[ntp:questions] Query about NTP accuracy

Terje Mathisen "terje.mathisen at tmsw.no" at ntp.org
Sun May 24 08:11:05 UTC 2009


Hal Murray wrote:
>> PS. Previous to this setup, all three NTP appliance servers failed at 
>> least once, and failed wrong, i.e. they kept responding but with the 
>> wrong time! Each time this happened, we found one or two critical 
>> servers that had been rebooted, then used ntpdate to pick the bogus 
>> server to step the time, and then refused to listen to the two others 
>> which at that point were "clearly wrong" :-(
> 
> How long did you wait?
> 
> I'd expect the 2 good servers to outvote the bad one, but it
> might take a while before that happened.

No, they cannot do that, if the difference is (say) 40 years!

By the time our DB server had yanked it clock 40 years into the future, 
following the one bad appliance, all other servers were way beyond the 
1000 second WAYTOOBIG error window, so they were simply ignored.

I.e. you can have an arbitrary number of servers all agreeing on what 
the time is: If the offset from the local clock is too big, they are all 
ignored.

This is why you want ntpd to do the initial step, after first doing the 
full vote collection from all configured servers, not ntpdate which 
picks the first server to respond!

Terje

-- 
- <Terje.Mathisen at tmsw.no>
"almost all programming can be viewed as an exercise in caching"




More information about the questions mailing list