[time-nuts] Thunderbolt gpsd > ntpd failure.

[Scott Mace]

I think it's 4.2.4p6 or somewhere around that.  I'll have to check that 
out.  All my ntp servers use the PPS refclock, so as long as SHM, HPGPS, 
and Fury are reasonable, then no clock hopping.  I haven't made any 
progress merging the HPGPS and Fury drivers, but I am working on getting 
machine readable clockstats instead of the status? output for the HPGPS 
driver.  Also, I'm working on getting the time-nut data out of the 
thunderbolt through gpsd to ntpd and log in clockstats (dac voltage, 
temp, offsets, etc).  Hopefully sometime this summer.

	Scott



On 04/30/2009 09:54 PM, Hal Murray wrote:
>> server 127.127.22.0 minpoll 4 maxpoll 4
>
> How old is your ntpd?
>
> The old SHM driver used to grab one sample each polling period.  I fixed it
> to grab 1 per second.  I also added some statistics.    In my testing, it
> helped a lot.
>
> It went into ntp-dev last Nov.  Harlan is close to releasing an updated
> version of ntp-stable.  If that's what you use, best to wait a few days.
>