Uploaded image for project: 'OpenShift Bugs'
  1. OpenShift Bugs
  2. OCPBUGS-19838

PTP HW Clock Class after disabling NMEA sentences

XMLWordPrintable

    • No
    • True
    • Hide

      Need to find a solution to identify different time source  when nmea string is lost

      Show
      Need to find a solution to identify different time source  when nmea string is lost
    • Hide
      One limitation of the current T-GM release is that it has only one NMEA sentence generator coming from the GNSS, without any backup NMEA sentence generator. If NMEA sentences are lost on their way to the E810 controller, the T-GM won't be able to synchronize the devices in the network synchronization chain. If these conditions are met, the PTP operator shall report an error. A fix will be implemented to report 'FREERUN' in the event of a loss of the NMEA string.
      Show
      One limitation of the current T-GM release is that it has only one NMEA sentence generator coming from the GNSS, without any backup NMEA sentence generator. If NMEA sentences are lost on their way to the E810 controller, the T-GM won't be able to synchronize the devices in the network synchronization chain. If these conditions are met, the PTP operator shall report an error. A fix will be implemented to report 'FREERUN' in the event of a loss of the NMEA string.
    • Hide
      2/8: on hold pending design decision for finding clock source when nmea lost.
      11/13: minimal fix for identify NMEA sentence lost PR is waiting to be verified in 4.14.z . This will remain open until we have actual solution to identify the time source on source lost
      10/5: will do minimal fix for identify NMEA sentence lost. The rest will be documented as known issue.
      10/2: This needs deeper discussion . Right now I can fix handling an event when nmea goes away , but can't handle holdover state when NMEA is lost. not sure what clock to source .
      Show
      2/8: on hold pending design decision for finding clock source when nmea lost. 11/13: minimal fix for identify NMEA sentence lost PR is waiting to be verified in 4.14.z . This will remain open until we have actual solution to identify the time source on source lost 10/5: will do minimal fix for identify NMEA sentence lost. The rest will be documented as known issue. 10/2: This needs deeper discussion . Right now I can fix handling an event when nmea goes away , but can't handle holdover state when NMEA is lost. not sure what clock to source .

      Description of problem:

      T-GM clock class announced with ptp4l  

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

      PTP Operator Version: 4.14.0-202309202002

      How reproducible:

      Always

      Steps to Reproduce:

      1. Deploy linuxptp-daemon
      2. Check PTP HW clock class state with PMC is LOCKED (6)
      # pmc -u -f /var/run/ptp4l.0.config  'GET GRANDMASTER_SETTINGS_NP'
      sending: GET GRANDMASTER_SETTINGS_NP
      	507c6f.fffe.1fb3f8-0 seq 0 RESPONSE MANAGEMENT GRANDMASTER_SETTINGS_NP 
      		clockClass              6
      		clockAccuracy           0x21
      		offsetScaledLogVariance 0x4e5d
      		currentUtcOffset        37
      		leap61                  0
      		leap59                  0
      		currentUtcOffsetValid   0
      		ptpTimescale            1
      		timeTraceable           0
      		frequencyTraceable      0
      		timeSource              0xa0 
      3. disable NMEA sentences from arriving to E810 CVL 
       # ubxtool -d NMEA
      
      4. Check Clock Class:
      
      # pmc -u -f /var/run/ptp4l.0.config  'GET GRANDMASTER_SETTINGS_NP'
      sending: GET GRANDMASTER_SETTINGS_NP
      	507c6f.fffe.1fb3f8-0 seq 0 RESPONSE MANAGEMENT GRANDMASTER_SETTINGS_NP 
      		clockClass              6
      		clockAccuracy           0x21
      		offsetScaledLogVariance 0x4e5d
      		currentUtcOffset        37
      		leap61                  0
      		leap59                  0
      		currentUtcOffsetValid   0
      		ptpTimescale            1
      		timeTraceable           0
      		frequencyTraceable      0
      		timeSource              0xa0

      Actual results:

      clockClass is 6 (LOCKED)

      Expected results:

      Without a way to inject NMEA sentences T-GM cant synchronize. We need a way to generate NMEA sentences so that ts2phc can synchronise if we lose NMEA sentences from GNSS. This is a problem that needs some work. Not perfect because ts2phc needs the seconds from somewhere (need time source) like generating synthetic NMEA sentences.

      Additional info:

      Note that when announcing a clockClass 6 or clockClass 7 ALWAYS the "timeTraceable" and freqencyTraceable" should be set to True.

            aputtur@redhat.com Aneesh Puttur
            jnunez@redhat.com Jose Nuñez
            Hen Shay Hassid Hen Shay Hassid
            Pasi Vaananen
            Votes:
            0 Vote for this issue
            Watchers:
            5 Start watching this issue

              Created:
              Updated: