1. 1 http://www.cs.wisc.edu/condor Using Condor An Introduction Condor Week 2009

2. 2 http://www.cs.wisc.edu/condor The Condor Project (Established ‘85) Distributed High Throughput Computing research performed by a team of ~35 faculty, full time staff and students who:  face software engineering challenges in a distributed UNIX/Linux/NT environment  are involved in national and international grid collaborations,  actively interact with academic and commercial entities and users,  maintain and support large distributed production environments,  and educate and train students.

3. 3 http://www.cs.wisc.edu/condor The Condor Project

4. 4 http://www.cs.wisc.edu/condor Some free software produced by the Condor Project › Condor System › VDT › Metronome › ClassAd Library › DAGMan › Hawkeye › GCB & CCB › MW › NeST / LotMan › And others… all as open source

5. 5 http://www.cs.wisc.edu/condor Full featured system › Flexible scheduling policy engine via ClassAds  Preemption, suspension, requirements, preferences, groups, quotas, settable fair-share, system hold… › Facilities to manage BOTH dedicated CPUs (clusters) and non-dedicated resources (desktops) › Transparent Checkpoint/Migration for many types of serial jobs › No shared file-system required › Federate clusters w/ a wide array of Grid Middleware

6. 6 http://www.cs.wisc.edu/condor Full featured system › Workflow management (inter-dependencies) › Support for many job types – serial, parallel, etc. › Fault-tolerant: can survive crashes, network outages, no single point of failure. › Development APIs: via SOAP / web services, DRMAA (C), Perl package, GAHP, flexible command- line tools, MW › Platforms: Linux i386/IA64, Windows 2k/XP/Vista, MacOS, FreeBSD, Solaris, IRIX, HP-UX, Compaq Tru64, … lots.  IRIX and Tru64 are no longer supported by current releases of Condor

7. 7 http://www.cs.wisc.edu/condor The Problem. Our esteemed scientist needs to run a “small” simulation.

8. 8 http://www.cs.wisc.edu/condor The Simulation Run a simulation of the evolution of the cosmos with various properties

9. 9 http://www.cs.wisc.edu/condor The Simulation Details Varying values for the value of:  G (the gravitational constant): 100 values  R μν (the cosmological constant): 100 values  c (the speed of light): 100 values  100 × 100 × 100 = 1,000,000 jobs

10. 10 http://www.cs.wisc.edu/condor Running the Simulation Each simulation:  Requires up to 4GB of RAM  Requires 20MB of input  Requires 2 – 500 hours of computing time  Produces up to 10GB of output Estimated total:  15,000,000 hours!

11. 11 http://www.cs.wisc.edu/condor NSF won’t fund the Blue Gene that I requested.

12. 12 http://www.cs.wisc.edu/condor While sharing a beverage with some colleagues, Carl asks “Have you tried Condor? It’s free, already installed for you, and you can use our pool.”

13. 13 http://www.cs.wisc.edu/condor Are my Jobs Safe? › No worries… Condor will take care of your jobs for you  Jobs are queued in a safe way More details later  Condor will make sure that your jobs run, return output, etc. You can specify what defines “OK”

14. 14 http://www.cs.wisc.edu/condor Condor will... › Keep an eye on your jobs and will keep you posted on their progress › Implement your policy on the execution order of the jobs › Keep a log of your job activities › Add fault tolerance to your jobs › Implement your policy on when the jobs can run on your workstation

15. 15 http://www.cs.wisc.edu/condor Condor Doesn’t Play Dice with My Jobs!

16. 16 http://www.cs.wisc.edu/condor Definitions › Job  The Condor representation of your work › Machine  The Condor representation of computers and that can perform the work › Match Making  Matching a job with a machine “Resource” › Central Manager  Central repository for the whole pool  Performs job / machine matching, etc.

17. 17 http://www.cs.wisc.edu/condor Job Jobs state their requirements and preferences: I require a Linux/x86 platform I prefer the machine with the most memory I prefer a machine in the chemistry department

18. 18 http://www.cs.wisc.edu/condor Machine Machines state their requirements and preferences: Require that jobs run only when there is no keyboard activity I prefer to run Albert’s jobs I am a machine in the physics department Never run jobs belonging to Dr. Heisenberg

19. 19 http://www.cs.wisc.edu/condor The Magic of Matchmaking › Jobs and machines state their requirements and preferences › Condor matches jobs with machines  Based on requirements  And preferences “Rank”

20. 20 http://www.cs.wisc.edu/condor Getting Started: Submitting Jobs to Condor › Overview:  Choose a “Universe” for your job  Make your job “batch-ready”  Create a submit description file  Run condor_submit to put your job in the queue

21. 21 http://www.cs.wisc.edu/condor 1. Choose the “Universe”  Todd’s Universe › Controls how Condor handles jobs › Choices include:  Vanilla  Standard  Grid  Java  Parallel  VM

22. 22 http://www.cs.wisc.edu/condor Using the Vanilla Universe The Vanilla Universe: – Allows running almost any “serial” job – Provides automatic file transfer, etc. – Like vanilla ice cream Can be used in just about any situation

23. 23 http://www.cs.wisc.edu/condor 2. Make your job batch- ready Must be able to run in the background No interactive input No GUI/window clicks – Who needs ‘em anyway? No music ;^)

24. 24 http://www.cs.wisc.edu/condor Make your job batch-ready (details)…  Job can still use STDIN, STDOUT, and STDERR (the keyboard and the screen), but files are used for these instead of the actual devices  Similar to UNIX shell: $./myprogram output.txt

25. 25 http://www.cs.wisc.edu/condor 3. Create a Submit Description File › A plain ASCII text file › Condor does not care about file extensions › Tells Condor about your job:  Which executable, universe, input, output and error files to use, command-line arguments, environment variables, any special requirements or preferences (more on this later) › Can describe many jobs at once (a “cluster”), each with different input, arguments, output, etc.

26. 26 http://www.cs.wisc.edu/condor Simple Submit Description File # Simple condor_submit input file # (Lines beginning with # are comments) # NOTE: the words on the left side are not # case sensitive, but filenames are! Universe = vanilla Executable = cosmos Output = cosmos.out Queue

27. 27 http://www.cs.wisc.edu/condor 4. Run condor_submit › You give condor_submit the name of the submit file you have created:  condor_submit cosmos.submit › condor_submit:  Parses the submit file, checks for errors  Creates a “ClassAd” that describes your job(s)  Puts job(s) in the Job Queue

28. 28 http://www.cs.wisc.edu/condor ClassAd ? › Condor’s internal data representation  Similar to classified ads (as the name implies)  Represent an object & its attributes Usually many attributes  Can also describe what an object matches with

29. 29 http://www.cs.wisc.edu/condor ClassAd Details › ClassAds can contain a lot of details  The job’s executable is “ cosmos”  The machine’s load average is 5.6 › ClassAds can specify requirements  I require a machine with Linux › ClassAds can specify preferences  This machine prefers to run jobs from the physics group

30. 30 http://www.cs.wisc.edu/condor ClassAd Details (continued) › ClassAds are:  semi-structured  user-extensible  schema-free  Attribute = Expression

31. 31 http://www.cs.wisc.edu/condor Example: MyType = "Job" TargetType = "Machine" ClusterId = 1377 Owner = “einstein" Cmd = “cosmos" Requirements = ( (Arch == "INTEL") && (OpSys == "LINUX") && (Disk >= DiskUsage) && ( (Memory * 1024)>=ImageSize ) ) … ClassAd Example String Number Boolean

32. 32 http://www.cs.wisc.edu/condor The Dog ClassAd Type = “Dog” Color = “Brown” Price = 12 ClassAd for the “Job”... Requirements = (type == “Dog”) && (color == “Brown”) && (price <= 15)...

