1. gLite Advanced Job Management
Africa Joint EUMEDGRID-Support/EPIKH School for Application Porting
Elisa Ingrà – Consortium GARR
Algiers, 4-16 July 2010

2. Log Lifetime Proxy
Do you need a long lifetime proxy for your application?
Use MyProxy!
How?!
Create a long proxy ( myproxy-init -d -n -s changeme-myproxyserver)
-d Use the proxy certificate subject (DN) as the default username
-n Don't prompt for passphrase
-s <hostname> Hostname of the myproxy-server
Create a normal proxy with vo extensions (voms-proxy-init --voms eumed/gilda)
In jdl file specify myproxy server name (MyProxyServer = “changeme-myproxyserver";)
Submit your job
For this tutorial you can use myproxy.ct.infn.it as MyProxyServer
Location, Meeting title, dd.mm.yyyy

3. Outline WMProxy Overview Special Jobs MPI Jobs DAG jobs
Job collections
Parametric jobs
MPI Jobs

4. WMProxy Overview
is a service providing access to the gLite Workload Management System (WMS) functionality through a simple Web Services based interface.
has been designed to handle a large number of requests for job submission
gLite 3.0=> ~180 secs for 500 jobs
goal is to get in the short term to ~60 secs for 1000 jobs
it provides additional features such as bulk submission and the support for shared and compressed sandboxes for compound jobs.
Now is co-existing with the WMS.

5. New request types
Support for new types strongly relies on newly developed JDL converters and on the DAG submission support
- all JDL conversions are performed on the server
- a single submission for several jobs
All new request types can be monitored and controlled through a single request id
- each sub-jobs can be however followed-up and controlled independently through its own id
“Smarter” WMS client commands/API
- allow submission of DAGs, collections and parametric jobs exploiting the concept of “shared sandbox”
- allow automatic generation and submission of collections and DAGs from sets of JDL files located in user specified directories on the UI
Crei un jdl per i diversi job e questo viene convertito in più job. Ogni job può essere controllato indipendentemente attraverso il suo id
Le api messe a disposizione permettono di sottomettere dag collectio e parametrici job sfruttando il concetto di sandbox condivise

6. Outline
Special Jobs
DAG Job
Job Collection
Parametric jobs

7. DAG jobs
A DAG job is a set of jobs where input, output, or execution of one or more jobs can depend on other jobs
Dependencies are represented through Directed Acyclic Graphs, where the nodes are jobs, and the edges identify the dependencies
nodeA
node
nodeC
nodeB
node D
nodeE

8. DAG -- JDL structure The JDL description of a DAG must
not contain any OutputSandbox attribute
occurrence also the ones in the descriptions
of the nodes. The OutputSandbox of the DAG
has to be considered as the sum
of the output sandboxes of all its nodes.
Max_node_running: the maximum number of resources involved in job running. You have not to specify the Output sandbox attribute here.
The most important attribute of a dag job is the nodes attribute. In this attribute you have to specify the subjobs of the dag and their dependencies.

9. Attribute: InputSandbox
All nodes that do not contain the InputSandbox attribute in their descriptions inherit the value of these attributes from the one specified for the DAG.
Nodes representing jobs without InputSandbox have to contain the following specification in their description (empy InputSandbox list):
The nodes attribute is composed by a set of attributes where you specify the jdl of the sub-job. You can specify the jdl of a sub-job or with a reference to a file or with a explicit description as in this case. [
…………….
InputSandbox = {};
…………… ]

10. Attribute: Nodes
The nodes attribute is composed by a set of attributes where you specify the jdl of the sub-job. You can specify the jdl of a sub-job or with a reference to a file or with a explicit description as in this case.

11. Attribute: File
The File attribute is a string representing the path on the local file system ofù a file containing the JDL description of a Job. It is important to note that this kind of representation can only be used when submitting to the WMS through a client (glite-wms-job-submit) able to resolve the path locally and to expand the JDL with the full description before passing it to the WMS.
The File attribute cannot be specified together with the Description attribute within the same node description!
The nodes attribute is composed by a set of attributes where you specify the jdl of the sub-job. You can specify the jdl of a sub-job or with a reference to a file or with a explicit description as in this case.

12. Attribute: Dependencies
With The attribute Dependencies you can specify the node dependencies. In the sample on the slide the subjod nodefilename2 depends on nodefilename1 than, the job nodefilename2 can start to run when the job nodefilename1 is done.

13. DAG jdl [ type = "dag"; max_nodes_running = 4; nodes = [ nodeA = [
file ="nodes/nodeA.jdl" ; ];
nodeB = [
file ="nodes/nodeB.jdl" ;
nodeC = [
file ="nodes/nodeC.jdl" ;
nodeD = [
file ="nodes/nodeD.jdl";
dependencies = {
{nodeA, nodeB},
{nodeA, nodeC},
{ {nodeB,nodeC}, nodeD } } ]
This is a complete dag job composed by 4 sub-jobs. In this jdl the sub-jobs are specified with a file name but it’s possible also to insert the description of the sub-job directly in the jdl.

14. Job Collection
A job collection is a set of independent jobs that user wants to submit and monitor as a single request
Jobs of a collection are submitted as DAG nodes without dependencies
JDL is a list of classad, which describes the subjobs [
Type = "collection";
VirtualOrganisation = “gilda";
nodes = {
[ <job descr 1 >],
[ <job descr 2 >], … }; ]
A collection is a special DAG without dependencies. In this case you have to set the Type attribute to collection. The node attribute is the same of the node attr. of the dag job but without the dependencies attribute.

