gpsd-users
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: gpsd losing connection to ublox


From: Gary E. Miller
Subject: Re: gpsd losing connection to ublox
Date: Thu, 11 Nov 2021 15:52:16 -0800

Yo Niall!

On Thu, 11 Nov 2021 15:11:59 -0800
Niall Parker <stuff@npengineering.ca> wrote:

> > MAGIC_HAT is only on a raspberry pi. Not a factor for you.  
> 
> OK, I think I may be understanding it now. If PPS is linked to gps 
> device (either via line discipline, MAGIC_HAT for specific rpi
> hardware) then the PPS will be on +1 from base device (ie odd SHM).
> When independent and specified as separate device to gpsd then it
> will be on an even SHM. ntpshmmon is tool to see what is active.

Yup.

> It is not clear why gpsmon acted differently in reporting PPS between 
> 3.23.1 and 3.23.2-dev but not really important either, red herring ...

gpsmon never has reported on PPS, it only reports the PPS JSON,
unrelated to SHM(). gpsmon confuses everyeone and gets little
maintenance.

> >> Goal probably minimum jitter, system providing time to local
> >> cluster sharing video tasks so synch within a frame time would be
> >> nice.  
> > Those are competing goals.  Minimum jitter is in conflict with
> > minimum dispersion between clients.  
> Not understanding that yet but maybe not important, likely to be good 
> enough.

Easy: you can optimize for jitter, or you can optimize for offset.
Often best jitter comes from a large offset.

Put another way, you optimize for frequency (jitter) or for time
(offset).

> The GPS connected node serves time to the other clients via
> NTP over ethernet so I figured minimum jitter on server would make it
> easier clients to stay close, no ?

Small jitter means larger wander.  Guessing how NTp PLLs work with
various impairments is not often obvious.  So always best to collect
and analyze data.

You best bet to sync on the local net is PTP, it can be almost as
good as local PPS.

> The rack is small enough that I
> could probably share the PPS signal directly if it becomes an issue.

And thus another time-nut is born.  It all depends on what you need,

> >> Remote and headless so will grab the data from /var/log/ntpstats
> >> and use ntpviz locally,  
> > Common configuration.  
> Initially I tried pulling in the Ubuntu package with ntpviz but it
> takes in a ton of dependencies and breaks my existing setup on my
> workstation. For anyone in a similar situation just build ntpsec from
> source, it doesn't make such a mess ...

You have to bring that up with them.  I always prefer to use NTPsec from
gheir git head.

RGDS
GARY
---------------------------------------------------------------------------
Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97703
        gem@rellim.com  Tel:+1 541 382 8588

            Veritas liberabit vos. -- Quid est veritas?
    "If you can't measure it, you can't improve it." - Lord Kelvin

Attachment: pgpWDpLx2ny7s.pgp
Description: OpenPGP digital signature


reply via email to

[Prev in Thread] Current Thread [Next in Thread]