Home

ARC Clients User Manual

1. a 46 4 5 Deprecated configuration 46 Chapter 1 Introduction The command line user interface of ARC consists of a set of commands necessary for job submission and manipulation and data management This manual replaces the older version in NORDUGRID MANUAL 1 and is valid for ARC release 11 05 client versions 1 0 0 and above Command line tools semantics are the same as in earlier 0 x versions of ARC roughly following that of basic Linux commands and most common batch system commands One obvious difference is change of the legacy prefix from ng to the more appropriate arc This is not only a cosmetic change behaviour of the commands also have changed as did their functionalities and options Users are strongly discouraged from modifying their old scripts by simply replacing ng with arc results may be unpredictable CHAPTER 1 INTRODUCTION Chapter 2 Commands 2 1 Proxy utilities ARC now comes complete with a set of utilities to create temporary user credentials proxies used to access Grid services 2 1 1 arcproxy In order to contact Grid services submit jobs copy data check information etc one has to present valid credentials These are commonly formalized as so called proxy certificates There are many different types of proxy certificates with different Grids and different services having own preferences arcproxy is
2. joblist filename the file storing information about active jobs default arc jobs dat i jobids from file filename file containing a list of job IDs status statusstr only select jobs whose status is statusstr P listplugins list the available plugins t timeout time timeout for queries default 20 sec d debug verbosity verbosity level is one of FATAL ERROR WARN ING INFO VERBOSE or DEBUG z conffile filename configuration file on Linux default HOME arc client conf v version print version information h 1 print help page Options in ARC 11 05 cluster name explicitly select or reject a specific site cluster Options in ARC 12 05 cluster designator select one or more computing elements by an alias for a single CE a group of CEs or a URL rejectmanagement URL skip jobs which are on a computing element with a given URL Arguments job list of job IDs and or job names 26 CHAPTER 2 COMMANDS Command line options have the same meaning as the corresponding ones of arcstat and others When several of the c lt cluster gt i lt jobidfile gt and job command line options are spec ified the command resumes all jobs listed on the command line plus all jobs on the specified clusters plus all jobs from the specified jobidfile However the c lt cluster gt for ARC 11 05 r lt URL gt for ARC 12 05 and s lt status gt options will filter the
3. nordugrid arc plugins globus package to be installed ARC1 Web service ARC execution service derived from OGSA BES standard CREAM CREAM BES compliant execution service BES Generic BES plugin consistent with the OGSA BES standard EMIES Web service following EMI Execution Service specifications Here are examples of full designators for ARC legacy index services ARCO ldap ce ng eu 2135 nordugrid cluster name ce ng eu Mds Vo name local o grid ARCO ldap index ng org 2135 mds vo name sweden 0 grid and for CREAM CREAM ldap cream glite org 2170 o grid In case interface part is missing every communication protocol interface corresponding to supported flavours and matching URL will be tried Because arcsub supports multiple Grid flavours and this number is continuously increasing it is strongly advisable not to skip interface part Example of such designator is ldap ce ng eu 2135 nordugrid cluster name ce ng eu Mds Vo name local o grid For convenience it is possible to shorten designator even more by skipping protocol and path parts of URL So designator may be as simple as hostname of service to be contacted Here is an example of such shorthand designator for index server index ng org mds vo name sweden and those suitable both for c and g options cream glite org ce ng eu If such short designators are used then rest of the URL is automatically generated according to the flavour which is currently trie
4. 7 NORDUGRID NORDUGRID MANUAL 13 27 3 2015 ARC CLIENTS User Manual for ARC 11 05 client versions 1 0 0 and above Contents 1 Introduction 2 1 Proxy utilities SACP ROK 2 2 Job submission and 221 GS we oe Ral A ce ee ee be castes erie Stes Gaia Dh ees Goo phe 2 2 9 sia ah ee ewe PA oe ee Berea we Go ee 4 ssa ish Pacis Wns et ok Ga a GA WS Shoe ge dh th So Se ee ee ee ee ee ye ee ee ee ee 226 Ae SB hee eae Be Oe Roe By eR Boe 2 2 oe ee a hee ee Rae eM ee hes 2218 Rey So ee ee Re ee eB 2 2 9 _arcrenew 2 2 11 arcresub 2 3 231 T 2 3 3 _ _arcrm 2 3 4 arcmkdir 2 4 Test suite 4 1 _arctest URLs 4 ARC Client Configuration 4 1 Block common Tejectservices 11 11 16 17 18 19 20 21 22 24 25 26 28 28 29 30 31 31 32 32 35 CONTENTS 2 eae ed to ee Beare bg 41 submissioninterface ooa a e e 41 eat a fee ae ae Pe Ae eee Ge gee ae a 41 vs ann a ass eee alae Siw BA OOS Bae Soe leds
5. httpgetpartial yes no for transfers to file destinations specifies whether the file should be read only unmodifiable or not default is yes indicates whether the GridF TP data channel should be encrypted default is no specifies size of chunks blocks buffers used in GridFTP or HTTP s g transactions default is protocol dependent specifies the algorithm for checksum to be computed for transfer verification or provided to the indexing server This is overridden by any metadata options specified see below If this option is not provided the default for the protocol is used checksum no disables checksum calculation make software try to overwrite existing file s before writing to destination tools will try to remove any information content associated with specified URL to distinguish between different kinds of https httpg and srm protocols Here gssapi stands for httpg implemention using only GSSAPI functions to wrap data and gsi uses additional headers as implmented in Globus IO The ss1 and 15 stand for usual https and especially usable only if used with srm protocol The ss13 is mostly same as ssl but uses SSLv3 hadshake while establishing https connection The default is gssapi for srm connections tls for https and gssapi for httpg In case of srm if default fails gsi is then tried specify the space token to be used for uploads to SRM storage elements supporting SRM version 2 2 or higher specify if before
6. lt jobid gt but only if it is finished Or arcstat i jobidfile r mycluster example com will return information about jobs which are in the jobidfile but not on mycluster example com If the 1 option is given extended information is printed Jobs can be sorted according to the job ID submission time or job name either in normal or reverse order By using the sort or rsort option followed by the desired ordering jobid submissiontime or jobname respectively jobs will be sorted in normal or reverse order Note that the options sort and rsort cannot be used at the same time Options s and j do not use job ID or names By specifying the a option the status of all active jobs will be shown If the option is used the list of jobs is read from a file with the specified filename instead of the default one HOME arc jobs dat on Linux Option c accepts arguments as explained in the description of arcsub that is in the GRID URL notation for ARC 11 05 or URLs aliases and groups from the configuration file in ARC 12 05 Different sites may report different job states depending on the installed grid middleware version Typical values can be e g Accepted Preparing Running Finished or Deleted Please refer to the respective middleware documentation for job state model description Command line option s will instruct the client to display information of only those jobs which
7. PUTing onto MyProxy server This option is specific for the GET command when contacting a MyProxy server R retrievable_by_cert string Allow specified entity to retrieve credential without passphrase this option is specific for the PUT com mand when contacting MyProxy server L URL URL of MyProxy server optionally followed by colon and port number e g example org 7512 if the port number is not specified 7512 is used by default myproxycmd command to server command can be PUT or GET PUT put put a delegated credential to MyProxy server 2 1 PROXY UTILITIES 9 GET get get a delegated credential from MyProxy server credential certificate and key is not needed in this case F nssdb use NSS credential DB in default Mozilla profiles including Firefox Seamonkey and Thunderbird constraint string proxy constraints see options below t timeout seconds timeout for network communication in seconds de fault 20 seconds d debug verbosity verbosity level is one of FATAL ERROR WARN ING INFO VERBOSE or DEBUG z conffile filename configuration file on Linux default HOME arc client conf v version print version information h help print help page MyProxy functionality can be used together with functionality a credential stored in a MyProxy server can receive a VOMS AC If destination location of
