Thread: [Linuxptp-users] RMS Errors Growing
PTP IEEE 1588 stack for Linux
Brought to you by:
rcochran
Menu
▾
▴
linuxptp-users
|
From: <ja...@as...> - 2023-06-19 17:33:06
|
Hi All, I'm working on a new computer build based on a Supermicro X13SAE motherboard and am seeing something unusual on the ptp4l rms output. When using the I225 interface, ptp works as expected with rms errors reducing to around 500ns. When using the I219 interface, the reported rms error grows with each sample and never converges. Ethtool reports the I219 interface supports ptp filter features so I usually use that interface when setting up computers. Running RHEL 7.9 with the kernel updated to v6.3.8, linuxptp at v4.0.0. Tried another live linux distro with the same results. Any thoughts as to why the errors would continue to grow on the one interface and not the other? Thanks, Jason |
|
From: egg c. <egg...@gm...> - 2023-06-20 06:41:36
|
Hi, Can you provide more info to reproduce the problem? E.g, your configuration, the network connection... Regards <ja...@as...> 于2023年6月20日周二 01:36写道： > > Hi All, > > I'm working on a new computer build based on a Supermicro X13SAE > motherboard and am seeing something unusual on the ptp4l rms output. > When using the I225 interface, ptp works as expected with rms errors > reducing to around 500ns. When using the I219 interface, the reported > rms error grows with each sample and never converges. Ethtool reports > the I219 interface supports ptp filter features so I usually use that > interface when setting up computers. Running RHEL 7.9 with the kernel > updated to v6.3.8, linuxptp at v4.0.0. Tried another live linux distro > with the same results. > > Any thoughts as to why the errors would continue to grow on the one > interface and not the other? > > Thanks, > > Jason > > > _______________________________________________ > Linuxptp-users mailing list > Lin...@li... > https://lists.sourceforge.net/lists/listinfo/linuxptp-users |
|
From: <ja...@as...> - 2023-06-20 15:05:25
|
Hello, The configuration is as listed below. Connect a TimeMachines TM2000B (fw 0.6.3) direct to the ethernet port on the motherboard. Also tested with a Netgear GS108 and still had the error. Boot to linux - I tried RHEL 7.9 with kernel 6.3.8 and RHEL 8.8 with kernel 4.18.0-477.13.1. Both versions had the error. Running ptp4l v4.0 eno1 = I219-LM (rev 11) eno2 = I225-V (rev 03) Ran the following command- sudo ptp4l -i eno2 -s -m ptp4l[840.573]: selected /dev/ptp0 as PTP clock ptp4l[840.574]: port 1 (eno2): INITIALIZING to LISTENING on INIT_COMPLETE ptp4l[840.574]: port 0 (/var/run/ptp4l): INITIALIZING to LISTENING on INIT_COMPLETE ptp4l[840.574]: port 0 (/var/run/ptp4lro): INITIALIZING to LISTENING on INIT_COMPLETE ptp4l[841.762]: port 1 (eno2): new foreign master 3484e4.fffe.ab0d0c-1 ptp4l[845.762]: selected best master clock 3484e4.fffe.ab0d0c ptp4l[845.762]: port 1 (eno2): LISTENING to UNCALIBRATED on RS_SLAVE ptp4l[847.506]: port 1 (eno2): minimum delay request interval 2^-2 ptp4l[848.100]: port 1 (eno2): UNCALIBRATED to SLAVE on MASTER_CLOCK_SELECTED ptp4l[848.350]: rms 10207961715021 max 14436237901989 freq -4614 +/- 3179 delay 6132 +/- 3 ptp4l[849.350]: rms 1239 max 2152 freq -6046 +/- 713 delay 5965 +/- 182 ptp4l[850.351]: rms 462 max 725 freq -4930 +/- 507 delay 5743 +/- 21 ptp4l[851.351]: rms 497 max 771 freq -4599 +/- 458 delay 5726 +/- 27 sudo ptp4l -i eno1 -s -m ptp4l[1061.649]: selected /dev/ptp1 as PTP clock ptp4l[1061.649]: port 1 (eno1): INITIALIZING to LISTENING on INIT_COMPLETE ptp4l[1061.649]: port 0 (/var/run/ptp4l): INITIALIZING to LISTENING on INIT_COMPLETE ptp4l[1061.649]: port 0 (/var/run/ptp4lro): INITIALIZING to LISTENING on INIT_COMPLETE ptp4l[1061.771]: port 1 (eno1): new foreign master 3484e4.fffe.ab0d0c-1 ptp4l[1065.772]: selected best master clock 3484e4.fffe.ab0d0c ptp4l[1065.772]: port 1 (eno1): LISTENING to UNCALIBRATED on RS_SLAVE ptp4l[1067.039]: port 1 (eno1): minimum delay request interval 2^-2 ptp4l[1067.688]: port 1 (eno1): UNCALIBRATED to SLAVE on MASTER_CLOCK_SELECTED ptp4l[1067.938]: rms 21654216012 max 30707819360 freq +17999999 +/- 10392304 delay -33216556 +/- 10188517 ptp4l[1068.939]: clockcheck: clock jumped forward or running faster than expected! ptp4l[1068.939]: rms 630332770 max 822762571 freq +23999999 +/- 0 delay -27609959 +/- 3559790 ptp4l[1068.939]: port 1 (eno1): SLAVE to UNCALIBRATED on SYNCHRONIZATION_FAULT ptp4l[1069.189]: port 1 (eno1): UNCALIBRATED to SLAVE on MASTER_CLOCK_SELECTED ptp4l[1069.939]: clockcheck: clock jumped forward or running faster than expected! ptp4l[1069.939]: rms 1191380561 max 1397186231 freq +23999999 +/- 0 delay -42764839 +/- 0 ptp4l[1069.939]: port 1 (eno1): SLAVE to UNCALIBRATED on SYNCHRONIZATION_FAULT ptp4l[1070.189]: port 1 (eno1): UNCALIBRATED to SLAVE on MASTER_CLOCK_SELECTED ptp4l[1070.939]: clockcheck: clock jumped forward or running faster than expected! ptp4l[1070.939]: rms 1756603386 max 1968528885 freq +23999999 +/- 0 delay -41795176 +/- 7986159 ptp4l[1070.939]: port 1 (eno1): SLAVE to UNCALIBRATED on SYNCHRONIZATION_FAULT Thanks, Jason On 2023-06-20 00:41, egg car wrote: > Hi, > > Can you provide more info to reproduce the problem? E.g, your > configuration, the network connection... > > Regards > > <ja...@as...> 于2023年6月20日周二 01:36写道： >> >> Hi All, >> >> I'm working on a new computer build based on a Supermicro X13SAE >> motherboard and am seeing something unusual on the ptp4l rms output. >> When using the I225 interface, ptp works as expected with rms errors >> reducing to around 500ns. When using the I219 interface, the reported >> rms error grows with each sample and never converges. Ethtool reports >> the I219 interface supports ptp filter features so I usually use that >> interface when setting up computers. Running RHEL 7.9 with the kernel >> updated to v6.3.8, linuxptp at v4.0.0. Tried another live linux >> distro >> with the same results. >> >> Any thoughts as to why the errors would continue to grow on the one >> interface and not the other? >> >> Thanks, >> >> Jason >> >> >> _______________________________________________ >> Linuxptp-users mailing list >> Lin...@li... >> https://lists.sourceforge.net/lists/listinfo/linuxptp-users |
|
From: Miroslav L. <mli...@re...> - 2023-06-20 15:28:11
|
On Tue, Jun 20, 2023 at 09:05:10AM -0600, ja...@as... wrote: > ptp4l[1068.939]: clockcheck: clock jumped forward or running faster than > expected! > ptp4l[1068.939]: rms 630332770 max 822762571 freq +23999999 +/- 0 delay > -27609959 +/- 3559790 This shows the frequency being stuck at the maximum. The server, HW or driver is broken, or something else is adjusting the clock. -- Miroslav Lichvar |
|
From: <ja...@as...> - 2023-06-21 21:44:21
|
Miroslav, Thanks for the suggestions- Found chrony running so I stopped and disabled chronyd. Still see the problem. I have another older computer running RHEL 7.9 with an Asus mb and a I219-LM chip on the same local net and it works properly. I tried the RHEL 8.8 OS that was not working with the Supermicro mb on the Asus mb with an I219-LM chip and it worked. I tried the RHEL 8.8 OS on a third computer that has the new Supermicro mb with the same hardware as the original and I still saw the error. Based on the testing steps above, the three main differences are motherboard, cpu and the I219-LM rev. The working 219 chip reports rev 10 where the one that reports an error is rev 11. It seems like a driver would be the most likely problem. A quick look at the Intel website indicates that the e1000e drivers have changed to a kernel-only support model. Jason On 2023-06-20 09:24, Miroslav Lichvar wrote: > On Tue, Jun 20, 2023 at 09:05:10AM -0600, ja...@as... wrote: >> ptp4l[1068.939]: clockcheck: clock jumped forward or running faster >> than >> expected! >> ptp4l[1068.939]: rms 630332770 max 822762571 freq +23999999 +/- 0 >> delay >> -27609959 +/- 3559790 > > This shows the frequency being stuck at the maximum. The server, HW or > driver > is broken, or something else is adjusting the clock. |
×
Want the latest updates on software, tech news, and AI?
Get latest updates about software, tech news, and AI from SourceForge directly in your inbox once a month.
Thanks for helping keep SourceForge clean.
X