alien tutorial

Post on 25-Feb-2016

59 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

DESCRIPTION

AliEn Tutorial. MODEL 2009 4 th May , 2009. Outline. Installation of the AliEn software AliEn and the GRID Authentication File Catalogue Submitting jobs. Installation. Please, do: w get http://alien.cern.ch/alien-torrent-installer c hmod +x alien-torrent-installer - PowerPoint PPT Presentation

TRANSCRIPT

AliEn Tutorial

MODEL 20094th May, 2009

2 Pablo.Saiz@cern.ch4 May 2009

Installation of the AliEn softwareAliEn and the GRIDAuthenticationFile CatalogueSubmitting jobs

Outline

3 Pablo.Saiz@cern.ch4 May 2009

Installation

Please, do:wget http://alien.cern.ch/alien-torrent-installerchmod +x alien-torrent-installer./alien-torrent-installer

Default installationFirst two commands fastLet’s leave the 3rd running...

4 Pablo.Saiz@cern.ch4 May 2009

AliEn, AliEn2

ALICE ENvironment on the GRIDGRID middleware

Developed as single entry point to the GRID for ALICE (CERN)All the components necessary to build a GRID and interact with other GRIDs

File System with metadataAuthorization, authentication, job optimization and execution, storage managementAudit, quotas, monitoringInterfaces to various GRID implementations

Used since:2002 for centrally managed productions2006 for user analysis

Used by several Virtual Organizations:ALICE, PANDA, CBM, GPCALMA

http://alien.cern.ch

5 Pablo.Saiz@cern.ch4 May 2009Pablo Saiz 5

CERN

6 Pablo.Saiz@cern.ch4 May 2009Pablo Saiz 6

LHC

27km particle accelerator100 meters underground

About to start collecting data

7 Pablo.Saiz@cern.ch4 May 2009Pablo Saiz 7

A Large Ion Collider Experiment

8 Pablo.Saiz@cern.ch4 May 2009

Why do we need a catalogue?

Loads of data (several PB per year!!)Distributed community

Same structure from everywhereContains location of the file

It does not contain the filePossibility of several copies of files

Select the closest oneMetadata information

We will see this later on...

9 Pablo.Saiz@cern.ch4 May 2009

AliEn2 File Catalogue

Mapping from LFN to SE and PFNUNIX-like file systemMetadataGUIDUsed by

all other services

Tier1

|--./| |--cern.ch/| | |--user/| | | |--a/| | | | |--admin/| | | | || | | | |--aliprod/| | | || | | |--f/| | | | |--fca/| | | || | | |--p/| | | | |--psaiz/| | | | | |--as/| | | | | || | | | | |--dos/| | | | | || | | | | |--local/

|--simulation/| |--2001-01/| | |--V3.05/| | | |--Config.C| | | |--grun.C

| |--36/| | |--stderr| | |--stdin| | |--stdout| || |--37/| | |--stderr| | |--stdin| | |--stdout| || |--38/| | |--stderr| | |--stdin| | |--stdout

| | | || | | |--b/| | | | |--barbera/ALICE

USERS

ALICE DATA

ALICE REMOTE

10 Pablo.Saiz@cern.ch4 May 2009

LFN and GUID catalogues

/

/alice

/alice/user/p/psaiz

/alice/simulation/2006

Index1-JAN-1970

1-JAN-2006

14-FEB-2007

23-AUG-2008

Index GUID PFNLFN Catalogue GUID CatalogueAliEn File & Metadata Catalogue

LFN GUID

LFN: Logical File Name (what the user sees)GUID: Global Unique Identifier (another way of identifying files)PFN: Physical File Name

11 Pablo.Saiz@cern.ch4 May 2009

Who uses the catalogue?

EverybodyIF A FILE IS NOT IN THE CATALOGUE,

IT DOES NOT EXIST ON THE GRIDUsersUser data, user executables ...Job executionExecutable, input, outputPackagesSource code, binary code, post

installation

12 Pablo.Saiz@cern.ch4 May 2009

Back to the exercises...