8. a powerful tool that can be used to generate most commonly used proxies It supports the following types e pre RFC GSI proxy e RFC compliant proxy default e VOMS extended proxy e MyProxy delegation arcproxy requires presence of user s private key and public certificate as well as the public certificate of their issuer CA These certificates can exist either as separate files or in an NSS certificate store such as that used by Firefox That is if your certificate is in your Firefox browser there is no need to export it into files as arcproxy can be used to create a proxy directly Options P path path to the generated proxy file defaults are de scribed below C path path to the certificate file defaults are described be low path to the key file defaults are described below T cadir path path to the trusted certificate directory only needed for VOMS client functionality defaults are described below CHAPTER 2 COMMANDS s vomsdir path path to the top directory of VOMS lsc files only needed for the VOMS client functionality V vomses path path to the VOMS server configuration file if the path is a directory rather than a file all the files under this directory will be searched S voms voms command Specify VOMS server more than VOMS server can be specified like this VOa command1 voms VOb command2 command is optional an
9. common and alias ARC 12 05 and above still makes use of the common block but does not recognise defaultservices and rejectservices options instead using rejectdiscovery rejectmanagement infointerface and 39 40 CHAPTER 4 ARC CLIENT CONFIGURATION submissioninterface options In addition each service in ARC 12 05 and above has its own block registry lt alias gt for registry services and computing lt alias gt for computing services where the lt alias gt has to be a unique name for the service without spaces In ARC versions 0 all configured services were used by default but from version 11 05 default services must be explicitly specified in the configuration either by the defaultservices 11 05 or default attributes 12 05 and above 4 1 Block common This attribute is multi valued This attribute is used to specify default services to be used Defining such in the user configuration file will override the default services set in the system configuration The value of this attribute should follow the format service_type grid service_url where service is type of service e g computing or index grid specifies type of middleware plugin to use when contacting the service e g ARCO ARC1 CREAM etc and service_url is the URL used to contact the service Several services can be listed separated with a blank space no line breaks allowed Example defaultservices index ARCO ldap index1 n
10. e g 43200 or 12h or 12H if not specified the default is the least value of 12 hours and validityPeriod which is the life time of the delegated proxy on a MyProxy server proxyPolicy policy content assigns specified string to proxy prolicy to limit its functionality proxyPolicyFile policy file e keybits number length of the key to generate Default is 1024 bits Special value inherit means using the same key length as the signing certificate e signingAlgorithm name signing algorithm to use for signing the public key of the proxy Possible values are shal sha2 alias for sha256 sha224 sha256 sha384 sha512 and inherit use algorithm of the signing certificate Default is inherit Proxy information items requested with the i option are printed in the requested order and are separated by newline If an item has multiple values they are printed in same line separated by a pipe sign Supported information item names are subject subject name of the proxy certificate identity identity subject name of the proxy certificate issuer issuer subject name of the proxy certificate ca subject name of the CA which issued the initial certificate path file system path to the file containing proxy type type of the proxy certificate e validityStart timestamp when proxy validity starts validityEnd timestamp when proxy validity ends validityPeriod duration of pr
11. is otherwise unreachable and yet your local job list file still has it listed Use s option with value Undefined to remove such stale job information from the local list Note that specifying a and f options together also removes such stale job information while also removing finished and cancelled jobs Other command line options have the same meaning as the corresponding ones of arcstat and others When several of the lt cluster gt i lt jobidfile gt and job command line options are spec ified the command cleans all jobs listed on the command line plus all jobs on the specified clusters plus all jobs from the specified jobidfile However the c lt cluster gt for ARC 11 05 r lt URL gt for ARC 12 05 and s lt status gt options will filter the jobs selected by the above mentioned options or if none of those are specified then these will filter all the jobs For example arcclean s FAILED c mycluster lt jobid gt will clean the failed jobs on mycluster plus lt jobid gt but only if it is failed Or arcclean i jobidfile r mycluster example comn will 24 CHAPTER 2 COMMANDS clean all jobs which are in the jobidfile but not on mycluster example com 2 2 9 arcrenew Quite often the user proxy expires while the job is still running or waiting in a queue In case such job has to upload output files to a Grid location a storage element it will fail By using the arcrenew command users can upload a
12. of alias value pairs Alias substitutions is performed in connection with the c command line switch of the ARC clients Aliases can refer to a list of services separated by a blank space Alias definitions can be recursive Any alias defined in a list that is read before a given list can be used in alias definitions in that list An alias defined in a list can also be used in alias definitions later in the same list Examples alias arcO comput ing ARCO 1dap ce ng org 2135 nordugrid cluster name ce ng org Mds Vo name local o grid arci computing ARC1 https arex ng org 60000 arex cream computing CREAM ldap cream glite org 2170 o grid crossbrokering arcO 1 cream 4 5 Deprecated configuration files ARC configuration file in releases 0 6 and 0 8 has the same name and the same format Only one attribute is preserved timeout other attributes unknown to newer ARC versions are ignored In ARC lt 0 5 48 configuration on Linux platforms was done via files HOME ngrc HOME nggiislist and HOME ngalias The main configuration file could contain user s default settings for the verbosity level the information system query timeout and the download directory used by ngget A sample file could be the following Sample ngrc file Comments starts with NGDEBUG 1 60 NGDOWNLOAD tmp If the environment variables NGDEBUG NGTIMEOUT NGDOWNLOAD were defined these
13. plugins protocols supported d debug verbosity verbosity level FATAL ERROR WARNING INFO VERBOSE or DEBUG default WARNING tonly available in version 12 05 and later Sonly available in version 13 02 and later 32 CHAPTER 2 COMMANDS z conffile filename configuration file on Linux default HOME arc client conf v version print version information h help print help page Arguments old URL current name of file or directory new URL new name for file or directory The arcrename command renames files or directories on grid storage elements and indexing services The path component of old URL and new URL must differ and it must be the only component of both URLs which is different arcrename will exit with an error if the paths are equivalent or other components of the URLs are different Renaming a URL to an existing URL will either fail or overwrite the existing URL depending on the protocol 2 4 Test suite 2 4 1 arctest arctest is a simple utility that tests very basic functionalities of the middleware It is convenient for e first time users who do not know job description languages and yet want to test e g their credentials or client setup e system administrators who d like to quickly test their installations without having to learn job description languages The arctest utility contains pre defined test jobs which can be submitted either to a specific test site or to a regular Grid infrastructu