15. Input Sandboxes Input Sandbox can contain:
pointer to other files within the DAG/collection
URI pointing to files on a remote gridFTP/HTTPS server
file paths on the UI machine (i.e. the usual way)
Only local files (file://) are uploaded to the WMS node
File pointed by URIs are directly downloaded on the WN by the JobWrapper just before the job is started
InputSandbox = {
"gsiftp://neo.datamat.it:2811/var/prg/sim.exe",
root.nodes.nodeA.description.OutputSandbox[0],
"file:///home/pacio/myconf“ };

16. Output Sandboxes
The OutputSandbox attribute lists the files destination of the job output
A base URI to be applied to all sandbox files can also be specified
Files are copied when the job has completed execution by the JobWrapper to the specified destination without transiting on the WMS node
OutputSandbox = { "jobOutput","run1/event1",
"jobError" };
OutputSandboxDestURI = {
"gsiftp://matrix.datamat.it/var/jobOutput",
"
"gsiftp://matrix.datamat.it/var/jobError" };
OutputSandboxBaseDestURI = "gsiftp://neo.datamat.it/home/run1/";

17. All nodes will share this Input Sandbox
Job collection example [
type = "collection";
InputSandbox = {"date.sh"};
RetryCount = 3;
nodes = {
file ="jobs/job1.jdl" ; ],
Executable = "/bin/sh";
Arguments = "date.sh";
StdOutput = "date.out";
StdError = "date.err";
OutputSandbox ={"date.out", "date.err"}; ]
file ="jobs/job3.jdl" ; };
All nodes will share this Input Sandbox

18. Parametric Job
A parametric job is a job where one or more of its attributes are parameterized
Values of attributes vary according to a parameter
Job monitoring / managing is always done through an unique jobID, as if the job was single (see submission of collection) [
JobType = "Parametric";
Executable = "/bin/sh";
Arguments = "md5.sh input_PARAM_.txt";
InputSandbox = {"md5.sh", "input_PARAM_.txt"};
StdOutput = "out_PARAM_.txt";
StdError = "err_PARAM_.txt";
Parameters = 4;
ParameterStart = 1;
ParameterStep = 1;
OutputSandbox = {"out_PARAM_.txt", "err_PARAM_.txt"}; ]

19. Parametric job
Parameter can be either a number, or a list of items (typically strings, but not enclosed within double quotes)
Input Sandbox (if present) has to be coherent with parameters
[ui-test] /home/giorgio/param > cat param2.jdl [
JobType = "Parametric";
Executable = “/bin/cat";
Arguments = “input_PARAM_.txt”;
InputSandbox = "input_PARAM_.txt";
StdOutput = "myoutput_PARAM_.txt";
StdError = "myerror_PARAM_.txt";
Parameters = {EARTH,MOON,MARS};
OutputSandbox = {“myoutput_PARAM_.txt”}; ]
[ui-test] /home/giorgio/param > ls
inputEARTH.txt inputMARS.txt inputMOON.txt param2.jdl
It is the list of the values the parameter must take.

20. The parameter is a number
Parametric job [
JobType = "Parametric";
Executable = "myjob.exe";
StdInput = "input_PARAM_.txt";
StdOutput = "output_PARAM_.txt";
StdError = "error_PARAM_.txt";
Parameters = 100;
ParameterStart = 1;
ParameterStep = 1;
InputSandbox = {"myjob.exe", "input_PARAM_.txt";
OutputSandbox = {"output_PARAM_.txt", "error_PARAM_.txt"}; ]
The parameter is a number
the initial number of the running paramenter,
the increment of the running parameter between consecutive jobs.
Both attributes, ParameterStart and ParameterStep, can be set only if Parameters is a number.

21. MPI Overview
Execution of parallel jobs is an essential issue for modern informatics and applications.
Most used library for parallel jobs support is MPI (Message Passing Interface)
At the state of the art, parallel jobs can run inside single Computing Elements (CE) only;
several projects are involved into studies concerning the possibility of executing parallel jobs on Worker Nodes (WNs) belonging to different CEs.
The source code must have been compiled with mpicc libraries

22. MPI JDL Type = "Job"; #Mandatory JobType = "MPICH";
> $ cat mpi.jdl [
Type = "Job";
#Mandatory
JobType = "MPICH";
#The number of CPU that will be used
NodeNumber = 2;
Executable = "cpi";
StdOutput = "cpi.out";
StdError = "cpi.err";
InputSandbox = {"cpi"};
OutputSandbox = {"cpi.err","cpi.out"};
RetryCount = 3; ]
First South Africa Grid Training in Catania

23. Practice... Compound jobs More on jdl attributes
More on jdl attributes
Jdl and data management (Is your input file>10 MB? )
A Grid application case 23
First South Africa Grid Training in Catania

24. References https://edms.cern.ch/file/722398/1.2/gLite-3-UserGuide.pdf
gLite 3 User Guide (pdf, html)
WMProxy and JDL Guide
Attributes-v0-9.pdf
Documentation
p/index.html
Examples
MPI

25. Questions…. 16-26 June 2008, Catania (Italy)
First South Africa Grid Training in Catania