I have been testing the 4.17 EC and RC releases on two different setups: LB (Ordinary Clock) and MB (Boundary Clock) for HPE DL110 Gen10 Plus hardware. We've been observing PTP jumps on both SNOs. Initially our primary focus was reliability tests where we install and remove DU 50 times, followed by basic sanity checks after each action. However after investigating this more, we confirmed that the PTP jumps are observed even on a steady phase without any workload.

4.18.0-rc.1

4.17.0-rc.6, 4.17.0-rc.5, 4.17.0-rc.2, 4.17.0-rc.1, 4.17.0-rc.0, 4.17.0-ec.3

    After running the SNO on a steady phase for some time, the issue can be seen randomly. The issue also can be seen after deploying/undeploying the workload on the SNOs randomly.

    1. Deploy SNO.
    2. Monitor the linuxptp-daemon-container for some time and jump can be seen.
    3. Alternatively, deploy/undeploy workload for some time (50 times by default in our tests) and ptp jump can be seen.  

    

Following PTP jump is observed:    
2024-10-08T21:19:55.956403391Z ptp4l[53433.707]: [ptp4l.0.config:4] clockcheck: clock jumped forward or running faster than expected!

    No PTP jump is expected.

    https://redhat-internal.slack.com/archives/C02Q5NTGH6W/p1727879330159119