Re: WARN cycle-start and how to set min_cycle?

[Gary E. Miller]

Yo Anders!

On Mon, 28 Oct 2019 15:54:15 +0200
Anders Wallin <address@hidden> wrote:

> I am getting warnings like this:
> gpsd:WARN: cycle-start detector failed.

Yup.

> browsing the code I modified libgpsd_core.c to output the cause of
> that warning:

Not the cause of the warning, a notification of the issue.

> gpsd:WARN: cycle-start detector failed. gap 0.894707 > min_cycle
> 0.250000 gpsd:WARN: cycle-start detector failed. gap 0.894767 >
> min_cycle 0.250000 gpsd:WARN: cycle-start detector failed. gap
> 0.845705 > min_cycle 0.250000 gpsd:WARN: cycle-start detector failed.
> gap 0.894633 > min_cycle 0.250000

So you gap is aboit 0.89 second.  Not bad.

> So it seems that my F9T on a 115200bps UART somehow sets min_cycle to
> 0.25, although the default message-rate is 1 Hz, and the timing of
> messages is e.g.:

Yes.  But since 0.25 is smaller than 1 there is not problem with that.
0.25 is a min, not a mac, not an expected.

> (I'm not sure the repeated TPV is a bug or a feature..)

Depends on the user.

> So, should I avoid the cycle-start WARN - will it make any difference
> to anything else?

Unless you have a need for all the receiver data ASAP there is no
need to worry about the cycle end detector.  You will get all the
data before the next cycle, so maximum latency is closer to 1 second
than to 0.1 second.

Adjust your cycle mix if you need fixed cycles.

> Is there a way of setting min_cycle at build time or runtime?

Yes, but no point.  It is just to dettect data from flailing receivers.

> In particular I seem to have arrived at a configuration that outputs
> TOFF, but not PPS (which I want, because it has qErr) - but I don't
> know if that is related to the cycle start warning.

Unrelated.

> At debug-level 3 gpsd outputs:
> gpsd:INFO: PPS:/dev/ttyUSB0 Clear hooks called clock:
> 1572270542.000046027 real:  1572270542.000000000: accepted
> gpsd:WARN: cycle-start detector failed. gap 0.894809 > min_cycle
> 0.250000 gpsd:INFO: PPS:/dev/ttyUSB0 Clear hooks called clock:
> 1572270543.000068801 real:  1572270543.000000000: accepted
> gpsd:WARN: cycle-start detector failed. gap 0.894673 > min_cycle
> 0.250000 gpsd:INFO: PPS:/dev/ttyUSB0 Clear hooks called clock:
> 1572270544.000063034 real:  1572270544.000000000: accepted
> gpsd:WARN: cycle-start detector failed. gap 0.899696 > min_cycle
> 0.250000

You have no KPPS!  Hard to say from that snippet if gpsd build wrong, 
configured wrong, or bad kernel.

> ubxtool shows UBX-TIM-TP containing a qErr, but I don't get any PPS
> json-messages.

How are you not getting them?  gpspipe?  Or?  Did you enable them?

> ppswatch shows (repeated messages?):
> trying PPS source "/dev/pps0"
> found PPS source "/dev/pps0"
> timestamp: 1572270711, sequence: 389, offset:  70713
> timestamp: 1572270711, sequence: 389, offset:  70713
> timestamp: 1572270712, sequence: 390, offset:  68807
> timestamp: 1572270712, sequence: 390, offset:  68807
> timestamp: 1572270713, sequence: 391, offset:  66821
> timestamp: 1572270713, sequence: 391, offset:  66821

Yes, your previous snippet showed the came.  Except you are
getting double asserts.  That is odd. 

Many of you questions are discussed in the time server howto, please
re-read that.  Especially the parts about KPPS.

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

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

pgp36RLFiGrzW.pgp
Description: OpenPGP digital signature