[HTCondor-users] zero cputime reported by startd 10.5.0

[Stefano Dal Pra <stefano.dalpra@xxxxxxxxxxxx>]

    Hi, 
    we are evaluating upgrade of our Execution Points from 9.0.17 to
    Current Feature Release (10.5.0).
    (while keeping 9.0.17 at the SCHEDDs _especially those with
    condor-ce_ and at Central Manager).
    
    So I upgraded a WN to 10.5.0, i've seen it working as expected and
    let it work during the weeend.
    Today i went looking in the accounting for jobs done in that machine
    and noticed that zero or very little cputime 
    is accounted.
    This looks strange, as running top from the machine shows machine
    cores "working hard".
    
    To verify a little further i executed a known test program who just
    crunch integer numbers for some 5 minutes.
    It run as expected, reports the correct final result (thus the
    program HAS run) but still the accounting reports zero cputime
    
    In the Job history file these are the relevant reported values:
    
    [root@ce06-htc
          2023-6]# cat history.9602771.0 | egrep -i
          '^RemoteWallClockTime|cpu'
        
        CPUsUsage =
          0.9997753100861237
        
        CpusProvisioned =
          1
        
        CumulativeRemoteSysCpu = 0.0
        
        CumulativeRemoteUserCpu = 0.0
        
        MATCH_EXP_numcpus = "32"
        
        MATCH_TotalSlotCpus = 32
        
        MachineAttrCpus0 = 1
        
        RemoteSysCpu = 0.0
        
        RemoteUserCpu = 0.0
        
        RemoteWallClockTime = 503.0
        
        RequestCpus = 1
        
        
        In general
      *SysCpu
        and *UserCpu are at 0.0, rarely few
    seconds are reported ( <= 5)
    
        [root@wn-200-10-11-02-a ~]# condor_version Â
      $CondorVersion: 10.5.0 2023-06-05 BuildID: 650732 PackageID:
      10.5.0-1 $
      
      $CondorPlatform: x86_64_CentOS7 $
      
    Is that a communication problem with the 9.0.17 schedd ?
    
    Cheers,
    Stefano