Conference humane::scheduler

1152.0. "Q: On demand Job, remote dependencies" by JURA::SCHATT () Mon Aug 26 1996 08:41

    Hi, I some questions about using the Scheduler on the right way:    
    
    I have the following Job dependencies:    

    Job1  ----> Job2  ----> Job3  ----> Job4  ----> Job5    
    Node A      Node B      Node A      Node C      Node A    
    VAX       (VAX remote)  VAX       (VAX remote)  VAX    

    Job2 on Node B must run only if Job1 on Node A has runned successfully,    
    Job3 on Node A must run only if Job2 on Node A has runned successfully,    
    Job4 on Node C must run only if Job3 on Node A has runned successfully,    
    Job5 on Node A must run only if Job4 on Node A has runned successfully,    
    
    and Job1 must be launched only 'on demand' by an admin user from a PC.    

    - First solution: The user can issue a SCHEDULE RUN command via the    
      Scheduler DECWindows interface launched through eXcursion.    

    jobs have been previously created by the following commands:

    schedule> create "@[]job1.com"                     
                     /nointerval/days=NONE/start_time=NEVER/retain=ALL
                     /name="job1"
    schedule> create "@[]job2.com"                     
                     /interval="+0 00:01"/days=ALL/start_time=NOW/retain=ALL
                     /name="job2"
                     /server="NodeB::qbr_mgr"
                     /synchronization="NodeA::job1=qbr_mgr"
    schedule> create "@[]job3.com"                     
                     /interval="+0 00:01"/days=ALL/start_time=NOW/retain=ALL
                     /name="job3"
                     /synchronization="NodeB::job2=qbr_mgr"
    schedule> create "@[]job4.com"                     
                     /interval="+0 00:01"/days=ALL/start_time=NOW/retain=ALL
                     /name="job4"
                     /server="NodeC::qbr_mgr"
                     /synchronization="NodeA::job3=qbr_mgr"
    schedule> create "@[]job5.com"                     
                     /interval="+0 00:01"/days=ALL/start_time=NOW/retain=ALL
                     /name="job5"
                     /synchronization="NodeC::job4=qbr_mgr"

    IS THAT RIGHT in the 'Scheduler philosophy' ?.

    - Second solution: The user launch the Job1 execution by copying 
      Job1.trigger file on NodeA (advantage: don't need to use eXcursion, just
      the file server which must be setup to run the admin application). 
      Job1 is scheduled to run every 5 minutes and checks if job1.trigger is
      present. If no the run finished, otherwise job1.trigger is deleted and
      job1 task is executed.

    job1 has been previously created by the following command:

    schedule> create "@[]job1.com"                     
                     /interval="+0 00:05"/days=ALL/start_time=NOW/retain=ALL
                     /name="job1"

    In this context, how I can schedule the other jobs which depends on Job1 ?.

    - Jobs are scheduled during the application instalation, how I can check
    that they have been already created (schedule show job)...


    Christian.