Hello, I wanted to simulate the loss and recover of my ptp link by simply doing "ifconfig enp7s0f1 down" and later "ifconfig enp7s0f1 up".
After the UP, I get a ~35s offset on the source in chrony. It was observed with chrony with ptp4l and direct PHC access (chrony PHC refclock) and with the complete suite with timemaster and (chrony SHM refclock). A ptp4l or timemaster restart fix the issue. I think the big jump is because the PHC was stopped and completely reset between a down/up sequence. But as the ptp4l is configured in slave only, we should perhaps jump to adjust the time as long as we are in uncalibrated state as ptp4l will do at restart ? (and not run BMCA ?) Chrony is correctly setting the UTC-TAI kernel offset. So the Intel (here) driver is not using it when initializing the PHC otherwise the offset would be minimum after putting the interface UP and I would not have seen the problem. Is it an expected behavior ? Could we do something on the ptp4l side to address this case ? Logs: Aug 20 14:48:51 ns2 ptp4l: [1046806.047] [0:enp7s0f1] port 1: link down Aug 20 14:48:51 ns2 ptp4l: [1046806.047] [0:enp7s0f1] port 1: SLAVE to FAULTY on FAULT_DETECTED (FT_UNSPECIFIED) Aug 20 14:48:51 ns2 ptp4l: [1046806.104] [0:enp7s0f1] selected local clock a0369f.fffe.f1b643 as best master Aug 20 14:48:51 ns2 ptp4l: [1046806.104] [0:enp7s0f1] assuming the grand master role Aug 20 14:48:51 ns2 ptp4l: [1046806.104] [0:enp7s0f1] port 1: master state recommended in slave only mode Aug 20 14:48:51 ns2 ptp4l: [1046806.104] [0:enp7s0f1] port 1: defaultDS.priority1 probably misconfigured Aug 20 14:49:12 ns2 ptp4l: [1046827.119] [0:enp7s0f1] selected local clock a0369f.fffe.f1b643 as best master Aug 20 14:49:12 ns2 ptp4l: [1046827.119] [0:enp7s0f1] assuming the grand master role Aug 20 14:49:12 ns2 ptp4l: [1046827.119] [0:enp7s0f1] port 1: master state recommended in slave only mode Aug 20 14:49:12 ns2 ptp4l: [1046827.119] [0:enp7s0f1] port 1: defaultDS.priority1 probably misconfigured Aug 20 14:49:15 ns2 ptp4l: [1046830.204] [0:enp7s0f1] port 1: link up Aug 20 14:49:15 ns2 kernel: [1046809.055056] igb 0000:07:00.1 enp7s0f1: igb: enp7s0f1 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: RX Aug 20 14:49:15 ns2 kernel: [1046809.055218] IPv6: ADDRCONF(NETDEV_CHANGE): enp7s0f1: link becomes ready Aug 20 14:49:15 ns2 ptp4l: [1046830.248] [0:enp7s0f1] port 1: FAULTY to LISTENING on INIT_COMPLETE Aug 20 14:49:15 ns2 ptp4l: [1046830.614] [0:enp7s0f1] clockcheck: clock jumped backward or running slower than expected! Aug 20 14:49:16 ns2 ptp4l: [1046831.440] [0:enp7s0f1] port 1: new foreign master ec4670.fffe.009fc5-1 Aug 20 14:49:20 ns2 ptp4l: [1046835.441] [0:enp7s0f1] selected best master clock ec4670.fffe.009fc5 Aug 20 14:49:20 ns2 ptp4l: [1046835.441] [0:enp7s0f1] port 1: LISTENING to UNCALIBRATED on RS_SLAVE Aug 20 14:49:20 ns2 ptp4l: [1046835.476] [0:enp7s0f1] port 1: minimum delay request interval 2^1 Aug 20 14:49:25 ns2 ptp4l: [1046840.617] [0:enp7s0f1] port 1: UNCALIBRATED to SLAVE on MASTER_CLOCK_SELECTED Version used: linuxptp 2.0 Emmanuel. ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, Slashdot.org! http://sdm.link/slashdot _______________________________________________ Linuxptp-users mailing list Linuxptp-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linuxptp-users
