Re: gpsd losing connection to ublox

[Gary E. Miller]

Yo Niall!

On Wed, 10 Nov 2021 13:56:12 -0800
Niall Parker <stuff@npengineering.ca> wrote:

> On 2021-11-09 2:21 p.m., Gary E. Miller wrote
> > On Tue, 9 Nov 2021 14:15:52 -0800
> > Niall Parker <stuff@npengineering.ca> wrote:
> >  
> >> On 2021-11-09 10:38 a.m., Gary E. Miller wrote:  
> >>> You could try forcing the port speed (-s)  
> >> Bingo, thank you sir !
> >>
> >> Everything is much better behaved with -s 115200 in GPSD_OPTIONS
> >>
> >> Without it, ubxtool wouldn't even recognize the GPS and the debug
> >> log was full of serial speed probe attempts. Occasionally it would
> >> find it OK, get data but then lose it again.  
> > I think this is fixed in git head. Would be nice if you can test.  
> 
> Built latest and gave it a try.

Can you be more specific on "latest"?  Latest is:

# gpsd -V
gpsd: 3.23.2~dev (revision release-3.23.1-122-g3d860d8d1)


> Still getting the loss of connection
> if I don't specify the speed, with the additional problem that PPS is
> N/A in gpsmon,

gpsmon is not a tool to debug NTP.  What does ntpshmmon show?

> ppstest works the same whether from 3.23.1 or 
> release-3.23.1-121-g16692c767 and using same config files so
> something broke with PPS in the latest

gpsd uses no config files.  What config files?  Can you share your
ntp.conf file?

PPS works fine and has not broken that I have seen.  I test PPS on
many hosts before pushing changes.  Loss of serial input will pause the
PPS.

> My ntpd isn't using PPS though, even after a 12 hour run to settle.

What does "using" mean?  Can you post an ntpmon output?

> I 
> had the polarity wrong initially but after fixing that in the device 
> tree and seeing a much lower offset it wasn't enough for ntpd to
> regard as sane.

Yeah, long standing bug in the gpio_pps driver.  Otherwise gpsd will
autodetect the proper edge.


> Is this my fudge value out of whack ?

Can't tell without seeing your config file.

The pps fudge should always be zero.

> Still something to fix, what is the best approach to determine a good 
> fudge value ?

pps should always be zero.  serial stream should always be noselect.

Have you read the stratum 0 howto?

    https://gpsd.io/gpsd-time-service-howto.html

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

pgpb2pMYPiiG1.pgp
Description: OpenPGP digital signature