External job scheduling within a distributed processing system having a local job control system

• US 6,694,345 B1
• Filed: 10/05/1999
• Issued: 02/17/2004
• Est. Priority Date: 10/05/1999
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Brelsford, David Paul, Skovira, Joseph Francis
• Primary Examiner(s): Follansbee, John
• Assistant Examiner(s): Opie, George L.
• Application Number:
  US09/412,932
• Time in Patent Office: 1,596 Days
• Field of Search: 709/100, 709/102, 709/104, 707/103.R, 717/127
• US Class Current: 718/100
• CPC Class Codes: G06F 9/485 Task life-cycle, e.g. stopping, restarting, resuming execution G06F9/4881 takes precedence
• CPC Class Codes: G06F 9/5072 Grid computing
• CPC Class Codes: G06F 9/52 Program synchronisation; Mutual exclusion, e.g. by means of semaphores

Abstract

External job scheduling capabilities are provided to a local job control system lacking or having insufficient job scheduling capabilities. This is accomplished by encapsulating running of a user job by the local job control system (LJCS) within running of a marker job at a node management system, which is responsive to an external resource scheduler. The technique includes starting a marker job by the resource scheduler external to the local job control system, wherein the marker job corresponds to a user job to be run by the LJCS; responsive to starting of the marker job, starting the user job by the LJCS; and upon termination of the user job, ending the marker job started by the resource scheduler, wherein starting of the user job by the LJCS is responsive to the resource scheduler'"'"'s starting of the marker job and wherein ending of the marker job is responsive to termination of the user job.

Claims

1. A method for processing jobs within a distributed processing system having a local job control system (LJCS) which receives user jobs to be processed, said method comprising:

starting a marker job by a resource scheduler external to said LJCS, wherein said marker job corresponds to a user job to be run by said LJCS;

responsive to starting of said marker job by said resource scheduler, starting said user job by said LJCS; and

upon termination of said user job, ending said marker job started by said resource scheduler, wherein starting of said user job by said LJCS is responsive to said resource scheduler'"'"'s starting of said marker job and wherein ending of said marker job is responsive to termination of said user job.

2. The method of claim 1, further comprising determining presence of said user job at said LJCS prior to said starting of said marker job by said resource scheduler.

3. The method of claim 2, wherein said determining is initiated by said resource scheduler.

4. The method of claim 1, wherein said resource scheduler starts said marker job utilizing a node management system which is external to said resource scheduler and external to said LJCS, and wherein said marker job exists at said node management system but not said LJCS.

5. The method of claim 4, wherein said starting of said marker job is preceded by said resource scheduler submitting said marker job to a batch job queue in said node management system, and wherein said starting of said user job by said LJCS is responsive to launching of a job command file from said node management system subsequent to starting of said marker job, said job command file requesting said LJCS to start said user job on a node set of said distributed processing system, wherein running of said marker job encompasses running of said user job.

6. The method of claim 4, further comprising determining presence of said user job at said LJCS prior to starting of said marker job by said resource scheduler, wherein said determining comprises comparing user jobs in a user job queue of said LJCS with marker jobs in a batch job queue maintained by said node management system, and identifying a new user job as any job listed in said user job queue but not in said batch job queue.

7. The method of claim 6, further comprising, responsive to identifying of a new user job in said user job queue of said LJCS, obtaining job information on said new user job, including reading a number of nodes required by said new user job, estimating running time required thereby, and obtaining a job name of said new user job, and employing said job information to build a command file to initiate a new marker job at said node management system.

8. The method of claim 7, further comprising monitoring starting of said user job by said LJCS responsive to said job command file, and if said user job fails to start within a first time period, recording a start failure in an error log maintained at said node management system.

9. The method of claim 8, further comprising after starting of said user job by said LJCS, starting a second time period and monitoring for completion of said user job, and wherein said method further comprises canceling said user job from running upon expiration of said second time period without successful completion of said user job, and removing said user job from a running queue of said LJCS.

10. The method of claim 1, further comprising monitoring for failure of completion of said user job by said LJCS, and upon detecting said failure, removing said user job from a running queue of said LJCS to thereby terminate said user job, and thereafter ending said marker job initiated by said resource scheduler.

11. A system for processing jobs within a distributed system having a local job control system (LJCS) which receives user jobs to be processed, said system comprising:

a resource scheduler external to said LJCS, said resource scheduler comprising means for starting a marker job, wherein said marker job corresponds to a user job to be run by said LJCS;

means for starting said user job by said LJCS responsive to starting of said marker job by said resource scheduler; and

means for ending said marker job started by said resource scheduler upon termination of said user job, wherein said means for starting of said user job by said LJCS is responsive to said resource scheduler'"'"'s means for starting said marker job and wherein said means for ending said marker job is responsive to termination of said user job.

12. The system of claim 11, further comprising means for determining presence of said user job at said LJCS prior to said starting of said marker job by said resource scheduler.

13. The system of claim 12, wherein said means for determining is initiated by said resource scheduler.

14. The system of claim 11, wherein said resource scheduler starts said marker job utilizing a node management system which is external to said resource scheduler and external to said LJCS, and wherein said marker job exists at said node management system but not said LJCS.