33. 33 http://www.cs.wisc.edu/condor The Job Queue › condor_submit sends your job’s ClassAd(s) to the schedd › The schedd (more details later):  Manages the local job queue  Stores the job in the job queue Atomic operation, two-phase commit “Like money in the bank” › View the queue with condor_q

34. 34 http://www.cs.wisc.edu/condor Example condor_submit and condor_q % condor_submit cosmos.submit Submitting job(s). 1 job(s) submitted to cluster 1. % condor_q -- Submitter: perdita.cs.wisc.edu : : ID OWNER SUBMITTED RUN_TIME ST PRI SIZE CMD 1.0 einstein 4/15 06:52 0+00:00:00 I 0 0.0 cosmos 1 jobs; 1 idle, 0 running, 0 held %

35. 35 http://www.cs.wisc.edu/condor Email about your job Condor sends email about job events to the submitting user Specify “notification” in your submit file to control which events: Notification= complete Notification= never Notification= error Notification= always Default

36. 36 http://www.cs.wisc.edu/condor Input, output & error files › Controlled by submit file settings › You can define the job’s standard input, standard output and standard error:  Read job’s standard input from “input_file”: Input= input_file Shell equivalent: program <input_file  Write job’s standard ouput to “output_file”: Output= output_file Shell equivalent: program >output_file  Write job’s standard error to “error_file”: Error= error_file Shell equivalent: program 2>error_file

37. 37 http://www.cs.wisc.edu/condor Feedback on your job › Create a log of job events › Add to submit description file: log = cosmos.log › Becomes the Life Story of a Job  Shows all events in the life of a job  Always have a log file

38. 38 http://www.cs.wisc.edu/condor Sample Condor User Log 000 (0001.000.000) 05/25 19:10:03 Job submitted from host:... 001 (0001.000.000) 05/25 19:12:17 Job executing on host:... 005 (0001.000.000) 05/25 19:13:06 Job terminated. (1) Normal termination (return value 0)...

39. 39 http://www.cs.wisc.edu/condor Example Submit Description File With Logging # Example condor_submit input file # (Lines beginning with # are comments) # NOTE: the words on the left side are not # case sensitive, but filenames are! Universe = vanilla Executable = /home/einstein/condor/cosmos.condor Log = cosmos.log ·Job log (from Condor) Input = cosmos.in ·Program’s standard input Output = cosmos.out ·Program’s standard output Error = cosmos.err ·Program’s standard error Arguments = -c=299792458 –G=6.67300e-112 ·Command line arguments InitialDir = /home/einstein/cosmos/run Queue

40. 40 http://www.cs.wisc.edu/condor “Clusters” and “Processes” › If your submit file describes multiple jobs, we call this a “cluster” › Each cluster has a unique “cluster number” › Each job in a cluster is called a “process”  Process numbers always start at zero › A Condor “Job ID” is the cluster number, a period, and the process number (i.e. 2.1)  A cluster can have a single process Job ID = 20.0 ·Cluster 20, process 0  Or, a cluster can have more than one process Job ID: 21.0, 21.1, 21.2·Cluster 21, process 0, 1, 2

41. 41 http://www.cs.wisc.edu/condor Submit File for a Cluster # Example submit file for a cluster of 2 jobs # with separate input, output, error and log files Universe = vanilla Executable = cosmos Arguments = -f cosmos_0.dat log = cosmos_0.log Input = cosmos_0.in Output = cosmos_0.out Error = cosmos_0.err Queue ·Job 2.0 (cluster 2, process 0) Arguments = -f cosmos_1.dat log = cosmos_1.log Input = cosmos_1.in Output = cosmos_1.out Error = cosmos_1.err Queue ·Job 2.1 (cluster 2, process 1)

42. 42 http://www.cs.wisc.edu/condor % condor_submit cosmos.submit-file Submitting job(s). 2 job(s) submitted to cluster 2. % condor_q -- Submitter: perdita.cs.wisc.edu : : ID OWNER SUBMITTED RUN_TIME ST PRI SIZE CMD 1.0 einstein 4/15 06:52 0+00:02:11 R 0 0.0 cosmos -c=299792458 2.0 einstein 4/15 06:56 0+00:00:00 I 0 0.0 cosmos –f cosmos_0.dat 2.1 einstein 4/15 06:56 0+00:00:00 I 0 0.0 cosmos –f cosmos_1.dat 3 jobs; 2 idle, 1 running, 0 held % Submitting The Job

43. 43 http://www.cs.wisc.edu/condor Back to our 1,000,000 jobs… › We could put all input, output, error & log files in the one directory  One of each type for each job  That’d be 4,000,000 files (4 files × 1,000,000 jobs)  Difficult (at best) to sort through › Better: Create a subdirectory for each run

44. 44 http://www.cs.wisc.edu/condor Organize your files and directories for big runs › Create subdirectories for each “run”  run_0, run_1, … run_999999 › Create input files in each of these  run_0/(cosmos.in,cosmos.dat)  run_1/(cosmos.in,cosmos.dat)  …  run_999999/(cosmos.in,cosmos.dat) › The output, error & log files for each job will be created by Condor from your job’s output

45. 45 http://www.cs.wisc.edu/condor Einstein’s simulation directory cosmos.exe cosmos.sub run_999999 cosmos.in cosmos.dat cosmos.out cosmos.err cosmos.log run_0 User creates these files Condor creates these files for you cosmos.in cosmos.dat cosmos.out cosmos.err cosmos.log

46. 46 http://www.cs.wisc.edu/condor Submit Description File for 1,000,000 Jobs # Cluster of 1,000,000 jobs with different directories Universe = vanilla Executable = cosmos Log = cosmos.log Output = cosmos.out Input = cosmos.in Arguments = –f cosmos.dat... InitialDir = run_0 ·Log, input, output & error files -> run_0 Queue ·Job 3.0 (Cluster 3, Process 0) InitialDir = run_1 ·Log, input, output & error files -> run_1 Queue ·Job 3.1 (Cluster 3, Process 1) ·Do this 3,999,998 more times…………

47. 47 http://www.cs.wisc.edu/condor Submit File for a Big Cluster of Jobs › We just submitted 1 cluster with 1,000,000 processes › All the input/output files will be in different directories › The submit file is pretty unwieldy (over 5,000,000 lines!) › Isn’t there a better way?

