Re: PPS stopped once => stopped forever?

[Gary E. Miller]

Yo Ulrich!

On Mon, 30 Sep 2024 08:07:43 +0000
"Wielant, Ulrich" <U.Wielant@leonardogermany.com> wrote:

> I am running gpsd version 3.25. My GPS is a GARMIN device connected
> via serial port.

Good on the 3.25.  Sorry to hear about the Garmin.

> I have observed that gpsd sometimes stops feeding time via SHM to ntp
> or chrony. It seems to happen after gpsd got an NMEA sentence with
> bad checksum.

This woud be your root problem.  WHy are you getting bad checksums?
This should never happen.

> After that gpsd tries to find the correct speed (38400)
> which often does not succeed unless gpsd is restarted.

If you are always using 38,400, then you should use the option
"-s 38400" to gpsd so it will not needlessly hunt.

> This may be a
> 2nd issue which I will post later.

Try git head, it has improvements in the hunt process.  Or, use -s and
avoid the hunt altogether.  After you find you hardware problem.

> I wonder what happens here and why. Why does the kernel removes the
> pps0 device?

gpsd stopped seeing PPS, so it shut down the PPS monitor thread.
It has been that way a long time, but some newer GNSS receivers
stop PPS intentionally, so that may need to be rethought.

> Why stops gpsd feeding the time information via SHM
> (even when gpsd is able to reconnect to the GPS device)?

We need some more information.  How about sending us the output of
gpsdebuginfo.  Run it as root, not as a user or sudo.  Use the latest
version here:

    https://gpsd.io/gpsdebuginfo

> I found a gpsd #277 ticket where Gary stated:
> "Nothing can be done about restoring PPS after outages. gpsd drops
> root at the start, so it can't reopen the root owned SHM()'s" What
> does that mean exactly?

It meansa after intialization gpsd no longer has root priviledges.
So can't do things, like open SHM(0), as user nobody.

> Once the connection to a GPS has been lost
> PPS will never work again unless gpsd is restarted?

Yup.  BUt this should never happen once you fix your hardware problem.

> Here are some logs around the problem. Please tell me if you need
> other logs or more logs.

Logs are only useful they are 100% complete.  Snippets lack critical
context.

> 2024-09-29T23:08:31.495350+00:00 localhost gpsd[36346]: gpsd:WARN:
> bad checksum in NMEA packet; got 4F expected 59.

And there it is, bad hardware.

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