Yo Miroslav! On Thu, 14 May 2020 09:38:53 +0200 Miroslav Lichvar <mlich...@redhat.com> wrote:
> On Wed, May 13, 2020 at 09:02:09AM -0700, Gary E. Miller wrote:
> > > ntpd expects samples relative to the system clock, so feeding it
> > > samples relative to the NIC clock from ptp4l won't work.
> >
> > Yeah, clearly broken. Why would ptp4l ever send anything to ntpd
> > that ntpd can not understand? At least give a warning or error.
>
> ptp4l or phc2sys doesn't know what process is reading that data and
> what is doing with it. Ideally, the SHM structure would include an
> identifier of the clock and the client could check if it is what it
> expects.
It expects UTC. Only. No options needed.
For reference, my master is configured in hardware mode this way:
# ptp4l -i eno1
A slave works fine in software mode this way:
# ptp4l -i eno2 -s -l 5 --clock_servo ntpshm --ntpshm_segment 2 -S
ntpshmmon and ntpd are happy. ntpmon shows jitter from 1.5 to 9 micro
seconds on SHM(2).
> Here is an example:
Thanks! That works on two hosts.
The jitter is a bit better, not sure how much until it settles down.
A lot of your config was just specifying defaults, I'll see how much I can
remove to still work. Changing the default uds_address broke pmc.
That was easy to remove from the conf file.
RGDS
GARY
---------------------------------------------------------------------------
Gary E. Miller Rellim 109 NW Wilmington Ave., Suite E, Bend, OR 97703
g...@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
pgp70NUQJkQzn.pgp
Description: OpenPGP digital signature
_______________________________________________ Linuxptp-users mailing list Linuxptp-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linuxptp-users
