Constraints/conditions: implement a functionality to run jobs against some conditions

User Requirements

• I would like to provide to customers an alternative to JobSynchronize and Events Management with something similar but easier to understand and to use. The principle is to use scheduler_history instead of scheduler_event table and evaluate the records with a SQL query in a pre-processing function.
• The use could be the following:
  • add a parameter CONSTRAINTS, containing a formula with all information for an SQL query.
  • add a Java class as a pre-processing function, the class transforms the rules in SQL and executes the request until it gets a result.
• The formula can have the following format:

         STATUS(SPOOLER/PATH/JOB) or EXIT(SPOOLER/PATH/JOB)=exit_code 
                   With STATUS = DONE, ERROR, SUCCESS, RUNNING 
  

• The JOB could be /PATH/JOB if it is on the same SPOOLER or SPOOLER/PATH/JOB if it is another one.
• The rule can use conditions: | & and parenthesis
• For different frequency, we can precise the time window: SUCCESS(JOB#10:00:00) if the success happened in the last 10 hours.
  • Examples:

    CONSTRAINTS : SUCCESS(SCHEDULER/test/job1) | FAILURE(/job2)
    	wait for the successful execution of /test/job1 on the spooler SCHEDULER or the failure of job2 on the same scheduler
    
    CONSTRAINTS : EXIT(SCHEDULER/test/job1)=2,4-8,12 & DONE(/job3)
    	wait for the job1 ending with one of the following exit codes : 2,4,5,6,7,8,13 and /job3 not running.
    
    CONTRAINTS:  FAILURE(/job3#01:00:00) 
    	wait for a failure happened in the last hour.
    

  • This kind of syntax will greatly simplify the migrations because we could solve problems with multiple dependencies, dependencies between nodes and "and/or" conditions.