14. took precedence over the values defined in this configuration Any command line options override the defaults 4 5 DEPRECATED CONFIGURATION FILES 47 The file HOME nggiislist was used to keep the list of default GIIS server URLs one line per see giis attribute description above The file HOME ngalias was used to keep list of site aliases one line per alias see alias attribute description above Acknowledgements This work was supported in parts by the Nordunet 2 program the Nordic DataGrid Facility the EU KnowARC project Contract nr 032691 the EU EMI project Grant agreement nr 261611 and the Swedish Research council via the eSSENCE strategic research program 48 CHAPTER 4 ARC CLIENT CONFIGURATION Bibliography 1 A Anjomshoaa et al Job Submission Description Language JSDL Specification Version 1 0 first errata update GFD R 136 July 2008 URL http www gridforum org documents GFD 136 pdf 2 M Ellert Mohn I Marton and R czei libarcclient A Client Library for ARC The NorduGrid Collaboration URL http www nordugrid org documents client_technical pdf NORDUGRID TECH 20 w I Foster and C Kesselman Globus A Metacomputing Infrastructure Toolkit International Journal of Supercomputer Applications 11 2 115 128 1997 Available at http www globus org 4 A Konstantinov The ARC Computational Job Management Module A REX The NorduGrid Collaboration URL ht
15. wee ae Al e hee eae 41 a te dg a 42 42 die Ata ee re nk Be Ge eee 42 Ibartender geiko BA ee 4 EA Mee eee A 42 ee BASEL ewe EER EEA Ee ee 42 2 ibd Sesh dad dhe de ob eb eee Bel ee he Bo 43 tages Gea BS See eg eee poe 43 Wag Be Sok a eG w AOE die ale 43 a Abita et do he Ow eh PON Soe We OE he ea 43 43 Godt aA we PH a ee 43 4 2 Service Blocks ao Ge Re a a bee 44 eran cee pega are ewe ene ee ics Sec eee ce ae 44 eee A Re Dba kee ab ee ew ee Se ee te oe 44 fea Apdo abe ae amp we Bla ee en 44 2 guise ao SS eae ha eRe eee AGO a Eee ER eS 45 1 45 E 45 51115 hie be ee YR ee eh a 46 44 Block alias 5 ork sou shaw ale emits we Goo Be Ee
16. when destination is an index ing service Must be specified for indexing services which do not automatically generate physical loca tions Can be specified multiple times locations will be tried in order until one succeeds 3 thirdparty perform third party transfer where the destination pulls from the source only available with GFAL plu gin and ARC 13 02 t timeout seconds timeout for network communication in seconds de fault 20 P plugins list the available plugins protocols supported d debug verbosity verbosity level FATAL ERROR WARNING INFO VERBOSE or DEBUG default WARNING 30 CHAPTER 2 COMMANDS z conffile filename configuration file on Linux default HOME arc client conf v version print version information h 1 print help page Arguments source source URL destination destination URL This command transfers contents of a file between 2 end points End points are represented by URLs or meta URLs or local file paths For supported endpoints please refer to Section 3 arccp can perform multi stream transfers if threads URL option is specified and server supports it Source can end with In that case the set of files under source will be copied into destination and destination must also end with Destination will be created if it does not exist If copying deeper than one level is required then r or D must be used If destination alone ends with it is extende
17. wish to re submit each job to the same site use option If the original job was successfully killed its traces will be removed from the execution site unless the k option is specified When several of the lt cluster gt i lt jobidfile gt and job command line options are speci fied the command resubmits all jobs listed on the command line plus all jobs on the specified clusters plus all jobs from the specified jobidfile However the c lt cluster gt for ARC 11 05 r lt URL gt for ARC 12 05 and s lt status gt options will filter the jobs selected by the above mentioned options or if none of those are specified then these will filter all the jobs For example arcresub s FAILED c mycluster lt jobid gt will resubmit the 28 CHAPTER 2 COMMANDS running jobs on mycluster plus lt jobid gt but only if it is failed Or arcresub jobidfile r mycluster example org will resubmit all jobs which are in the jobidfile but not on mycluster example org 2 3 Data manipulation ARC provides basic data management tools to copy create list rename and remove files and directories to from and between Grid storage elements and index services 2 3 1 arcls arcls is a simple utility that allows to list contents and view some attributes of objects of a specified by a URL remote directory Options 1 long detailed listing L locations detailed listing including URLs from which files can
18. with a new name This job thus demonstrates usage of inputfiles and outputfiles attributes of job description languages arctest is complementary to the arcinfo utility which extracts information about Grid resources with out submitting test jobs and to arcproxy which provides more detailed information about user credentials 34 CHAPTER 2 COMMANDS Chapter 3 URLs File locations in ARC can be specified both as local file names and as Internet standard Uniform Resource Locators URL There are also some additional URL options that can be used Depending on the installed ARC components some or all of the following transfer protocols and metadata services are supported ftp ordinary File Transfer Protocol FTP gsiftp GridFTP the Globus enhanced FTP protocol with security encryption etc developed by The Globus Alliance 3 http ordinary Hyper Text Transfer Protocol HTTP with PUT and GET methods using multiple streams https with SSL v3 httpg HTTP with Globus GSI ldap ordinary Lightweight Data Access Protocol LDAP srm Storage Resource Manager SRM service 6 root Xrootd protocol available in ARC 2 0 0 and later read only 4 2 0 and later full function ality rucio Next generation ATLAS data management system read only available in ARC 4 1 0 and later acix ARC Cache Index read only available in ARC 4 1 0 and later file local to the host file name with a full path An URL can be us
19. writing to specified location software should try to create all directories mentioned in specified URL Currently this applies to FTP and GridFTP only Default for those protocols is yes controls the use of the TCP_NODELAY socket option which dis ables the Nagle algorithm Applies to http s only Default is no supported only in arcls and other arc tools specify transfer protocols for meta URLs such as SRM Multiple protocols can be specified as a comma separated list in order of preference specify the Rucio account to use when authenticating with Rucio while storing file on HTTP S server software will try to send it in chunks parts If server reports error for partial PUT command software will fall back to tranfering file in single piece This be havior is non standard and not all servers report error properly Hence default is safer while retrieving file from HTTP S server software will try to read in in chunks parts If server does not support partial GET com mand it usually ignores request for partial transfer range and file is transfered in one piece Default is yes 37 Local files are referred to by specifying either a location relative to the job submission working directory or by an absolute path the one that starts with preceded with a file prefix URLs also support metadata options which can be used for registering additional metadata attributes or querying the service using metadata
20. CE BLOCKS 45 Example group favs group fast The interface of the service through which the computing element information should be retrieved Ex ample infointerface org nordugrid 1ldapng Possible information interfaces are e org nordugrid ldapng org nordugrid ldapgluel e org nordugrid ldapglue2 e org nordugrid wsrfglue2 org ogf glue emies resourceinfo The interface of the service with which the jobs should be submitted Example submissioninterface org nordugrid gridftpjob Possible submission interfaces are e org nordugrid gridftpjob e org ogf bes e org ogf glue emies activitycreation The interface of the service with which the registry can be queried Example registryinterface org nordugrid ldapegiis Possible registry interfaces are org nordugrid ldapegiis org nordugrid emir subject to change 46 CHAPTER 4 ARC CLIENT CONFIGURATION 4 3 srms conf If any data management commands are used with the Storage Resource Management SRM 6 protocol the file or its analogue on non Linux platforms may be created to store cached information on these services For more information see the description inside this file 4 4 Block alias This block is deprecated in ARC 12 05 Users often prefer to submit jobs to a specific site since contact URLs and especially end point references are very long it is very convenient to replace them with aliases Block alias simply contains a list