48. 48 http://www.cs.wisc.edu/condor Submit File for a Big Cluster of Jobs (the better way) › We can queue all 1,000,000 in 1 “Queue” command  Queue 1000000 › Condor provides $(Process) and $(Cluster)  $(Process) will be expanded to the process number for each job in the cluster 0, 1, … 999999  $(Cluster) will be expanded to the cluster number Will be 4 for all jobs in this cluster

49. 49 http://www.cs.wisc.edu/condor $(Process) › The initial directory for each job can be specified using $(Process)  InitialDir = run_$(Process)  Condor will expand these to: “ run_0 ”, “ run_1 ”, … “ run_999999 ” directories › Similarly, arguments can be variable  Arguments = -n $(Process)  Condor will expand these to: “-n 0”, “-n 1”, … “-n 999999”

50. 50 http://www.cs.wisc.edu/condor Better Submit File for 1,000,000 Jobs # Example condor_submit input file that defines # a cluster of 100000 jobs with different directories Universe = vanilla Executable = cosmos Log = cosmos.log Input = cosmos.in Output = cosmos.out Error = cosmos.err Arguments = –f cosmos.dat ·All share arguments InitialDir = run_$(Process) ·run_0 … run_999999 Queue 1000000 ·Jobs 4.0 … 4.999999

51. 51 http://www.cs.wisc.edu/condor Now, we submit it… $ condor_submit cosmos.submit Submitting job(s)............................................................................................................................................................................................................................................................... Logging submit event(s)............................................................................................................................................................................................................................................................... 1000000 job(s) submitted to cluster 4.

52. 52 http://www.cs.wisc.edu/condor And, Check the Queue $ condor_q -- Submitter: x.cs.wisc.edu : : x.cs.wisc.edu ID OWNER SUBMITTED RUN_TIME ST PRI SIZE CMD 4.0 einstein 4/20 12:08 0+00:00:05 R 0 9.8 cosmos –f cosmos.dat 4.1 einstein 4/20 12:08 0+00:00:03 I 0 9.8 cosmos –f cosmos.dat 4.2 einstein 4/20 12:08 0+00:00:01 I 0 9.8 cosmos –f cosmos.dat 4.3 einstein 4/20 12:08 0+00:00:00 I 0 9.8 cosmos –f cosmos.dat... 4.999998 einstein 4/20 12:08 0+00:00:00 I 0 9.8 cosmos –f cosmos.dat 4.999999 einstein 4/20 12:08 0+00:00:00 I 0 9.8 cosmos –f cosmos.dat 999999 jobs; 999998 idle, 1 running, 0 held

53. 53 http://www.cs.wisc.edu/condor 1,000,000 Condor jobs Job Queue Condor Pool Condor Schedd Einstein’s workstation Physics Department Condor Pool Einstein and the other scientists can use the compute resources of each other’s machines

54. 54 http://www.cs.wisc.edu/condor My Biggest Blunder Ever › Albert removes R μν (Cosmological Constant) from his equations, and needs to remove his running jobs › We’ll just ignore that modern cosmologists may have re-introduced R μν (so called “dark energy”)

55. 55 http://www.cs.wisc.edu/condor Removing Jobs › If you want to remove a job from the Condor queue, you use condor_rm › You can only remove jobs that you own › Privileged user can remove any jobs  “root” on UNIX  “administrator” on Windows

56. 56 http://www.cs.wisc.edu/condor Removing jobs (continued) › Remove an entire cluster:  condor_rm 4 ·Removes the whole cluster › Remove a specific job from a cluster:  condor_rm 4.0 ·Removes a single job › Or, remove all of your jobs with “-a”  condor_rm -a ·Removes all jobs / clusters

57. 57 http://www.cs.wisc.edu/condor condor_status % condor_status Name OpSys Arch State Activ LoadAv Mem ActvtyTime antipholus.cs LINUX INTEL Unclaimed Idle 0.020 511 0+02:28:42 coral.cs.wisc LINUX INTEL Claimed Busy 0.990 511 0+01:27:21 doc.cs.wisc.e LINUX INTEL Unclaimed Idle 0.260 511 0+00:20:04 dsonokwa.cs.w LINUX INTEL Claimed Busy 0.810 511 0+00:01:45 ferdinand.cs. LINUX INTEL Claimed Suspe 1.130 511 0+00:00:55 vm1@pinguino. LINUX INTEL Unclaimed Idle 0.000 255 0+01:03:28 vm2@pinguino. LINUX INTEL Unclaimed Idle 0.190 255 0+01:03:29

58. 58 http://www.cs.wisc.edu/condor How can my jobs access their data files?

59. 59 http://www.cs.wisc.edu/condor Access to Data in Condor › Use shared filesystem if available › No shared filesystem?  Condor can transfer files Can automatically send back changed files Atomic transfer of multiple files Can be encrypted over the wire  Remote I/O Socket (parrot)  Standard Universe can use remote system calls (more on this later)

60. 60 http://www.cs.wisc.edu/condor Condor File Transfer › Transfer_Input_Files  List of files that you want Condor to transfer to the execute machine › Transfer_Output_Files  List of files that you want Condor to transfer from the execute machine  If not specified, Condor will transfer back all “new” files in the execute directory › ShouldTransferFiles  YES: Always transfer files to execution site  NO: Always rely on a shared filesystem  IF_NEEDED: Condor will automatically transfer the files if the submit and execute machine are not in the same FileSystemDomain ( Use shared file system if available)

61. 61 http://www.cs.wisc.edu/condor Condor File Transfer Example # Example submit file using file transfer Universe = vanilla Executable = cosmos Log = cosmos.log ShouldTransferFiles = IF_NEEDED Transfer_input_files = dataset.$(Process), cosmos.dat Transfer_output_files = TheAnswer.dat Queue 1000000

62. 62 http://www.cs.wisc.edu/condor We Always Want More Condor is managing and running our jobs, but:  Our CPU requirements are greater than our resources  Jobs are preempted more often than we like

63. 63 http://www.cs.wisc.edu/condor Happy Day! The Physics Department purchased a dedicated cluster! The administrator installs Condor on all these new dedicated cluster nodes

64. 64 http://www.cs.wisc.edu/condor Adding nodes › Einstein installs Condor on the desktop machines, and configures them with his machine as the central manager  The central manager: Central repository for the whole pool Performs job / machine matching, etc. › These are “non-dedicated” nodes, meaning that they can't always run Condor jobs

65. 65 http://www.cs.wisc.edu/condor 1,000,000 Condor jobs Job Queue Workstations Condor Schedd Einstein’s workstation Physics Department Condor Pool Central Manager Dedicated Cluster With the additional resources, Albert and his colleagues can get their jobs completed faster.

66. 66 http://www.cs.wisc.edu/condor Some Good Questions… What are all of these Condor Daemons running on my machine? What do they do?

