Uploaded image for project: 'OpenShift Bugs'
  1. OpenShift Bugs
  2. OCPBUGS-49826

T-GM announces "Locked" too fast after holdover

XMLWordPrintable

    • None
    • CNF RAN Sprint 266, CNF RAN Sprint 267
    • 2
    • False
    • Hide

      None

      Show
      None
    • Hide
      There is a known issue when the grandmaster clock (T-GM) transitions to the Locked state too soon. This happens before the Digital Phase-Locked Loop (DPLL) completes its transition to the "Locked-HO-Acquired" state, and after the Global Navigation Satellite Systems (GNSS) time source is restored.
      (link:https://issues.redhat.com/browse/OCPBUGS-49826[*OCPBUGS-49826*])
      Show
      There is a known issue when the grandmaster clock (T-GM) transitions to the Locked state too soon. This happens before the Digital Phase-Locked Loop (DPLL) completes its transition to the "Locked-HO-Acquired" state, and after the Global Navigation Satellite Systems (GNSS) time source is restored. (link: https://issues.redhat.com/browse/OCPBUGS-49826 [* OCPBUGS-49826 *])
    • Known Issue
    • Done
    • 2025-02-20: PR in review upstream. 4.18 release note added

      Description of problem:

          When T-GM is on holdover and the GNSS source is re-acquired, the T-GM-STATUS is announced as S2 before the DPLL is locked:
      
      025-02-04T12:56:32.000210458+00:00 stdout F GM[1738673792]:[ts2phc.0.config] ens4f0 T-GM-STATUS s2
      -- 37 seconds of wrongly announcing class 6 --
      2025-02-04T12:57:09.470315110+00:00 stderr F I0204 12:57:09.470299  322778 dpll.go:362] {3 ice automatic  locked 0x507c6fffff1fb218 pps}

       

      Version-Release number of selected component (if applicable):

          Upstream master

      How reproducible:

      100%    

      Steps to Reproduce:

      Reset GNSS module from time to time to enter the holdover mode:

      sh-5.1# ubxtool -p COLDBOOT -v 3 -P 29.20
      Monitor log for DPLL state and for T-GM state

      Actual results:

          T-GM state moves to S2 after GNSS returns regardless of DPLL state being still in Holdover

      Expected results:

          T-GM state moves to S2 after ValidSourceAvailable + any of the phase offsets (ptp4l, DPLL, ts2phc) <= ptpClockThreshold AND 
      DPLL is in “Locked Holdover Acquired” state

      Additional info:

      Upstream issue: https://github.com/k8snetworkplumbingwg/ptp-operator/issues/9
      Might also fix this customer case: https://access.redhat.com/support/cases/#/case/04040710

              vgrinber@redhat.com Vitaly Grinberg
              vgrinber@redhat.com Vitaly Grinberg
              Daniel Popsuevich Daniel Popsuevich
              Padraig OGrady Padraig OGrady
              Votes:
              0 Vote for this issue
              Watchers:
              7 Start watching this issue

                Created:
                Updated: