Master-Worker Tutorial Condor Week 2006

1
Master-WorkerTutorialCondor Week 2006
2
Agenda

• What is M-W
• When to use M-W
• How to build a simple M-W application
• Q A

3
Why M-W?

• M-W addresses a weakness in Condor
• Short jobs
• Also, for dynamic, parallel workflows

4
A Condor Job
5
An easy solution

• Why not just wrap up smaller jobs into a bigger
  Condor job?
• Partial failures?
• Load balancing?
• Dynamic creation of work?

6
Solution Lightweight TasksMultiplexed on top of
Jobs

• Process Thread Condor Job MW Task
• MWTask dispatch in milliseconds, Condor job can
  take minutes

7
MW is

• C Framework
• To re-use condor worker jobs
• To each run many tasks
• Results in very parallel application

8
MW is not

• MPI
• General parallel programming scheme

9
MW in action
T
Worker
Master exe
T T T T T T T T
T
Worker
T
condor_submit
Worker
Submit machine
10
You Must Write 3 Classes

• Subclasses of
• MWDriver
• MWTask
• MWWorker

Master exe
Worker exe
11
Your_MWTask

• Subclass MWTask
• Data members for inputs
• Data member for results
• Serialization of inputs and results
• Distinct instances on each side

12
The Four Task Methods

• void MyTaskpack_work(void)
• void MyTaskunpack_work(void)
• void MyTaskpack_results(void)
• void MyTaskunpack_results(void)
• Also ctor/dtor!

13
RMComms

• Abstraction for communication
• (and some other stuff)
• RMC-gtpack(int array, int length)
• RMC-gtunpack(int array, int length)

14
MWWorker

• Just one method
• executeTask(MWTask t)
• Also ctor/dtor!

15
MWDriver

• get_userinfo(int argc, char argv)
• RMC-gtadd_executable(char exe, char
  requirements)
• setup_initial_tasks(int num_tasks, MWTask
  init_tasks)
• act_on_completed_task(MWTask t)
• RMC-gtadd_task(MWTask t)
• Also ctor/dtor

16
Putting it all togethernew_skel

• ./new_skel MY_PROJECT
• Use configure help for options
• make

17
Debugging with Independent Mode

• Special RMComm for debugging
• Single process, can run under gdb

18
Running on the Grid

• Just launch the appropriate master
• condor_q to see it in action

19
Advice for Large Runs

• Use personal condor
• Flock, glide-in, schedd-on-side, hobblein
• Use checkpointing!
• Set_worker_increment high

20
User-level Checkpointing

• MWTaskwrite_chkpt_info(FILE )
• MWTaskread_chkpt_info(FILE )
• MWDriverread_master_state(FILE )
• MWDriverwrite_master_state(FILE )

21
Example codes with MW

• Matmul
• Blackbox
• knapsack

22
MW Philosophy

• Reuse either code or concept
• Key idea Late binding

23
Other resources

• http//www.cs.wisc.edu/condor/mw
• Online manual
• MW-users mailing list

24
Thank You!

• Questions?
• MW Home page
• http//www.cs.wisc.edu/condor/mw