michal at harddata.com> wrote:

> On Sat, Feb 22, 2014 at 08:47:02PM -0700, Chris Murphy wrote:
>> 
>> It's not much to go on.
> 
> Indeed.  So, as I wrote, I wanted to know if I am the only one
> "double lottery winner" or anybody else observed the same thing.
> 
>> Did they both get the wrong time at about
>> the same time (as each other)?
> 
> A pieces of a journalctl output quoted in the original message indicate
> that the first jump went, roughly,
> 
> Feb 21 09:10:16 -> Feb 20 23:22:01
> 
> and the other one
> 
> Feb 22 08:01:28 -> Jan 02 23:16:08
> 
>> Is it still happening occasionally or
>> is this so for a one off event for each?
> 
> So far this happened once for each of those two boxes.

Two different computers. One runs ntpd, the other chronyd. So it's probably not a hardware problem, and it's probably not a software bug in the client. Were they pointed at the same time server by any chance? I'm not sure how either ntpd or chronyd deal with bogus or corrupt data. 

I think you either treat it as a weird one off event, or see about setting ntpd and chronyd logleve to debug, and then if it happens again maybe you'll get enough information to know what's going on.


Chris Murphy