21. Definition xmlns http schemas ggf org jsd1 2005 11 jsdl xmlns posix http schemas ggf org jsd1 2005 11 jsdl posix gt 2 2 JOB SUBMISSION AND MANAGEMENT 13 lt JobDescription gt lt JobIdentification gt lt JobName gt Hello World job lt JobName gt lt JobIdentification gt lt Application gt lt posix POSIXApplication gt lt posix Executable gt bin echo lt posix Executable gt lt posix Argument gt Hello World lt posix Argument gt lt posix Output gt out txt lt posix Output gt lt posix Error gt err txt lt posix Error gt lt posix POSIXApplication gt lt Application gt lt JobDescription gt lt JobDefinition gt The job ID uniquely identifies the job Job IDs differ strongly for different computing service flavours but basically they have a form of a URL You should use job ID as a handle to refer to the job when doing other job manipulations such as querying job status arcstat killing it 11 re submitting arcresub or retrieving the result arcget Usually job ID is a valid URL for the job session directory You can almost always use it to access the files related to the job by using data management tools see Chapter 2 3 There may be exceptions for some computing service flavours like CREAM which do not support listing job session directory The option can be used to manually select known computing sites for example arcsub c alias c group c URL job xrsl Th
22. L r rejectmanagement URL skip jobs which are on a computing element with a given URL Arguments job list of job IDs and or job names If a job is killed its traces are being cleaned from the Grid If you wish to keep the killed job in the system e g for a post mortem analysis use the option Job cancellation is an asynchronous process such that it may take a few minutes before the job is actually cancelled Command line options have the same meaning as the corresponding ones of arcstat and others When several of the lt cluster gt i lt jobidfile gt and job command line options are spec ified the command kills all jobs listed on the command line plus all jobs on the specified clusters plus all jobs from the specified jobidfile However the c lt cluster gt for ARC 11 05 r lt URL gt for ARC 12 05 and s lt status gt options will filter the jobs selected by the above mentioned options or if none of those are specified then these will filter all the jobs For example arckill s INLRMS R c mycluster lt jobid gt will kill the running jobs on mycluster plus lt jobid gt but only if it is running Or arckill i jobidfile r mycluster example com will kill all jobs which are in the jobidfile but not on mycluster example com 2 2 8 arcclean If a job fails or gets killed with k option or when you are not willing to retrieve the results for some reasons a good practice for users is no
23. a eae Pee eee ee 7 arckill oa eae wa arcrenew oe defaultservices arcresub arcproxy deprecated files arcresume arcrename See Bice ALCL eens a ee arcrenew infointerface oats ate A ee arcresub jobdownloaddirectory APCS arcresume joblist ov agi eae ee aoe ae ahs joblisttype 50 keypath proxypath registryinterface rejectdiscovery rejectmanagement rejectservices srms conf submissioninterface timeout tte hates verbosity vomsserverpath D data management JOO ED das job management Index 5 SCCUTILY submit job gon options URLs
24. arcrm srm grid uio no grid atlas A0D_0947 pool root 2 3 4 arcmkdir The arcmkdiz command allows users to create directories if the protocol of the specified URL supports it Options make parent directories as needed t timeout seconds timeout for network communication in seconds de fault 20 P plugins list the available plugins protocols supported d debug verbosity verbosity level FATAL ERROR WARNING INFO VERBOSE or DEBUG default WARNING z conffile filename configuration file on Linux default HOME arc client conf v version print version information h help print help page Arguments URL directory to create arcmkdir creates directories on grid storage elements and indexing services If the parent directory does not exist and p is not specified then arcmkdir will probably fail but it depends on the protocol The permissions on the new directory are the default of the server or if the protocol requires them to be specified then the directory is only readable writable searchable by the user the equivalent of 700 on a file system Example arcmkdir srm grid uio no grid atlas newdir 2 3 5 arcrename The command allows users to rename files and directories if the protocol of the specified URL supports it Options t timeout seconds timeout for network communication in seconds de fault 20 P plugins list the available
25. ation information the global configuration files if exist or default values are used instead Some client tools may be able to create the default HOME arc client conf on Linux if it does not exist The ARC configuration file consists of several configuration blocks Each configuration block is identified by a keyword and contains configuration options for a specific part of the ARC middleware The configuration file is written in a plain text format known as INI Configuration blocks start with identifying keywords inside square brackets Typically first comes a common block common Thereafter follows one or more attribute value pairs written one on each line in the following format common attributei valuel attribute2 value2 attribute3 value3 value4 comment line i comment line 2 Note that values must not be enclosed in quotes Most attributes have counterpart command line options Command line options always overwrite configuration attributes Client configuration is different for ARC 11 05 client versions 1 and ARC 12 05 client versions 2 and above Newer clients will work with the old configuration Older clients will not understand most new configuration options In addition configuration files from ARC 0 versions will not work with newer client versions The ngclientZarc tool is provided to help with migrating configuration to the new format ARC 11 05 clients recognise two configuration blocks
26. attribute corresponds to the command line option of Example keypath home username key pem Specifies a non standard location of user s public certificate It is used by arcproxy or similar tools during proxy generation This attribute corresponds to the C command line option of arcproxy Example certificatepath home username cert pem Specifies non standard location of the directory containing CA certificates This attribute corresponds to the T command line option of arcproxy Example cacertificatesdirectory home user cacertificates Specifies an explicit path to the certificate of the CA that issued user s credentials Example cacertificatepath home user myCA 0 Specifies non standard path to the file which contians list of VOMS services and associated configuration parameters This attribute corresponds to the V command line option of arcproxy Example vomsserverpath etc voms vomses Sets directory which will be used as the default job download directory This attribute corresponds to the D command line option of arcget Example jobdownloaddirectory home my jobs 44 CHAPTER 4 ARC CLIENT CONFIGURATION 4 2 Service blocks These blocks are only available in ARC 12 05 and above Each service is configured through its own block Each service has to have unique alias name which is used to refer to this service The services can be grouped into multiple groups Then the name of the group ca
27. attributes These options are specified at the end of the LFN and consist of name and value pairs separated by colons The following attributes are supported 38 CHAPTER 3 URLS checksumtype Type of checksum Supported values are cksum default md5 and adler32 checksumvalue The checksum of the file The checksum attributes may also be used to validate files that were uploaded to remote storage Examples of URLs are http grid domain org dir script sh gsiftp grid domain org 2811 threads 10 secure yes dir input_12378 dat ldap grid domain org 389 lc collection1 rc Nordugrid dc nordugrid dc org file home auser griddir steer cra srm srm domain org griddir user file1 checksumtype adler32 checksumvalue 12345678 srm srm domain org transferprotocol https data file2 This is a destination URL The file will be copied to srm domain org at the path griddir user filet and the checksum will be compared to what is reported by the SRM service after the transfer This is a source or destination URL When getting a TURL from SRM the HTTPS transfer protocol will be requested Chapter 4 ARC Client Configuration The default behaviour of an ARC client can be configured by specifying alternative values for some parameters in the client configuration file The file is called client conf and is located in directory arc in user s home area e g on Linux If this file is not present or does not contain the relevant configur