Did the installation finish?Let’s get ready to access the GRIDX509 Certificate authentication

Public vs Private keyCertificate AuthorityProxy certificates

13 Pablo.Saiz@cern.ch4 May 2009

Preparing certificates

In $HOME/.globus:userkey.pemusercert.pem

Readable only by the user (chmod 400)Make alien use those files:

mkdir $HOME/.alienln –s $HOME/.globus $HOME/.alien/globus

14 Pablo.Saiz@cern.ch4 May 2009

Environment files$ALIEN_ROOT/.Environment$HOME/.alien/Environment

Bash files sourced before starting aliencat <<EOF > $HOME/.alien/Environment

export ALIEN_ORGANISATION=PANDA export ALIEN_USER=<username> EOF

Put alien in the PATHmkdir –p $HOME/binln –s $HOME/alien/bin/alien $HOME/bin/

15 Pablo.Saiz@cern.ch4 May 2009

alien proxy-init, alien proxy-infoGetting a GRID proxy

*****

16 Pablo.Saiz@cern.ch4 May 2009

Globus related:Permissions on $HOME/.globus/userkey.pem are not private to the user – chmod 400 userkey.pemYour certificate authority is exotic and not known to the server.Your certificate has expired.Clock skew:

• Your local computer time is in the future with respect to the server's time.

• Your local computer time is more in the past than the certificate life time.

Authentication problems

17 Pablo.Saiz@cern.ch4 May 2009

Check the certificates in your machines:Store them under e.g: $HOME/.globus/Check that the key file is read only

Setup the environment$HOME/.alien/EnvironmentPut alien in the path

Get a valid alien tokenalien proxy-init

Try it out !!

18 Pablo.Saiz@cern.ch4 May 2009

Standard bash shell with grid comandsMain bash features are available.Not all shell helper programs are available.Some local commands (like ls, cat etc) are overwritten with the corresponding GRID commands.File/path tab completion in the virtual GRID directory.

Accessing the shell

19 Pablo.Saiz@cern.ch4 May 2009

Tab completion working!!!

Basic commands I

20 Pablo.Saiz@cern.ch4 May 2009

Basic commands II

21 Pablo.Saiz@cern.ch4 May 2009

whereis command

Gives (list of) SE and pfns for a file

22 Pablo.Saiz@cern.ch4 May 2009

Viewing the files

23 Pablo.Saiz@cern.ch4 May 2009

Shell – Copying files (add/get)

24 Pablo.Saiz@cern.ch4 May 2009

Access the alien shell.alien

Check your user name:whoami

List the contents of your home directoryls

Print the working directorypwd

Create the following directory structure in your AliEn space:$HOME/bin (if it is not there)$HOME/macros and $HOME/jdl

Get the information of the file: whereis /bin/date

Shell – Try it out I

25 Pablo.Saiz@cern.ch4 May 2009

See the file /bin/datecat /bin/date

Copy the file to your local machinecp /bin/date /tmp/my_file

Exit the shell and check the file

Shell – Try it out II

26 Pablo.Saiz@cern.ch4 May 2009

Job & Transfer models

JobsTaskQueue Job

Broker

Job Manager

CE

JA

Central servicesSite services

SplittingExpired

PrioritiesMerging

Zombies

Job optimizers

File catalogue

LFN GUIDMetadataFTD

PackmanMonALISA

JA

JA

CE

SEPackman

MonALISA

FTD

SE

Site ASite B

Transfer Manager TransferQueue Transfer

Broker

Transfer optimizers

fts

bbftp

xrdcp

27 Pablo.Saiz@cern.ch4 May 2009

ALICE sitesEurope

Asia

North America

Africa

See Costin’s presentation

28 Pablo.Saiz@cern.ch4 May 2009

Running jobs

More than 25 CPU/years per day

29 Pablo.Saiz@cern.ch4 May 2009

AliEn

AliEn user interface

VTD EDGLCG /GLITECONDORARC/

NORDUGRID

Nice! I STILL do not have to

worry about ever changing GRID environment…

Interconnecting GRIDs

30 Pablo.Saiz@cern.ch4 May 2009

Job status

31 Pablo.Saiz@cern.ch4 May 2009

Your job is submitted and a <JOBID>.is assigned to it.

