[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference humane::scheduler

Title:SCHEDULER
Notice:Welcome to the Scheduler Conference on node HUMANEril
Moderator:RUMOR::FALEK
Created:Sat Mar 20 1993
Last Modified:Tue Jun 03 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:1240
Total number of notes:5017

1145.0. "NT agent jobs unable to start until SCHED CHECK/ALL finishes." by CSC32::BSS::SAUL () Thu Aug 01 1996 19:56

Hi,

I have a customer who says that jobs on their NT agent will not start
until a SCHED CHECK/ALL has finished on the server.  Is this normal 
behavior?  Would the NSCHED process be getting hung up due to the
check being done.  Thanks for any info...

Ted
T.RTitleUserPersonal
Name
DateLines
1145.1explanationRUMOR::FALEKex-TU58 KingFri Aug 02 1996 15:0320
    I think what the code in SCHED CHECK/ALL does is to cause the scheduler
    to read every job record in the database (in job # order), and "process" it,
    doing any neccessary actions (this includes re-checking schedule and all
    dependencies for each job, etc.) If a job is encountered during the
    check that "should" run, then it will be run. The SCHED CHECK operation
    involves a lot of processing and if you have a lot of jobs, it can take
    a while.  If, while the check is running, run is requested for a job
    via the user interface, and the job hasn't been processed yet by the
    "check" command, then the job will be run as soon as the scheduler gets
    to that job number during processing. If the job wasn't runnable
    at the moment it was processed by SCHED CHECK, and then becomes
    runnable while SCHED CHECK is still processing, (for example if run was
    requested via the user interface, dependencies complete, or the
    scheduled time arrives) than it won't run untill SCHED CHECK completes
    and normal processing resumes.
    
    In theory there should be no need to run $ SCHED CHECK command at
    all - in fact it wasn't documented in early versions of the scheduler.
    (Because of the large amount of overhead and job processing delay)
    Are they doing this for any particular reason ? 
1145.2/allRUMOR::FALEKex-TU58 KingFri Aug 02 1996 15:053
    The /ALL in in $ SCHED CHECK/ALL means that all scheduler in the
    cluster do the $SCHED CHECK command against the job database (at the same
    time) - which could slow I/O and make the comamnd take even longer