28. be downloaded metadata display all available metadata r recursive operate recursively if possible D depth recursion_level operate recursively if possible up to specified level 0 no recursion available in ARC 13 02 nolist show only description of requested object do not list content of directories like 15 d f forcelist treat requested object as directory and always try to list content check readability of object Retrieving and showing information about object is supressed t timeout seconds timeout for network communication in seconds de fault 20 P plugins list the available plugins protocols supported d debug verbosity verbosity level FATAL ERROR WARNING INFO VERBOSE or DEBUG default WARNING z conffile filename configuration file on Linux default HOME arc client conf v version print version information h 1 print help page Arguments URL file or directory URL This tool is very convenient not only because it allows to list files at a Storage Element or records in an indexing service but also because it can give a quick overview of a job s working directory which is explicitly given by job ID 2 3 DATA MANIPULATION 29 Usage examples can be as follows arcls l gsiftp lscf nbi dk 2811 jobs 1323842831451666535 arcls srm grid uio no 8446 srm managerv2 SFN johndoe log2 Examples of URLs accepted by
29. conf X509_VOMSES or X509_VOMS_FILE arc vomses voms vomses ARC_LOCATION etc vomses for Windows environment ARC_LOCATION etc grid security vomses for Windows environment PWD vomses CO w etc vomses 10 etc grid security vomses arc client conf the overall ARC client configuration file see Section 4 Some options can be given default values by specifying them in such ARC client configuration file By using the conffile option a different configuration file can be used rather than the default 2 2 Job submission and management The following commands are used for job submission and management such as status check results retrieval cancellation re submission and such The jobs must be described using a job description language ARC supports the following languages JSDL I xRSL 7 and JDL 5 2 2 1 arcsub The arcsub command is the most essential one as it is used for submitting jobs to the Grid resources arcsub matches user s job description to the information collected from the Grid and the optimal site is being selected for job submission The job description is then being submitted to that site and usually is then forwarded to a local Resource Management System LRMS which can be e g PBS or Condor or SGE etc Options e jobdescrstring filename string describing the job to be submitted f jobdescrfile filename file describing the job to be submitted j joblist file
30. ct a specific site Options in ARC 12 05 cluster designator select one or more computing elements by an alias for a single CE a group of CEs or a URL rejectmanagement URL skip jobs which are on a computing element with a given URL Arguments job list of job IDs and or job names The arccat command returns the standard output of a job o option the standard error e option or errors reported by either Grid Manager or A REX 1 option Other command line options have the same meaning as in arcstat When several of the lt cluster gt 1 lt jobidfile gt and job command line options are spec ified the command prints logs of all jobs listed on the command line plus all jobs on the specified clusters plus all jobs from the specified jobidfile However the c lt cluster gt for ARC 11 05 r lt URL gt for ARC 12 05 and s lt status gt options will filter the jobs selected by the above mentioned options or if none of those are specified then these will filter all the jobs For example arccat s Finished c mycluster lt jobid gt will print logs of the finished jobs on mycluster plus of lt jobid gt but only if it is finished Or arccat 1 jobidfile mycluster example org will print logs of jobs which are in the jobidfile but not on mycluster example org 2 2 4 arcget To retrieve the results of a finished job the arcget command should be used It will transfer the files specified
31. d debug verbosity verbosity level is one of FATAL ERROR WARN ING INFO VERBOSE or DEBUG z conffile filename configuration file on Linux default HOME arc client conf version print version information h help print help page Options in ARC 11 05 g 1 url explicitly select or reject a specific index server cluster name explicitly select or reject a specific source site q qluster name explicitly select or reject a specific site submission target 2 2 JOB SUBMISSION AND MANAGEMENT 27 Options in ARC 12 05 g 1 designator select one or more registries by an alias for a single registry a group of registries or a URL c cluster designator select one or more computing elements by an alias for a single CE a group of CEs or a URL q qluster designator select one or more computing elements for the new jobs by an alias for a single CE a group of CEs or a URL rejectmanagement URL skip jobs which are on a computing element with a given URL R rejectdiscovery URL skip the service with the given URL during service discovery S submissioninterface InterfaceName use this interface for submit ting e g org nordugrid gridftpjob org ogf glue emies activitycreation org ogf bes Options in ARC 13 02 I infointerface InterfaceName the computing element specified by URL at the com mand line should be queried using this informat
32. d For example in the case of ARC1 https communication protocol is assumed If you are using some services frequently enough it is recommended to use aliases for these URLs Aliases are specified in the configuration file see Section In ARC 12 05 and higher it is advisable to configure each site in your client configuration see Section 4 2 and use its alias Specifying designators full URLs on the command line is possible but not recom mended Flavours as used in ARC 11 05 are not recognised in ARC 12 05 To list possible submission interfaces to use with S option use arcinfo command In order to keep track of submitted jobs ARC client stores information in a dedicated file on Linux platforms by default located in HOME arc jobs dat It is sometimes convenient to keep separate lists e g for different kinds of jobs to be used later with e g arcstat This is achieved with the help of the j command line option The ARC client transforms input job description into a format that can be understood by the Grid services to which it is being submitted By specifying the dumpdescription option such transformed description is written to the standard output instead of being submitted for execution Possible broker values for the arcsub command line option are Random ranks targets randomly default 2 2 JOB SUBMISSION AND MANAGEMENT 15 FastestQueue ranks targets according to their queue length B
33. d with the part of source after last thus allowing users to skip the destination file or directory name if it is meant to be identical to the source Usage examples of arccp are arccp i gsiftp lscf nbi dk 2811 jobs 1323842831451666535 job out job out arccp http www nordugrid org data somefile gsiftp hathi hep lu se data arccp gsiftp pgs02 grid upjs sk 2811 jobs 13331297786445657047863 output arccp R 3 my file srm srm host org spacetoken MYTOKEN my file 1 2 3 3 arcrm The arcrm command allows users to erase files and directories at any location specified by a valid URL Options force remove logical file name registration even if not all physical instances were removed t timeout seconds timeout for network communication in seconds de fault 20 P plugins list the available plugins protocols supported d debug verbosity verbosity level FATAL ERROR WARNING INFO VERBOSE or DEBUG default WARNING z conffile filename configuration file on Linux default HOME arc client conf v version print version information h help print help page Arguments URL URL file or directory URL mutliple URLs are supported in ARC 13 02 and above A convenient use for arcrm is to erase the files in a data indexing catalog as it will not only remove the physical instance but also will clean up the database record 2 3 DATA MANIPULATION 31 Here is an arcrm example
34. d is used to ask for specific attributes e g roles Command options are all put all of this DN s attributes into AC list list all of the DN s attribute will not create AC extension Role yourRole specify the role if this DN has such a role the role will be put into AC voname groupname Role yourRole specify the VO group and role if this DN has such a role the role will be put into AC o group role Specify ordering of attributes examples o knowarc eu coredev Developer knowarc eu testers Tester G gsicom use GSI communication protocol for contacting VOMS services H httpcom use HTTP communication protocol for contacting VOMS services that provide RESTful access Note that for the RESTful access 186 command and mul tiple VOMS servers are not supported 0 old use GSI proxy default is RFC 3820 compliant proxy I info print all information about this proxy in order to show the Identity DN without CN as suffix for proxy of the certificate the trusted certificates di rectory T cadir is needed r removes the proxy file U user string username for the MyProxy server N nopassphrase don t prompt for a credential passphrase when re trieving a credential from on MyProxy server The precondition of this choice is that the credential is PUT onto the MyProxy server without a passphrase by using R retrievable_by_cert option when being
