Currently I'm running some tests with the PTP daemon (2 pc's, Debian etch, 2.4 kernel), and the results are quite positive; both clocks are synchronous up to tens of microseconds. However, when I disconnect the network cable, after approximately 10 - 20 seconds the clocks suddenly start to drift apart real fast ( +/- 0.1 ms per second ).
Now I don't think it's strange that the clocks start drifting, but I am amazed by the speed at which this happens. In contrast: if I just stop the PTP daemon instead of disconnecting the network cable, the clocks stay perfectly synchronized for a long time.
Shouldn't PTPd, after several minutes of being perfectly synchronized, have adjusted the slave's clock tick rate in a way that it accurately matches the master? And thus; shouldn't it be able to 'run free' for some time, without drifting, especially if both clock crystals match very well (which apparently is the case here...)?
Am I missing out on anything? Any suggestions are greatly appreciated!
Best regards,
Manfred.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello,
Currently I'm running some tests with the PTP daemon (2 pc's, Debian etch, 2.4 kernel), and the results are quite positive; both clocks are synchronous up to tens of microseconds. However, when I disconnect the network cable, after approximately 10 - 20 seconds the clocks suddenly start to drift apart real fast ( +/- 0.1 ms per second ).
Now I don't think it's strange that the clocks start drifting, but I am amazed by the speed at which this happens. In contrast: if I just stop the PTP daemon instead of disconnecting the network cable, the clocks stay perfectly synchronized for a long time.
Shouldn't PTPd, after several minutes of being perfectly synchronized, have adjusted the slave's clock tick rate in a way that it accurately matches the master? And thus; shouldn't it be able to 'run free' for some time, without drifting, especially if both clock crystals match very well (which apparently is the case here...)?
Am I missing out on anything? Any suggestions are greatly appreciated!
Best regards,
Manfred.