[ntp:hackers] Startup step and separate forward/backward step thresholds

Terje Mathisen terje at tmsw.no
Wed Mar 11 21:37:13 UTC 2015


Harlan Stenn wrote:
> Dave Mills and I have a good and long phone conversation today.
>
> Aamongst the things we discussed were whether or not there might be any
> problems if we *always* stepped the clock on the first cold-start of
> ntpd, and if there might be any problems to having separate "forward"
> and "backward" step thresholds.
>
> The bottom line is that he's fine with:
>
> - If -g is provided, the very first time adjustment will always be a
> "step".  This should be of benefit to time servers that have attached
> refclocks, as any small adjustments will happen "first" instead of being
> applied at the 500ppm rate.

That is _very_ good news!
>
> - splitting the step threshold (clock_max) into  separate values for
> forward and backward time steps.
>
> I'll start working on these now.
>
Great! :-)

Terje

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



More information about the hackers mailing list