35. e org 2 2 5 arcsync It is advised to start every grid session by running arcsync especially when changing workstations The reason is that your job submission history is cached on your machine and if you are using ARC client 20 CHAPTER 2 COMMANDS installations on different machines your local lists of submitted jobs will be different To synchronise these lists with the information in the Information System use the arcsync command Options j joblist filename the file storing information about active jobs default arc jobs dat f force don t ask for confirmation T truncate truncate the job list before synchronising P listplugins list the available plugins t timeout seconds timeout for network communication in seconds de fault 20 d debug verbosity verbosity level FATAL ERROR WARNING INFO VERBOSE or DEBUG default WARNING z conffile filename configuration file on Linux default HOME arc client conf v version print version information h 1 print help page Options in ARC 11 05 cluster name explicitly select or reject a specific site g index url explicitly select or reject a specific index server Options in ARC 12 05 cluster designator select one or more computing elements by an alias for a single CE a group of CEs or a URL g 1 designator select one or more registries by an alias for a single registry a group of registri
36. e to actually create the new proxy by running arcproxy Command line options have the same meaning as the corresponding ones of arcstat and others 2 2 JOB SUBMISSION AND MANAGEMENT 25 When several of the lt cluster gt 1 lt jobidfile gt and job command line options are speci fied the command renews proxies of all jobs listed on the command line plus all jobs on the specified clusters plus all jobs from the specified jobidfile However the lt cluster gt for ARC 11 05 r lt URL gt for ARC 12 05 and s lt status gt options will filter the jobs selected by the above mentioned options or if none of those are specified then these will filter all the jobs For example arcrenew s FAILED c mycluster lt jobid gt will renew proxies of the failed jobs on mycluster plus lt jobid gt but only if it is failed Or arcrenew 1 jobidfile r mycluster example com will renew proxies of all jobs which are in the jobidfile but not on mycluster example com 2 2 10 arcresume In some cases a user may want to restart a failed job for example when input files become available or the storage element for the output files came back online or when a proxy is renewed with arcrenew This can be done using the arcresume command Make sure your proxy is still valid or when uncertain run arcproxy followed by arcrenew before arcresume The job will be resumed from the state where it has failed 11 all jobs j
37. ectory with the name jobs dat Example joblist home user run jobs dat joblist C run jobs dat The format to use when creating a new job list file for storing information about active jobs Two possible formats can be used XML and BDB When using the former job information is stored in plain text as XML while the latter will use a Berkeley database to store job information There is no command line option corresponding to this attribute The default format is BDB Note that this option is only used by tools writing job information such as arcsub arcsync arctest etc and if they are instructed to use non existing new job list file i e create a new Example joblisttype XML joblisttype BDB Specifies default Bartender services Multiple Bartender URLs should be separated with a blank space These URLs are used by the chelonia command line tool the Chelonia FUSE plugin and by the data tool commands arcls etc Example bartender http my bar com tender 4 1 BLOCK COMMON 43 Specifies non standard location of proxy certificate It is used by arcproxy or similar tools during proxy generation and all other tools during establishing of a secure connection This attribute corresponds to the P command line option of arcproxy Example proxypath tmp my proxy Specifies a non standard location of user s private key It is used by arcproxy or similar tools during proxy generation This
38. ed in a standard form Or to enhance the performance or take advantage of various features it can have additional options For a metadata service URL construction is the following where the nested URL s are physical replicas Options are passed on to all replicas but if it is desired to use the same option with a different value for all replicas the option can be specified as a common option using the following syntax 35 36 CHAPTER 3 URLS In user level tools URLs may be expressed in this syntax or there may simpler ways to contruct complex URLs In particular command line tools such as arccp and the xRSL and JSDL job description languages provide methods to express URLs and options in a simpler way For the SRM service the syntax is Versions 1 1 and 2 2 of the SRM protocol are supported The default service_path is srm managerv2 when the server supports v2 2 srm managervl otherwise For Rucio the URLs look like The Rucio authorisation URL can be specified with the environment variable RUCIO_AUTH_URL The Rucio account to use can be specified either through the rucioaccount URL option or RUCIO_ACCOUNT environment variable If neither are specified the account is taken from the VOMS nickname attribute For ACIX the URLs look like The URL components are host port Hostname or IP address and port of a server lfn Logical File Name url URL of the file as registered in indexing service service_pat
39. enchmark name ranks targets according to a given benchmark as specified by the name If no benchmark is specified 2000 used Data ranks targets according the amount of megabytes of the requested input files that are already in the computing resources cache Note that only targets running the A REX BES interface can supply this information PythonBroker lt module gt lt class gt arguments ranks targets using any user supplied custom Python broker module optionally with broker arguments Such module can reside anywhere in user s PYTHONPATH lt otherbroker gt arguments ranks targets using any user supplied custom C broker plu gin optionally with broker arguments Default location for broker plugins on Linux systems is usr lib arc may depend on the operating system or the one specified by the ARC_PLUGIN_PATH To write a custom broker in one has to write a new specialization of the BrokerPlugin base class and implement the methods set match and operator in the new class The class should be compiled as a loadable module that has the proper ARC plugin descriptor for the new broker For example to build a broker plugin MyBroker one executes g I arc install include L arc install lib pkg config cflags glibmm 2 4 libxml1 2 0 o libaccmybroker so shared MyBroker cpp For more details refer to libarclib documentation 2 It often happens that some sites tha
40. erfaceName 21 long format extended information print a list of services configured in the client conf list the available plugins timeout for network communication in seconds de fault 20 verbosity level FATAL ERROR WARNING INFO VERBOSE or DEBUG default WARNING configuration file default HOME arc client conf print version information on Linux print help page explicitly select or reject a specific site explicitly select or reject a specific index server select one or more computing elements by an alias for a single CE a group of CEs or a URL select one or more registries by an alias for a single registry a group of registries or a URL skip the service with the given URL during service discovery only get information about executon tar gets which supports this job submission interface e g org nordugrid gridftpjob org ogf glue emies activitycreation org ogf bes the computing element specified by URL at the com mand line should be queried using this information interface possible options org nordugrid ldapng org nordugrid ldapglue2 org nordugrid wsrfglue 2 org ogf glue emies resourceinfo The arcinfo command is used to obtain information about clusters and queues targets available on the Grid Either the cluster or index flag should be used to specify the target s which should be queried for information Both of these flags take a service endpoint as argument See arcsub and