67. 67 http://www.cs.wisc.edu/condor Condor Daemon Layout Personal Condor / Central Manager Master collector negotiator startd = Process Spawned schedd

68. 68 http://www.cs.wisc.edu/condor condor_master › Starts up all other Condor daemons › Runs on all Condor hosts › If there are any problems and a daemon exits, it restarts the daemon and sends email to the administrator › Acts as the server for many Condor remote administration commands:  condor_reconfig, condor_restart  condor_off, condor_on  condor_config_val  etc.

69. 69 http://www.cs.wisc.edu/condor Central Manager: condor_collector › Central manager: central repository and match maker for whole pool › Collects information from all other Condor daemons in the pool  “Directory Service” / Database for a Condor pool  Each daemon sends a periodic update ClassAd to the collector › Services queries for information:  Queries from other Condor daemons  Queries from users (condor_status) › Only on the Central Manager(s) › At least one collector per pool

70. 70 http://www.cs.wisc.edu/condor Condor Pool Layout: Collector = ClassAd Communication Pathway = Process Spawned Central Manager Master Collector negotiator

71. 71 http://www.cs.wisc.edu/condor Central Manager: condor_negotiator › Performs “matchmaking” in Condor › Each “Negotiation Cycle” (typically 5 minutes):  Gets information from the collector about all available machines and all idle jobs  Tries to match jobs with machines that will serve them  Both the job and the machine must satisfy each other’s requirements › Only one Negotiator per pool  Ignoring HAD › Only on the Central Manager(s)

72. 72 http://www.cs.wisc.edu/condor Condor Pool Layout: Negotiator = ClassAd Communication Pathway = Process Spawned Central Manager Master Collector negotiator

73. 73 http://www.cs.wisc.edu/condor Execute Hosts: condor_startd › Execute host: machines that run user jobs › Represents a machine to the Condor system › Responsible for starting, suspending, and stopping jobs › Enforces the wishes of the machine owner (the owner’s “policy”… more on this in the administrator’s tutorial) › Creates a “starter” for each running job › One startd runs on each execute node

74. 74 http://www.cs.wisc.edu/condor Condor Pool Layout: startd = ClassAd Communication Pathway = Process Spawned Central Manager Master Collector schedd negotiator Cluster Node Master startd Cluster Node Master startd Workstation Master startd schedd Workstation Master startd schedd

75. 75 http://www.cs.wisc.edu/condor Submit Hosts: condor_schedd › Submit hosts: machines that users can submit jobs on › Maintains the persistent queue of jobs › Responsible for contacting available machines and sending them jobs › Services user commands which manipulate the job queue:  condor_submit, condor_rm, condor_q, condor_hold, condor_release, condor_prio, … › Creates a “shadow” for each running job › One schedd runs on each submit host

76. 76 http://www.cs.wisc.edu/condor Condor Pool Layout: schedd = ClassAd Communication Pathway = Process Spawned Cluster Node Master startd Cluster Node Master startd Central Manager Master Collector schedd negotiator Workstation Master startd schedd Workstation Master startd schedd

77. 77 http://www.cs.wisc.edu/condor Condor Pool Layout: master = ClassAd Communication Pathway = Process Spawned Central Manager Master Collector schedd negotiator Cluster Node Master startd Cluster Node Master startd Cluster Node Master startd schedd Cluster Node Master startd schedd

78. 78 http://www.cs.wisc.edu/condor What about these “condor_shadow” processes? › The Shadow processes are Condor’s local representation of your running job  One is started for each job › Similarly, on the “execute” machine, a condor_starter is run for each job

79. 79 http://www.cs.wisc.edu/condor Condor Pool Layout: running a job = Communication Pathway = Process Spawned Submit Host Master schedd shadow Execute Host Master startd starter Job Execute Host Master startd starter Job shadow

80. 80 http://www.cs.wisc.edu/condor Now what? › Some of the machines in the pool can’t run my jobs  Not enough RAM  Not enough scratch disk space  Required software not installed  Etc.

81. 81 http://www.cs.wisc.edu/condor Specify Requirements › An expression (syntax similar to C or Java) › Must evaluate to True for a match to be made Universe = vanilla Executable = cosmos Log = cosmos.log InitialDir = run_$(Process) Requirements = Memory >= 4096 && Disk > 10000 Queue 1000000

82. 82 http://www.cs.wisc.edu/condor Advanced Requirements › Requirements can match custom attributes in your Machine Ad  Can be added by hand to each machine  Or, automatically using the “Hawkeye” mechanism Universe = vanilla Executable = cosmos Log = cosmos.log InitialDir = run_$(Process) Requirements =(CosmosData =!= UNDEFINED) Queue 1000000

83. 83 http://www.cs.wisc.edu/condor CosmosData =!= UNDEFINED ??? › What’s this “ =!= ” and “ UNDEFINED” business ? Is this like the state of Schrödinger’s Cat? › Introducing ClassAd Meta Operators:  Allow you to test if a attribute is in a ClassAd  Is identical to operator: “ =?= ”  Is not identical to operator: “ =!= ”  Behave similar to == and !=, but are not strict  Somewhat like “ is NONE ” and “ is not NONE ” in Python  Without these, ANY expression with an UNDEFINED in it will always evaluate to UNDEFINED

84. 84 http://www.cs.wisc.edu/condor Meta Operator Examples › 10 == UNDEFINED evaluates to UNDEFINED › UNDEFINED == UNDEFINED evaluates to UNDEFINED › 10 =?= UNDEFINED evaluates to FALSE › UNDEFINED =?= UNDEFINED evaluates to TRUE › UNDEFINED =!= UNDEFINED evaluates to FALSE

85. 85 http://www.cs.wisc.edu/condor More Meta Operator Examples › Value == 10:  Evaluates to TRUE if Value is in the ad and is equal to 10  Evaluates to FALSE if Value is in the ad and is not equal to 10  Evaluates to UNDEFINED if Value is not in the ad › Value =!= UNDEFINED:  Evaluates to FALSE if Value is in the ad  Evaluates to TRUE if Value is not in the ad

86. 86 http://www.cs.wisc.edu/condor Final Meta Operator Examples › ( (Value =!= UNDEFINED) && (Value == 10) )  Evaluates to TRUE if Value is in the ad and is equal to 10  Evaluates to FALSE if Value is in the ad and is not equal to 10  Evaluates to FALSE if Value is not in the ad › ( (Value =?= UNDEFINED) || (Value != 10) )  Evaluates to FALSE if Value is in the ad and is equal to 10  Evaluates to TRUE if Value is in the ad and is not equal to 10  Evaluates to TRUE if Value is not in the ad

87. 87 http://www.cs.wisc.edu/condor Using Attributes from the Machine Ad › You can use attributes from the matched Machine Ad in your job submit file  $$(Name) will be replaced by the value of the specified attribute in the Machine Ad › Example:  Matching Machine Ad has: CosmosData = “/local/cosmos/data”  Submit file has: Executable = cosmos Requirements = (CosmosData =!= UNDEFINED) Arguments = -d $$(CosmosData)  Resulting command line: cosmos –d /local/cosmos/data

88. 88 http://www.cs.wisc.edu/condor And, Specify Rank › All matches which meet the requirements can be sorted by preference with a Rank expression. › Higher the Rank, the better the match Universe = vanilla Executable = cosmos Log = cosmos.log Arguments = -arg1 –arg2 InitialDir = run_$(Process) Requirements = Memory >= 4096 && Disk > 10000 Rank = (KFLOPS*10000) + Memory Queue 1000000

89. 89 http://www.cs.wisc.edu/condor My jobs aren’t running!!