Submitting jobs

32 Pablo.Saiz@cern.ch4 May 2009

Checking the job status

33 Pablo.Saiz@cern.ch4 May 2009

Getting the output

34 Pablo.Saiz@cern.ch4 May 2009

Other useful commands

jobListMatchList of CE that can execute the job

SpyOutput of the job while it is running

MasterjobGive information about

35 Pablo.Saiz@cern.ch4 May 2009

Connect to alienalien login

Submit the jobsubmit <<EOF

Executable=“date”;EOF

Follow the executiontop –id, ps trace, ps jdl

Get the outputcp /proc/<user>/<id>/job-output/stdout /tmp/out

Try it out!!

36 Pablo.Saiz@cern.ch4 May 2009

Executable: Compulsory field where we give the lfn of the executable that should be stored in /bin or $V0/bin or $HOME/bin.Arguments: They will be passed to the executable.Packages: Type packages in the shell to see what kind of packages are installed.InputFile: The files that will be transported to the node where the job will run.InputData: It will require that the job will be executed in a site close to the files specified here.InputDataList: The filename in which the Job Agent will write the InputData list.InputDataListFormat: The format of the InputData list.

JDL fields I

37 Pablo.Saiz@cern.ch4 May 2009

OutputFile: The files that will be registered in the catalog once the job finishes.OutputArchive: What files will be archived in a zip file.Validationcommand: Specifies the script to be used as a validation script.Email: Receive a mail when the job finishes.TTL: The maximum run time of your job.Split: Split the jobs in several sub jobs.

JDL fields II

38 Pablo.Saiz@cern.ch4 May 2009

Packages

packmangives a list of the packages defined/installedinstalls packages

39 Pablo.Saiz@cern.ch4 May 2009

Add:Packages=“pbarprod@pandaroot::nov08“;

Try other commands:Executable=“pandaroot”;

Multiple submissionSplit=“production:1-5”;

Create your own executables

Try it out !!

40 Pablo.Saiz@cern.ch4 May 2009

Conclusion

AliEnInterface to the GRIDDeveloped by ALICEAlso used by PANDA, CBM, GPCALMAIn production since 2002

File CatalogueFile and Metadata

Job submission modelData movement and replication

41 Pablo.Saiz@cern.ch4 May 2009

Registration – Certificates:http://alien.cern.ch/twiki/bin/view/Alice/UserRegistration

https://ca.cern.ch/ca/

AliEn: http://alien.cern.ch

Gshell: http://alien.cern.ch/twiki/bin/view/AliEn/GAPI

User's guide: http://project-arda-dev.web.cern.ch/project-arda-dev/alice/apiservice/AA-UserGuide-0.0m.pdf

References I

42 Pablo.Saiz@cern.ch4 May 2009

aliensh Grid Command Online Reference V1.0 http://project-arda-dev.web.cern.ch/project-arda-dev/alice/apiservice/guide/guide-1.0.htm

Previous tutorials:http://aliceinfo.cern.ch/Offline/Analysis/Tutorial/

Event Tag System:http://pcaliweb02.cern.ch/Offline/Analysis/RunEventTagSystem/EventTags.html#Event%20tag%20system

https://edms.cern.ch/document/788315/1 (INTERNAL NOTE)

References II

43 Pablo.Saiz@cern.ch4 May 2009

http://alien.cern.ch

Any questions?

Thank you very much!!

top related