41. eries default 20 sec verbosity level is one of FATAL ERROR WARN ING INFO VERBOSE or DEBUG configuration file on Linux default HOME arc client conf print version information print help page explicitly select or reject a specific site select one or more computing elements by an alias for a single CE a group of CEs or a URL skip jobs which are on a computing element with a given URL list of job IDs and or job names The arcstat command returns the status of jobs in the Grid and is typically issued with a job ID as returned by arcsub as an argument It is also possible to use job name instead of ID but if several jobs have identical names information will be collected about all of them More than one job ID and or name can be given When several of the lt cluster gt i lt jobidfile gt and job command line options are spec ified the command returns information about all jobs listed on the command line plus all jobs on the specified clusters plus all jobs from the specified jobidfile However the c lt cluster gt for ARC 11 05 r lt URL gt for ARC 12 05 and s lt status gt options will filter the jobs selected by the above mentioned options or if none of those are specified then 2 2 JOB SUBMISSION AND MANAGEMENT 17 these will filter all the jobs For example arcstat s Finished c mycluster lt jobid gt will return information about the finished jobs on mycluster plus about
42. es or a URL R rejectdiscovery URL skip the service with the given URL during service discovery The ARC client keeps a local list of jobs in the user s home directory If this file is lost corrupt or the user wants to recreate the file on a different workstation the arcsync command will recreate this file from the information available in the Information System Since the information about a job retrieved from a cluster can be slightly out of date if the user very recently submitted or removed a job a warning is issued when this command is run The f option disables this warning If the job list is not empty when invoking syncronisation the old jobs will be merged with the new jobs unless the T option is given note the capital T in which case the job list will first be truncated and then the new jobs will be added 2 2 6 arcinfo The arcinfo command is used to obtain status information about clusters on the Grid 2 2 JOB SUBMISSION AND MANAGEMENT Options 1 long L list configured services P listplugins t timeout d debug z conffile 1 h 1 Options in ARC 11 05 cluster g index Options in ARC 12 05 cluster g index R rejectdiscovery S submissioninterface Options in ARC 13 02 D infointerface seconds verbosity filename name url designator designator URL InterfaceName Int
43. for download in job description to the user s computer Options 11 all jobs j joblist filename the file storing information about active jobs default arc jobs dat i jobids from file filename file containing a list of job IDs status statusstr only select jobs whose status is statusstr D dir dirname download path the job directory will be created in that location J usejobname dirname use the job name instead of the short ID as the job directory name 2 2 JOB SUBMISSION AND MANAGEMENT 19 k keep files in the Grid do not clean f force force download overwrite existing job directory P listplugins list the available plugins t timeout time timeout for queries default 20 sec d debug verbosity verbosity level is one of FATAL ERROR WARN ING INFO VERBOSE or DEBUG z conffile filename configuration file on Linux default HOME arc client conf v 1 print version information h 1 print help page Options in ARC 11 05 cluster name explicitly select or reject a specific site cluster Options in ARC 12 05 c cluster designator select one or more computing elements by an alias for a single CE a group of CEs or a URL r rejectmanagement URL skip jobs which are on a computing element with a given URL Arguments job list of job IDs and or job names Only the results of jobs that have finished can be dow
44. g org 2135 Mds Vo name testvo o grid windex ARC1 https index2 ng org 50000 isis computing ARC1 https ce arc org 60000 arex computing CREAM ldap ce glite org 2170 o grid This attribute is multi valued This attribute can be used to indicate that a certain service should be rejected blacklisted Several services can be listed separated with a blank space no line breaks allowed Example rejectservices computing ARC1 https bad service org arex Domain name of a service which should be rejected during service discovery Jobs will not be submitted to any interface of that service Multiple rejectdiscovery commands can be used Example rejectdiscovery bad server org rejectdiscovery bad2 server org 4 1 BLOCK COMMON 41 During job management operations the jobs belonging to this service will be skipped Multiple reject management commands can be used Example rejectmanagement bad server org rejectmanagement bad2 server org Default information interface used by service discovery and status query operations If a service has no infointerface specified in its configuration block see Section 4 2 this will be used by default Example infointerface org nordugrid ldapng Default submission interface used by job management operations If a service has submissioninterface specified in its configuration block see Section 4 2 this will be used by default Example submissioninterface org nordug
45. g elements by an alias for a single CE a group of CEs or a URL select one or more registries by an alias for a single registry a group of registries or a URL skip the service with the given URL during service discovery only use this interface for submit ting e g org nordugrid gridftpjob org ogf glue emies activitycreation org nordugrid xbes the computing element specified by URL at the com mand line should be queried using this information interface possible options org nordugrid ldapng org nordugrid ldapglue2 org nordugrid wsrfglue2 org ogf glue emies resourceinfo test job 1 runtime in minutes There are currently three test jobs defined Once submitted their results can be inspected and retrieved in a usual manner using arccat arcstat arcget etc Test job descriptions 1 ARC 11 05 and 12 05 A classical Hello World job printing hello grid text to standard output at a remote execution site ARC 13 02 The job calculates prime numbers for a number of minutes given by r default 5 and outputs the list to stderr The source code for the prime number program the Makefile and the executable are downloaded to the cluster from HTTP and FTP servers and the program is compiled before running 2 Lists all environment variables defined for the Grid user at the remote site using standard output 3 Downloads the pre defined input file from the HTTP server and produces an output file by copy ing input
46. h End point path of the web service file File name with full path option URL option commonoption URL option for all replicas metadataoption Metadata option The following URL options are supported threads lt number gt specifies number of parallel streams to be used by GridFTP or HTTP s g default value is 1 maximal value is 10 exec yes means the file should be treated as executable specify if file must be uploaded to this destination even if job processing failed default is no renew check invariant indicates whether the file should be cached default for input files in A REX is yes renew forces a download of the file even if the cached copy is still valid copy forces the cached file to be copied rather than linked to the session dir this is useful if for example the file is to be modified check forces a check of the permission and modification time against the original source invariant disables checking the original source modification time check option is available in ARC 2 0 0 and above invariant option is available in ARC 3 0 0 and above readonly yes no secure yes no blocksize lt number gt checksum cksum adler32 no overwrite yes no protocol gsi gssapi ss1 tls ss13 spacetoken lt pattern gt autodir yes no tcpnodelay yes no transferprotocol protocols rucioaccount account httpputpartial yes no
47. ion interface possible options org nordugrid ldapng org nordugrid ldapglue2 org nordugrid wsrfglue2 org ogf glue emies resourceinfo Arguments job list of job IDs and or job names More than one job ID and or job name can be given If several jobs were submitted with the same job name all those jobs will be resubmitted If the job description of a job to be resubmitted contained any local input files checksums of these was calculated and stored in the job list and those will be used to check whether the files has changed If local input files has changed the job will not be resubmitted In case the job description is not found in the job list an attempt will be made to retrieve it from the cluster holding the orignal job This however may fail since both the submission client and the cluster can have made modifications to the job description Upon resubmision the job will receive a new job ID and the old job ID will be stored in the local job list file enabling future back tracing of the resubmitted job Upon resubmision the job will receive a new job ID The old job ID will be kept in the local job list file enabling future back tracing of the resubmitted job Regarding command line options arcresub behaves much like arcsub except that in this case indicates not the submission target site but on the contrary the site from which the jobs will be resubmitted Submission target site is specified with option q If you
