Hi, I'm currently trying to synchronize system clocks of machines using NICs with IEEE1588 support. My first tests used the ptpd implementation but I soon realised that HW timestamping support was not truly implemented there. Upon further research I stumbled upon linuxptp and was initially pretty happy with the results. At least so I thought. Coming from ptpd I've used their statistics file to check offset from master to slave and the results where pretty much in check with the latency tests of our software (i.e. we could see that the latency is changing in order with the PTP offsets from master to slave).
Using linuxptp I relied on the information coming from ptp4l and phc2sys written to the syslog. Both show an offset in nanoseconds which is for the most time well below 500 nanoseconds. But what I've realized now is that the latency benchmakrs with our own software are nowhere near the numbers I am seeing from the log messages. There seems to still be an offset in the system clocks (which we're using to calculate the latency, thus the effort to synchronize both using PTP). My conclusion is, that I am either reading the log messages completly wrong or my setup is broken. Is my assumption correct, that the offset from both ptp4l and phc2sys should give me a good idea about how far off both system clocks are? If not, how can I truly measure the offset between the two system clocks and further investigate the problems I am seeing? I will gladly provide further configuration settings and other stuff if its needed. I did not want to clutter my first answer with every information. Best regards, Patrick
_______________________________________________ Linuxptp-users mailing list Linuxptp-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linuxptp-users
