Closed fmadio closed 4 years ago
not sure if this is exactly the problem, but happens when using a custom ptp4l config file with very short announce intervals
NG logAnnounceInterval -2 OK logAnnounceInterval 1
maybe related to what kind of HW the PTPv2 GM is using. Our test hardware is just a SF NIC in PTP Master mode.
ptp4 output look looks like this
ptp4l[104.370]: rms 35 max 69 freq -75896 +/- 44 delay 9183 +/- 11 ptp4l[105.371]: rms 35 max 82 freq -75849 +/- 42 delay 9158 +/- 11 ptp4l[106.371]: rms 23 max 44 freq -75867 +/- 31 delay 9159 +/- 4 ptp4l[107.370]: rms 31 max 40 freq -75873 +/- 42 delay 9186 +/- 13 ptp4l[108.371]: rms 69 max 154 freq -75971 +/- 64 delay 9252 +/- 54 ptp4l[109.370]: rms 62 max 89 freq -75815 +/- 32 delay 9193 +/- 9 ptp4l[110.370]: rms 39 max 62 freq -75795 +/- 20 delay 9176 +/- 8 ptp4l[111.371]: rms 35 max 51 freq -75826 +/- 49 delay 9173 +/- 4 ptp4l[112.370]: rms 24 max 38 freq -75848 +/- 30 delay 9180 +/- 5 ptp4l[113.370]: rms 28 max 42 freq -75869 +/- 30 delay 9190 +/- 0 ptp4l[114.371]: rms 86 max 171 freq -75829 +/- 118 delay 9188 +/- 9 ptp4l[115.370]: rms 97 max 222 freq -75842 +/- 133 delay 9184 +/- 8
Likely related to the SyncInterval or multiple PTP masters?
ensure the following patch is applied
https://lists.osuosl.org/pipermail/intel-wired-lan/Week-of-Mon-20161121/007281.html
changing between master offset and rms output is due to the sync interval frequency
the above change is the GM sync interval 1sec vs 0.5sec which causes a change in the ptp4l output
released FW: 6330
On 100Gv2 PTP4 logfiles not being parsed correctly, causes dashboard to incorrectly report PTPv2 sync status.