90. 90 http://www.cs.wisc.edu/condor Check the queue › Check the queue with condor_q: bash-2.05a$ condor_q -- Submitter: x.cs.wisc.edu : :x.cs.wisc.edu ID OWNER SUBMITTED RUN_TIME ST PRI SIZE CMD 5.0 einstein 4/20 12:23 0+00:00:00 I 0 9.8 cosmos -arg1 –n 0 5.1 einstein 4/20 12:23 0+00:00:00 I 0 9.8 cosmos -arg1 –n 1 5.2 einstein 4/20 12:23 0+00:00:00 I 0 9.8 cosmos -arg1 –n 2 5.3 einstein 4/20 12:23 0+00:00:00 I 0 9.8 cosmos -arg1 –n 3 5.4 einstein 4/20 12:23 0+00:00:00 I 0 9.8 cosmos -arg1 –n 4 5.5 einstein 4/20 12:23 0+00:00:00 I 0 9.8 cosmos -arg1 –n 5 5.6 einstein 4/20 12:23 0+00:00:00 I 0 9.8 cosmos -arg1 –n 6 5.7 einstein 4/20 12:23 0+00:00:00 I 0 9.8 cosmos -arg1 –n 7 6.0 einstein 4/20 13:22 0+00:00:00 H 0 9.8 cosmos -arg1 –arg2 8 jobs; 8 idle, 0 running, 1 held

91. 91 http://www.cs.wisc.edu/condor Look at jobs on hold % condor_q –hold -- Submiter: x.cs.wisc.edu : :x.cs.wisc.edu ID OWNER HELD_SINCE HOLD_REASON 6.0 einstein 4/20 13:23 Error from starter on skywalker.cs.wisc.edu 9 jobs; 8 idle, 0 running, 1 held Or, See full details for a job % condor_q –l 6.0

92. 92 http://www.cs.wisc.edu/condor Check machine status › Verify that there are idle machines with condor_status: bash-2.05a$ condor_status Name OpSys Arch State Activity LoadAv Mem ActvtyTime vm1@tonic.c LINUX INTEL Claimed Busy 0.000 501 0+00:00:20 vm2@tonic.c LINUX INTEL Claimed Busy 0.000 501 0+00:00:19 vm3@tonic.c LINUX INTEL Claimed Busy 0.040 501 0+00:00:17 vm4@tonic.c LINUX INTEL Claimed Busy 0.000 501 0+00:00:05 Total Owner Claimed Unclaimed Matched Preempting INTEL/LINUX 4 0 4 0 0 0 Total 4 0 4 0 0 0

93. 93 http://www.cs.wisc.edu/condor Look in Job Log › Look in your job log for clues: bash-2.05a$ cat cosmos.log 000 (031.000.000) 04/20 14:47:31 Job submitted from host:... 007 (031.000.000) 04/20 15:02:00 Shadow exception! Error from starter on gig06.stat.wisc.edu: Failed to open '/scratch.1/einstein/workspace/v67/condor- test/test3/run_0/cosmos.in' as standard input: No such file or directory (errno 2) 0 - Run Bytes Sent By Job 0 - Run Bytes Received By Job...

94. 94 http://www.cs.wisc.edu/condor Still not running? Exercise a little patience › On a busy pool, it can take a while to match and start your jobs › Wait at least a negotiation cycle or two (typically a few minutes)

95. 95 http://www.cs.wisc.edu/condor Look to condor_q for help: condor_q -analyze bash-2.05a$ condor_q -ana 29 --- 029.000: Run analysis summary. Of 1243 machines, 1243 are rejected by your job's requirements 0 are available to run your job WARNING: Be advised: No resources matched request's constraints Check the Requirements expression below: Requirements = ((Memory > 8192)) && (Arch == "INTEL") && (OpSys == "LINUX") && (Disk >= DiskUsage) && (TARGET.FileSystemDomain == MY.FileSystemDomain)

96. 96 http://www.cs.wisc.edu/condor Better analysis: condor_q –better-analyze bash-2.05a$ condor_q -better-ana 29 The Requirements expression for your job is: ( ( target.Memory > 8192 ) ) && ( target.Arch == "INTEL" ) && ( target.OpSys == "LINUX" ) && ( target.Disk >= DiskUsage ) && ( TARGET.FileSystemDomain == MY.FileSystemDomain ) Condition Machines Matched Suggestion --------- ---------------- ---------- 1 ( ( target.Memory > 8192 ) ) 0 MODIFY TO 4000 2 ( TARGET.FileSystemDomain == "cs.wisc.edu" )584 3 ( target.Arch == "INTEL" ) 1078 4 ( target.OpSys == "LINUX" ) 1100 5 ( target.Disk >= 13 ) 1243

97. 97 http://www.cs.wisc.edu/condor Learn about available resources: bash-2.05a$ condor_status –const 'Memory > 8192' (no output means no matches) bash-2.05a$ condor_status -const 'Memory > 4096' Name OpSys Arch State Activ LoadAv Mem ActvtyTime vm1@s0-03.cs. LINUX X86_64 Unclaimed Idle 0.000 5980 1+05:35:05 vm2@s0-03.cs. LINUX X86_64 Unclaimed Idle 0.000 5980 13+05:37:03 vm1@s0-04.cs. LINUX X86_64 Unclaimed Idle 0.000 7988 1+06:00:05 vm2@s0-04.cs. LINUX X86_64 Unclaimed Idle 0.000 7988 13+06:03:47 Total Owner Claimed Unclaimed Matched Preempting X86_64/LINUX 4 0 0 4 0 0 Total 4 0 0 4 0 0

98. 98 http://www.cs.wisc.edu/condor Job Policy Expressions › User can supply job policy expressions in the submit file. › Can be used to describe a successful run. on_exit_remove = on_exit_hold = periodic_remove = periodic_hold =

99. 99 http://www.cs.wisc.edu/condor Job Policy Examples › Do not remove if exits with a signal:  on_exit_remove = ExitBySignal == False › Place on hold if exits with nonzero status or ran for less than an hour:  on_exit_hold = ( (ExitBySignal==False) && (ExitSignal != 0) ) || ( (ServerStartTime - JobStartDate) < 3600) › Place on hold if job has spent more than 50% of its time suspended:  periodic_hold = CumulativeSuspensionTime > (RemoteWallClockTime / 2.0)

100. 100 http://www.cs.wisc.edu/condor Insert ClassAd attributes › Special purpose usage › In the submit description file, introduce an attribute for the job +Department = “physics” causes the ClassAd to contain Department = “physics”