15. The system of claim 14, wherein said resource scheduler further comprises means for submitting said marker job to a batch job queue in said node management system, and wherein said means for starting of said user job by said LJCS is responsive to launching of a job command file from said node management system subsequent to starting of said marker job, said job command file requesting said LJCS to start said user job on a node set of said distributed processing system, wherein running of said marker job encompasses running of said user job.

16. The system of claim 14, further comprising means for determining presence of said user job at said LJCS prior to starting of said marker job by said resource scheduler, wherein said means for determining comprises means for comparing user jobs in a user job queue of said LJCS with marker jobs in a batch job queue maintained by said node management system, and means for identifying a new user job as any job listed in said user job queue but not in said batch job queue.

17. The system of claim 16, further comprising, responsive to identifying of a new user job in said user job queue of said LJCS, means for obtaining job information on said new user job, including for reading a number of nodes required by said new user job, estimating running time required thereby, and obtaining a job name of said new user job, and means for employing said job information to build a command file to initiate a new marker job at said node management system.

18. The system of claim 17, further comprising means for monitoring starting of said user job by said LJCS responsive to said job command file, and if said user job fails to start within a first time period, for recording a start failure in an error log maintained at said node management system.

19. The system of claim 18, further comprising means for starting a second time period after starting of said user job by said LJCS, and for monitoring for completion of said user job, and wherein said system further comprises means for canceling said user job from running upon expiration of said second time period without successful completion of said user job, and for removing said user job from a running queue of said LJCS.

20. The system of claim 11, further comprising means for monitoring for failure of completion of said user job by said LJCS, and upon detecting said failure, for removing said user job from a running queue of said LJCS to thereby terminate said user job, and thereafter for ending said marker job initiated by said resource scheduler.

21. A system for processing jobs within a distributed processing system having a local job control system (LJCS) which receives user jobs to be processed, said system comprising;

a scheduler module for controlling scheduling of a job for execution within said LJCS, said scheduler module being external to said LJCS and comprising computer code for;

starting a marker job external to said LJCS, wherein said marker job corresponds to a user job to be run by said LJCS;

responsive to starting of said marker job, initiating starting of said user job by said LJCS; and

upon termination of said user job, ending said marker job, wherein starting of said user job by said LJCS is responsive to said scheduler'"'"'s starting of said marker job and wherein ending of said marker job is responsive to termination of said user job.

22. At least one program storage device, readable by a machine, tangibly embodying at least one program of instructions executable by the machine to perform a method for processing jobs within a distributed processing system having a local job control system (LJCS) which receives user jobs to be processed, said method comprising:

starting a marker job by a resource scheduler external to said LJCS, wherein said marker job corresponds to a user job to be run by said LJCS;

responsive to starting of said marker job by said resource scheduler, starting said user job by said LJCS; and

upon termination of said user job, ending said marker job started by said resource scheduler, wherein starting of said user job by said LJCS is responsive to said resource scheduler'"'"'s starting of said marker job and wherein ending of said marker job is responsive to termination of said user job.

23. The at least one program storage device of claim 22, further comprising determining presence of said user job at said LJCS prior to said starting of said marker job by said resource scheduler.

24. The at least one program storage device of claim 23, wherein said determining is initiated by said resource scheduler.

25. The at least one program storage device of claim 22, wherein said resource scheduler starts said marker job utilizing a node management system which is external to said resource scheduler and external to said LJCS, and wherein said marker job exists at said node management system but not said LJCS.

26. The at least one program storage device of claim 25, wherein said starting of said marker job is preceded by said resource scheduler submitting said marker job to a batch job queue in said node management system, and wherein said starting of said user job by said LJCS is responsive to launching of a job command file from said node management system subsequent to starting of said marker job, said job command file requesting said LJCS to start said user job on a node set of said distributed processing system, wherein running of said marker job encompasses running of said user job.

27. The at least one program storage device of claim 25, further comprising determining presence of said user job at said LJCS prior to starting of said marker job by said resource scheduler, wherein said determining comprises comparing user jobs in a user job queue of said LJCS with marker jobs in a batch job queue maintained by said node management system, and identifying a new user job as any job listed in said user job queue but not in said batch job queue.

28. The at least one program storage device of claim 27, further comprising, responsive to identifying of a new user job in said user job queue of said LJCS, obtaining job information on said new user job, including reading a number of nodes required by said new user job, estimating running time required thereby, and obtaining a job name of said new user job, and employing said job information to build a command file to initiate a new marker job at said node management system.

29. The at least one program storage device of claim 28, further comprising monitoring starting of said user job by said LJCS responsive to said job command file, and if said user job fails to start within a first time period, recording a start failure in an error log maintained at said node management system.

30. The at least one program storage device of claim 29, further comprising after starting of said user job by said LJCS, starting a second time period and monitoring for completion of said user job, and wherein said method further comprises canceling said user job from running upon expiration of said second time period without successful completion of said user job, and removing said user job from a running queue of said LJCS.

31. The at least one program storage device of claim 22, further comprising monitoring for failure of completion of said user job by said LJCS, and upon detecting said failure, removing said user job from a running queue of said LJCS to thereby terminate said user job, and thereafter ending said marker job initiated by said resource scheduler.