cDSCHEDULE Automated Job Submission System Guide and Reference �ManualD

SCHEDULE
Automated Job Submission System
Guide and Reference Manual


 o T Y  
PreviousContentsIndex

9

2.3.5.4 Marking a calendar



CEach day of a calendar can contain up to two letters as marks. The 4marked calendar rescheduling method specifies that aHjob is to run on all days marked by a certain letter. Either of the two letters can be this mark.

EFor example to mark January 10 with the letter "A" use the following command:

 

" 
 B$ SCHEDULE MODIFY/CALENDAR [DEFAULT]STANDARD/DATE=10-JAN/MARK="A"    



GFor example to mark January 1 as a day that no runs are allowed on use the following command:

 

" 
 >$ SCHEDULE MODIFY/CALENDAR [DEFAULT]STANDARD/DATE=1-JAN/NORUN  



=

2.3.5.5 Setting up holidays

ETo set up system wide holidays use the [DEFAULT]HOLIDAY calendar. To Gset up a local holiday calendar just create one by that name in either Fyour top-level or sub-level directories. Holiday calendars are marked up just like any other calendar.

?For example to mark January 1 as a system wide holiday use the following command:

 

" 
 =$ SCHEDULE MODIFY/CALENDAR [DEFAULT]HOLIDAY/DATE=1-JAN/NORUN    



*

2.3.6 Reports

EAn important component of the SCHEDULE is a general purpose Creport writer. The SCHEDULE system interfaces this report ?writer to the SCHEDULE database. The report writer is Gcontrolled by a group of report definitions. A group of 80 (32 history data, 48 control data)Freport definitions are provide with the system. All these definitions =can be customized as needed or new ones defined. 9

BThe format and content of these reports are documented in a later section of this manual.

FThere are two types of data that can be used by the report writer. 1) HControl data which contains all the static information about what a job Bis and how to schedule it. 2) History data which contains all the 6historical information about scheduling state changes.8

2.3.7 Scheduling queues



DAll job scheduling activity is controlled by a series of scheduling Hqueues. Each queue represents a state. For example one queue is used to Dhold jobs waiting for disk space. The various scheduling queues are Gidentified by a single letter. In general jobs start at the top (queue #A) and progress downward (queue S).

)The various queue names are listed below.

    I
  1. COMPUTE NEXT START Any job that is to be examined H to determine when it will start again is entered here. The next start + time is computed and the job proceeds on.I
  2. WAIT FOR NEXT START TIME Any job that has a start J time set is entered into this queue and stays here until the start time occurs.E
  3. INITIATE WITH NO SET ID Any job that has been I initiated either manually or automatically that does NOT have a SET ID B assigned is entered here. The job is assigned a number and then proceeds.A
  4. WORKGROUP MANAGER Workgroup processing is J performed in this state. Node analysis and selection is performed. Only , affects jobs that are part of a workgroup.D
  5. WAIT FOR LOCAL PREREQUISITE Any job that has F prerequisites on the local system (or cluster) is entered into this H queue. Once all the prerequisite jobs have successfully completed the  job proceeds.F
  6. REQUEST REMOTE PREREQUISITE Any job that has a E prerequisite specified that is on a remote node is entered here. A F request is transmitted to the remote node and then the job proceeds.C
  7. WAIT FOR REMOTE PREREQUISITE Once a job has H requested a completion event from a remote node it then waits in this J queue until the event occurs. Once all responses have been received the  job proceeds on.H
  8. WAIT FOR A REQUIRED FILE Any job that requires a H file to be present is entered into this queue. Once the file has been  created the job proceeds.E
  9. WAIT FOR DISK SPACE REQUIREMENTS Any job that J requires a certain amount of disk space to be present before proceeding 7 will stay in this queue until the space is available.B
  10. READY TO SUBMIT Once a job has met all the I resource and other requirements it is entered into this queue. The job D is then entered into a batch execution queue and the job proceeds.E
  11. SUBMITTED FOR EXECUTION Once the job has been ; submitted for execution an entry is made into this queue.H
  12. EXECUTION STARTED Once the job starts to execute # an entry is made into this queue.I
  13. JOB COMPLETED When the job completes execution it  is entered into this queue.C
  14. INITIATE DESCENDANTS JOBS On completion the A initiate list of each job is examined. Any jobs that are to be ) activated are entered into the B queue.I
  15. SEND INITIATE REQUEST TO A REMOTE NODE Any remote F initiate job request is placed into this queue. The request is then & transmitted over to the remote node.I
  16. HOLDING PREVIOUS COMPLETIONS When a job completes F it is entered into this queue. Entries are held in this queue to be I used to satisfy future prerequisite requirements. After a short period 3 typically 1 day the entries are removed.C
  17. EVENT MESSAGES Foreground processing signal ? messages are entered into this queue before dispatching them.I
  18. JOB COMPLETION MESSAGES Foreground job completion I messages containing complete history information are entered into this  queue before dispatching.B
  19. REMOTE COMPLETION MESSAGES Remote messages I requesting information when a job has completed are kept in this queue F until the job completes. At that time a message is sent back to the  requesting node.

    A
  1. EVALUATING VARIABLES The job is currently J evaluating variables and determining the course that the job group will G go depending on the resource list, initiates, prerequisites, and end  action list.




/   
N
Note

 P

9 The report definitions are contained 9 in the file SCHEDULE_LIBRARY:SCHRPTCLB.CLB.







 o V T Y  
PreviousNextContentsIndex