101. 101 http://www.cs.wisc.edu/condor We’ve seen how Condor can: › Keep an eye on your jobs  Keep you posted on their progress › Implement your policy on the execution order of the jobs › Keep a log of your job activities

102. 102 http://www.cs.wisc.edu/condor Getting Condor › Available as a free download from http://www.cs.wisc.edu/condor › Download Condor for your operating system  Available for most UNIX (including Linux and Apple’s OS/X) platforms  Also for Windows XP / 2003 / Vista

103. 103 http://www.cs.wisc.edu/condor Condor Releases › Stable / Developer Releases  Version numbering scheme similar to that of the (pre 2.6) Linux kernels … › Major.minor.release  Minor is even (a.b.c): Stable Examples: 7.0.5, 7.2.2 –7.2.2 just released Very stable, mostly bug fixes  Minor is odd (a.b.c): Developer New features, may have some bugs Examples: 7.1.5, 7.3.1 –7.3.1 “real soon now”

104. 104 http://www.cs.wisc.edu/condor Albert installs a “Personal Condor” on his laptop… › What do we mean by a “Personal” Condor?  Condor on your own workstation  No root / administrator access required  No system administrator intervention needed › His Personal Condor can joins the local Condor pool  He can now access to the pool’s resources (machines) › Can also be “Stand Alone”  Can still run his own jobs, maintain job queue, etc.  Can even access grid resources if needed!

105. 105 http://www.cs.wisc.edu/condor Multiple Universes

106. 106 http://www.cs.wisc.edu/condor Einstein submits some MPI jobs # Example condor_submit input file that for MPI # jobs universe = parallel executable = mp1script arguments = my_mpich_linked_executable arg1 arg2 machine_count = 4 should_transfer_files = yes when_to_transfer_output = on_exit transfer_input_files = my_mpich_linked_executable queue

107. 107 http://www.cs.wisc.edu/condor My new jobs can run for 20 days… What happens when a job is forced off it’s CPU? – Preempted by higher priority user or job – Vacated because of user activity How can I add fault tolerance to my jobs?

108. 108 http://www.cs.wisc.edu/condor Condor’s Standard Universe to the rescue! › Support for transparent process checkpoint and restart › Remote system calls (remote I/O)  Your job can read / write files as if they were local

109. 109 http://www.cs.wisc.edu/condor Remote System Calls in the Standard Universe › I/O system calls are trapped and sent back to the submit machine Examples: open a file, write to a file › No source code changes typically required › Programming language independent

110. 110 http://www.cs.wisc.edu/condor Process Checkpointing in the Standard Universe › Condor’s process checkpointing provides a mechanism to automatically save the state of a job › The process can then be restarted from right where it was checkpointed  After preemption, crash, etc.

111. 111 http://www.cs.wisc.edu/condor Checkpointing: Process Starts checkpoint: the entire state of a program, saved in a file  CPU registers, memory image, I/O time

112. 112 http://www.cs.wisc.edu/condor Checkpointing: Process Checkpointed time 123

113. 113 http://www.cs.wisc.edu/condor Checkpointing: Process Killed time 3 3 Killed!

114. 114 http://www.cs.wisc.edu/condor Checkpointing: Process Resumed time 3 3 goodputbadput goodput

115. 115 http://www.cs.wisc.edu/condor When will Condor checkpoint your job? › Periodically, if desired  For fault tolerance › When your job is preempted by a higher priority job › When your job is vacated because the execution machine becomes busy › When you explicitly run condor_checkpoint, condor_vacate, condor_off or condor_restart command

116. 116 http://www.cs.wisc.edu/condor Making the Standard Universe Work › The job must be relinked with Condor’s standard universe support library › To relink, place condor_compile in front of the command used to link the job: % condor_compile gcc -o myjob myjob.c - OR - % condor_compile f77 -o myjob filea.f fileb.f - OR - % condor_compile make –f MyMakefile

117. 117 http://www.cs.wisc.edu/condor Limitations of the Standard Universe › Condor’s checkpointing is not at the kernel level.  Standard Universe the job may not: Fork() Use kernel threads Use some forms of IPC, such as pipes and shared memory › Must have access to source code to relink › Many typical scientific jobs are OK

118. 118 http://www.cs.wisc.edu/condor Death of the Standard Universe

119. 119 http://www.cs.wisc.edu/condor DMTCP & Parrot › DMTCP (Checkpointing)  “Distributed MultiThreaded Checkpointing”  Developed at Northeastern University  http://dmtcp.sourceforge.net/ http://dmtcp.sourceforge.net/  See Pete’s talk tomorrow @ 11:40 › Parrot (Remote I/O)  “Parrot is a tool for attaching existing programs to remote I/O system”  Developed by Doug Thain (now at Notre Dame)  http://www.cse.nd.edu/~ccl/software/parrot/  dthain@nd.edu

120. 120 http://www.cs.wisc.edu/condor Connecting Condors Albert knows people with their own Condor pools, and gets permission to use their computing resources… How can Condor help him do this?

121. 121 http://www.cs.wisc.edu/condor Connect Condors with Flocking › Flocking is a Condor-specific technology flock › Albert configures his Condor schedd to “flock” to his friend’s pool.

122. 122 http://www.cs.wisc.edu/condor Condor Pool 1,000,000 Condor jobs Einstein’s Condor Pool Friendly Condor Pool

123. 123 http://www.cs.wisc.edu/condor Albert meets The Grid › Albert also has access to grid resources he wants to use  He has certificates and access to Globus or other resources at remote institutions › But Albert wants Condor’s queue management features for his jobs! › He installs Condor so he can submit “Grid Universe” jobs to Condor

124. 124 http://www.cs.wisc.edu/condor “Grid” Universe › All handled in your submit file › Supports a number of “back end” types:  Globus: GT2, GT4  NorduGrid  UNICORE  Condor  PBS  LSF  EC2  NQS

125. 125 http://www.cs.wisc.edu/condor Grid Universe & Globus 2 › Used for a Globus GT2 back-end  “Condor-G” › Format: Grid_Resource = gt2 Head-Node Globus_rsl = › Example: Universe = grid Grid_Resource = gt2 beak.cs.wisc.edu/jobmanager Globus_rsl = (queue=long)(project=atom-smasher)

126. 126 http://www.cs.wisc.edu/condor Grid Universe & Globus 4 › Used for a Globus GT4 back-end › Format: Grid_Resource = gt4 Globus_XML = › Example: Universe = grid Grid_Resource = gt4 beak.cs.wisc.edu Condor Globus_xml = long atom- smasher

127. 127 http://www.cs.wisc.edu/condor Grid Universe & Condor › Used for a Condor back-end  “Condor-C” › Format: Grid_Resource = condor Remote_ =  “Remote_” part is stripped off › Example: Universe = grid Grid_Resource = condor beak condor.cs.wisc.edu Remote_Universe = standard

128. 128 http://www.cs.wisc.edu/condor Grid Universe & NorduGrid › Used for a NorduGrid back-end Grid_Resource = nordugrid › Example: Universe = grid Grid_Resource = nordugrid ngrid.cs.wisc.edu

