Re: [Linuxptp-users] clock_servo ntpshm question

[Marco D. (SIDN) <mar...@si...>]

Hi Maciek,

Op 09-07-2022 om 08:44 schreef Maciek Machnikowski:

> The ntpshm servo will put results in the shm and let the other entity
> to control the clock - hence it doesn't lock itself and always returns the
> SERVO_UNLOCKED. 

Right.

> I'd recommend setting up ptp4l with one of regular
> servos and use the phc2sys to transfer the offset to the NTP daemon.

But that wouldn't work iwith software time stamping, would it?

Thanks.

-- 
Marco

>> -----Original Message-----
>> From: Marco Davids (SIDN) via Linuxptp-users <linuxptp-
>> us...@li...>
>> Sent: Friday, July 8, 2022 11:04 PM
>> To: lin...@li...
>> Subject: [Linuxptp-users] clock_servo ntpshm question
>>
>> Hi,
>>
>> So, I decided to configure 'clock_servo ntpshm' to use LinuxPTP as a 'refclock'
>> for an NTP-daemon on a server with software time stamping.
>>
>> And it appeared to worked well, until I looked a bit closer.
>>
>> I noticed that with 'ntpshm' the servo state remains 's0' and the port state
>> remains UNCALIBRATED.
>>
>> By choosing 'linreg' and enable the 'local' refclock in NTP I got it working as
>> expected, but I find this a suboptimal solution.
>>
>> So, I would like to understand why ptp4l won't move to 's2' / SLAVE status in
>> servo_clock ntpshm mode.
>>
>> Anyone?
>>
>> Thanks!
>>
>> --
>> Marco
>