129. 129 http://www.cs.wisc.edu/condor Grid Universe & UNICORE › Used for a UNICORE back-end › Format: Grid_Resource = unicore › Example: Universe = grid Grid_Resource = unicore uhost.cs.wisc.edu vhost

130. 130 http://www.cs.wisc.edu/condor Grid Universe & PBS › Used for a PBS back-end › Format: Grid_Resource = pbs › Example: Universe = grid Grid_Resource = pbs

131. 131 http://www.cs.wisc.edu/condor Grid Universe & LSF › Used for a LSF back-end › Format: Grid_Resource = lsf › Example: Universe = grid Grid_Resource = lsf

132. 132 http://www.cs.wisc.edu/condor Credential Management › Condor will do The Right Thing™ with your X509 certificate and proxy › Override default proxy:  X509UserProxy = /home/einstein/other/proxy › Proxy may expire before jobs finish executing  Condor can use MyProxy to renew your proxy  When a new proxy is available, Condor will forward the renewed proxy to the job  This works for non-grid jobs, too

133. 133 http://www.cs.wisc.edu/condor Condor Universes: Scheduler and Local › Scheduler Universe  Plug in a meta-scheduler  Similar to Globus’s fork job manager  Developed for DAGMan (next slide) › Local  Very similar to vanilla, but jobs run on the local host  Has more control over jobs than scheduler universe

134. 134 http://www.cs.wisc.edu/condor My jobs have have dependencies… Can Condor help solve my dependency problems?

135. 135 http://www.cs.wisc.edu/condor Einstein learns DAGMan › Directed Acyclic Graph Manager › DAGMan allows you to specify the dependencies between your Condor jobs, so it can manage them automatically for you.  E.G.: “Don’t run job ‘B’ until job ‘A’ has completed successfully.”

136. 136 http://www.cs.wisc.edu/condor What is a DAG? › A DAG is the data structure used by DAGMan to represent these dependencies. › Each job is a “node” in the DAG. › Each node can have any number of “parent” or “children” nodes – as long as there are no loops! Job D Job A Job B Job C

137. 137 http://www.cs.wisc.edu/condor DAGs large and small › DAGMan can scale up to huge DAGs  LIGO runs 500,000 node DAGs › DAGMan is useful even in the small cases!  What about a simple 2 job DAG? Run database query Process output Can all be run on one machine--no Condor pool necessary › DAGMan can be used as a Condor job throttle  Lots of nodes with no dependencies

138. 138 http://www.cs.wisc.edu/condor Defining a DAG › A DAG is defined by a.dag file, listing each of its nodes and their dependencies: # diamond.dag Job A a.sub Job B b.sub Job C c.sub Job D d.sub Parent A Child B C Parent B C Child D › each node will run the Condor job specified by its accompanying Condor submit file Job D Job A Job B Job C

139. 139 http://www.cs.wisc.edu/condor Submitting a DAG › To start your DAG, just run condor_submit_dag with your.dag file  Condor will start a personal DAGMan daemon which to begin running your jobs: % condor_submit_dag diamond.dag › condor_dagman is then run by the schedd  Thus, DAGMan daemon itself is “watched” by Condor, so you don’t have to

140. 140 http://www.cs.wisc.edu/condor DAGMan DAG Running: A › DAGMan acts as a “meta-scheduler”, managing the submission of your jobs to Condor based on the DAG dependencies. Condor Job Queue.dag File B A C D A

141. 141 http://www.cs.wisc.edu/condor DAGMan DAG Running: B & C › DAGMan holds & submits jobs to the Condor queue at the appropriate times. Condor Job Queue BC D A B C

142. 142 http://www.cs.wisc.edu/condor DAG Running: C failed › In case of a job failure, DAGMan continues until it can no longer make progress, and then creates a “rescue” file with the current state of the DAG. Condor Job Queue DAGMan Rescue File A BC D

143. 143 http://www.cs.wisc.edu/condor DAG Recovery: C › Once the failed job is ready to be re-run, the rescue file can be used to restore the prior state of the DAG. Condor Job Queue Rescue File DAGMan A BC DC

144. 144 http://www.cs.wisc.edu/condor DAG Recovery: D › Once that job completes, DAGMan will continue the DAG as if the failure never happened. Condor Job Queue DAGMan A BC DD

145. 145 http://www.cs.wisc.edu/condor DAG Complete › Once the DAG is complete, the DAGMan job itself is finished, and exits. Condor Job Queue DAGMan A BC D

146. 146 http://www.cs.wisc.edu/condor Additional DAGMan Features › Provides other handy features for job management…  Nodes can have PRE & POST scripts  Failed nodes can be automatically re- tried a configurable number of times  Job submission can be “throttled”

147. 147 http://www.cs.wisc.edu/condor Want more from DAGMan? › What if:  You want to define workflows that can flexibly take advantage of different grid resources?  You want to register data products in a way that makes them available to others?  You want to use the grid without a full Condor installation?

148. 148 http://www.cs.wisc.edu/condor Introducing Pegasus WMS (Workflow Management System) › A higher level on top of DAGMan › User creates an abstract workflow › Pegasus maps abstract workflow to executable workflow › DAGMan runs executable workflow › Doesn’t need full Condor (schedd only)

149. 149 http://www.cs.wisc.edu/condor Pegasus Features › Workflow has inter-job dependencies (similar to DAGMan) › Pegasus can map jobs to grid sites › Pegasus handles discovery and registration of data products › Pegasus handles data transfer to/from grid sites

150. 150 http://www.cs.wisc.edu/condor *The full moon is 0.5 deg. sq. when viewed form Earth, Full Sky is ~ 400,000 deg. sq. Generating mosaics of the sky (Bruce Berriman, Caltech) Size of the mosaic in degrees square* Number of jobs Number of input data files Number of intermediate files Total data footprint Approx. execution time (20 procs) 1232535881.2GB40 mins 21,4442123,9065.5GB49 mins 44,85674713,06120GB1hr 46 mins 68,5861,44422,85038GB2 hrs. 14 mins 1020,6523,72254,43497GB6 hours

151. 151 http://www.cs.wisc.edu/condor Abstract Workflow (DAX) › Pegasus workflow description—DAX  In XML Java API to generate XML for you  Workflow “high-level language”  Devoid of resource descriptions  Devoid of data locations  Refers to codes as logical transformations  Refers to data as logical files  Specifies dependencies Akin to a high-level DAG file

152. 152 http://www.cs.wisc.edu/condor Basic Workflow Mapping › Mapping is Translating abstract DAX into an executable DAG › Select where to run the computations  Change task nodes into nodes with executable descriptions › Select which data to access  Add stage-in and stage-out nodes to move data › Add nodes that register the newly-created data products › Add nodes to create an execution directory on a remote site › Write out the workflow in a form understandable by a workflow engine  Include provenance capture steps

153. 153 http://www.cs.wisc.edu/condor Pegasus Workflow Mapping Original workflow: 15 compute nodes devoid of resource assignment Resulting workflow mapped onto 3 Grid sites: 11 compute nodes (4 reduced based on available intermediate data) 13 data stage-in nodes 8 inter-site data transfers 14 data stage-out nodes to long-term storage 14 data registration nodes (data cataloging) 41 85 10 9 13 12 15 9 4 8 3 7 10 13 12 15

154. 154 http://www.cs.wisc.edu/condor Pegasus WMS Pegasus Workflow Mapper Condor DAGMan TeraGrid Open Science Grid Campus resources Local machine Transformation Catalog Site Catalog Workflow Description in XML (DAX) Condor Schedd Submit Host Replica Catalog Pegasus WMS restructures and optimizes the workflow, provides reliability Properties

155. 155 http://www.cs.wisc.edu/condor Mapping & Running a Workflow › To map a workflow, run pegasus-plan: % pegasus-plan -Dpegasus.user.properties=pegasus- wms/config/properties --dir dags --sites viz -- output local --force --nocleanup --dax pegasus- wms/dax/montage.dax  Creates executable workflow › To run a workflow, run pegasus-run: % pegasus-run -Dpegasus.user.properties=pegasus- wms/dags/train01/pegasus/montage/run0001/pegasus. 51773.properties pegasus- wms/dags/train01/pegasus/montage/run0001  Runs condor_submit_dag and other tools

156. 156 http://www.cs.wisc.edu/condor There’s much more… › We’ve only scratched the surface of Pegasus’s capabilities › Resources:  Pegasus home page: pegasus.isi.edupegasus.isi.edu  Tutorial materials available at: http://pegasus.isi.edu/tutorials.php http://pegasus.isi.edu/tutorials.php  Questions: pegasus@isi.edupegasus@isi.edu  Also, talk to Kent Wenger (Condor Team) wenger@cs.wisc.edu

157. 157 http://www.cs.wisc.edu/condor General User Commands › condor_status View Pool Status › condor_qView Job Queue › condor_submitSubmit new Jobs › condor_rmRemove Jobs › condor_prioIntra-User Prios › condor_historyCompleted Job Info › condor_submit_dagSubmit new DAG › condor_checkpointForce a checkpoint › condor_compileLink Condor library

158. 158 http://www.cs.wisc.edu/condor Condor Job Universes Vanilla Universe Standard Universe Grid Universe Scheduler Universe Local Universe Virtual Machine Universe Java Universe Parallel Universe MPICH-1 MPICH-2 LAM …

159. 159 http://www.cs.wisc.edu/condor Why have a special Universe for Java jobs? › Java Universe provides more than just inserting “java” at the start of the execute line of a vanilla job:  Knows which machines have a JVM installed  Knows the location, version, and performance of JVM on each machine  Knows about jar files, etc.  Provides more information about Java job completion than just JVM exit code Program runs in a Java wrapper, allowing Condor to report Java exceptions, etc.

160. 160 http://www.cs.wisc.edu/condor Universe Java Job # Example Java Universe Submit file Universe = java Executable = Main.class jar_files = MyLibrary.jar Input = infile Output = outfile Arguments = Main 1 2 3 Queue

161. 161 http://www.cs.wisc.edu/condor Java support, cont. bash-2.05a$ condor_status –java Name JavaVendor Ver State Actv LoadAv Mem abulafia.cs Sun Microsy 1.5.0_ Claimed Busy 0.180 503 acme.cs.wis Sun Microsy 1.5.0_ Unclaimed Idle 0.000 503 adelie01.cs Sun Microsy 1.5.0_ Claimed Busy 0.000 1002 adelie02.cs Sun Microsy 1.5.0_ Claimed Busy 0.000 1002 … Total Owner Claimed Unclaimed Matched Preempting INTEL/LINUX 965 179 516 250 20 0 INTEL/WINNT50 102 6 65 31 0 0 SUN4u/SOLARIS28 1 0 0 1 0 0 X86_64/LINUX 128 2 106 20 0 0 Total 1196 187 687 302 20 0

162. 162 http://www.cs.wisc.edu/condor Albert wants Condor features on remote resources › He wants to run standard universe jobs on Grid-managed resources  For matchmaking and dynamic scheduling of jobs  For job checkpointing and migration  For remote system calls

163. 163 http://www.cs.wisc.edu/condor Condor GlideIn › Albert can use the Grid Universe to run Condor daemons on Grid resources › When the resources run these GlideIn jobs, they will temporarily join his Condor Pool › He can then submit Standard, Vanilla, PVM, or MPI Universe jobs and they will be matched and run on the remote resources › Currently only supports Globus GT2  We hope to fix this limitation

164. 164 http://www.cs.wisc.edu/condor your workstation Friendly Condor Pool personal Condor 600 Condor jobs Globus Grid PBS LSF Condor Condor Pool glide-in jobs

165. 165 http://www.cs.wisc.edu/condor How It Works Manager LSF User Job Startd Personal CondorRemote Resource Condor jobs GlideIn jobs Starter ScheddCollector & Negotiator Grid Manager Shadow Master

166. 166 http://www.cs.wisc.edu/condor GlideIn Concerns › What if the remote resource kills my GlideIn job?  That resource will disappear from your pool and your jobs will be rescheduled on other machines  Standard universe jobs will resume from their last checkpoint like usual › What if all my jobs are completed before a GlideIn job runs?  If a GlideIn Condor daemon is not matched with a job in 10 minutes, it terminates, freeing the resource

167. 167 http://www.cs.wisc.edu/condor In Review With Condor’s help, Albert can:  Manage his compute job workload  Access local machines  Access remote Condor Pools via flocking  Access remote compute resources on the Grid via “Grid Universe” jobs  Carve out his own personal Condor Pool from the Grid with GlideIn technology

168. 168 http://www.cs.wisc.edu/condor Administrator Commands › condor_vacateLeave a machine now › condor_onStart Condor › condor_offStop Condor › condor_reconfigReconfig on-the-fly › condor_config_valView/set config › condor_userprioUser Priorities › condor_statsView detailed usage accounting stats

169. 169 http://www.cs.wisc.edu/condor My boss wants to watch what Condor is doing

170. 170 http://www.cs.wisc.edu/condor Use CondorView! › Provides visual graphs of current and past utilization › Data is derived from Condor's own accounting statistics › Interactive Java applet › Quickly and easily view:  How much Condor is being used  How many cycles are being delivered  Who is using them  Utilization by machine platform or by user

171. 171 http://www.cs.wisc.edu/condor CondorView Usage Graph

172. 172 http://www.cs.wisc.edu/condor I could also talk lots about… › GCB: Living with firewalls & private networks › Federated Grids/Clusters › APIs and Portals › MW › Database Support (Quill) › High Availability Fail-over › Compute On-Demand (COD) › Dynamic Pool Creation (“Glide-in”) › Role-based prioritization and accounting › Strong security, incl privilege separation › Data movement scheduling in workflows › …

173. 173 http://www.cs.wisc.edu/condor Thank you! Check us out on the Web: http://www.condorproject.org Email: condor-admin@cs.wisc.edu