48. is not set then the cacertificatesdirectory attribute in the client configuration file client conf will be used The o order attribute can be used several times e g knowarc eu coredev Developer knowarc eu testers Tester Note that it does not make sense to specify the order if you have two or more different VOMS server specified When getting the delegated credentials from a MyProxy server using the option regular credentials certificate and key are not needed MyProxy functionality can be used together with VOMS function ality Options voms and vomses can be used for the GET command if VOMS attributes are required to be included in the proxy Supported constraints to be specified with the option are validityStart time e g 2008 05 29T10 20 30Z time when certificate becomes valid Default is now e validityEnd time time when certificate becomes invalid Default is 43200 12 hours from start e validityPeriod time e g 43200 or 12h or 12H for how long certificate is valid If neither validityPeriod nor validityEnd are specified default is 12 hours for local proxy and 168 hours for delegated proxy on MyProxy server 10 CHAPTER 2 COMMANDS vomsACvalidityPeriod time e g 43200 or 12h or 12H for how long VOMS AC is valid Default is the least value of 12 hours and validityPeriod myproxyvalidityPeriod time duration of proxies delegated by MyProxy server
49. is will submit a job to either of the sites known by the alias alias or member of the group group or having the URL URL In ARC 11 05 to submit a job to any site except badsite use sign in front of the name arcsub c badsite job xrsl See below for description of different kinds of designators which can be used with c and g options in ARC 11 05 In ARC 12 05 and higher to submit a job to any site except badsite example com use the R option with the site s URL which can be shortened to a domain name arcsub R badsite example com job xrsl The arcsub command locates the available sites by querying the information system unless option is used in which case only the listed sites are queried Default index services for the information system are specified in the configuration template distributed with the middleware and can be overwritten both in the user s configuration see Section 4 and from the command line using option g Different Grid middlewares may use different notation for such index services In ARC 11 05 client versions 1 the designators for and are either alias names group names or URLs interface URL where interface is optional specifying the computing service flavour and the corresponding plugin to be used when handling the URL Possible flavours are eroups are only available in ARC 12 05 14 CHAPTER 2 COMMANDS ARCO Legacy ARC execution and index services requires
50. jobs selected by the above mentioned options or if none of those are specified then these will filter all the jobs For example arcresume s FAILED c mycluster lt jobid gt will resume the failed jobs on mycluster plus lt jobid gt but only if it is failed Or arcresume 1 jobidfile mycluster example com will resume all jobs which are in the jobidfile but not on mycluster example com 2 2 11 arcresub Quite often it happens that a user would like to re submit a job but has difficulties recovering the original job description file e g the xRSL file This happens when job description files are created by scripts on fly and matching of job description to the job ID is not straightforward The utility called arcresub helps in such situations allowing users to resubmit jobs Options a all all jobs j joblist filename the file storing information about active jobs default arc jobs dat i jobids from file filename file containing a list of job IDs jobids to file filename the IDs of the submitted jobs will be appended to this file m re submit to the same site M not same do not resubmit to the same cluster status statusstr only select jobs whose status is statusstr k keep keep files in the Grid do not clean b broker string select broker method default is Random P listplugins list the available plugins t timeout time timeout for queries default 20 sec
51. more registries by an alias for a single registry a group of registries or a URL skip the service with the given URL during service discovery only use this interface for submit ting e g org nordugrid gridftpjob org ogf glue emies activitycreation org ogf bes submit directly no resource discovery or matchmak ing the computing element specified by URL at the com mand line should be queried using this information interface possible options org nordugrid ldapng org nordugrid ldapglue2 org nordugrid wsrfglue2 org ogf glue emies resourceinfo file s describing the job s to be submitted Here myjob jsd1 is a file containing job description Note that in this example f is omitted since the job description file is the last item on the command line Please remember that you must have valid credentials see Section 2 1 and be authorised at at least one Grid site The job must be described using one of the supported job description languages The description can be entered either as an argument on the command line or can be read from a file as in the example above Several jobs can be requested at the same time by giving more than one file name as an argument or by repeating the f or e options It is possible to mix e and f options in the same arcsub command simple Hello World job description myjob jsdl using the standard JSDL language is shown below lt xml version 1 0 encoding UTF 8 gt lt Job
52. n be used in the command line to select all members of the group Possible names of blocks are registry lt alias gt for registry services and computing lt alias gt for computing services Example registry index1 url ldap index1 nordugrid org 2135 Mds Vo name NorduGrid o grid registryinterface org nordugrid ldapegiis default yes group favs registry index2 url ldap index2 nordugrid org 2135 Mds Vo name NorduGrid o grid computing myce url myce example com computing ce url https ce example com 60000 arex infointerface org ogf glue emies resourceinfo submissioninterface org ogf glue emies activitycreation default yes group favs The URL of the service The URL can be shortened to the domain name in which case the client will try to guess the missing parts protocol port paths This is the only mandatory option in a service block Example url https example com 60000 arex Setting this to yes indicates that this service should be used by default by the commands if there were no computing elements or registries given as command line arguments If default is not set then the service can only be enabled through command line options Example default yes The name of the group the service belongs to Services can be selected by specifying the name of the group as command line arguments instead of the alias of the service A service can belong to multiple groups 4 2 SERVI
53. name the file storing information about active jobs on Linux default arc jobs dat jobids to file filename the IDs of the submitted jobs will be appended to this file D dryrun add dryrun option to the job description x dumpdescription do not submit dump transformed job description to stdout b broker string select broker method default is Random 12 P listplugins t timeout d z conffile v version h help Options in ARC 11 05 c cluster 1 Options in ARC 12 05 c cluster g 1 R rejectdiscovery S submissioninterface Options in ARC 13 02 direct I infointerface Arguments filename seconds verbosity filename designator designator designator designator URL InterfaceName InterfaceName A typical Grid job submission looks like arcsub myjob jsdl CHAPTER 2 COMMANDS list the available plugins timeout for network communication in seconds de fault 20 verbosity level FATAL ERROR WARNING INFO VERBOSE or DEBUG default WARNING configuration file default HOME arc client conf print version information on Linux print help page explicitly select or reject a specific site explicitly select or reject a specific index server select one or more computing elements by an alias for a single CE a group of CEs or a URL select one or
54. needs to take into account that this ID is probably different from the one provided by arcsub An example is an ID provided by the A REX computing service That service provides OGSA BES compatible interface for job management and the ID contains an XML document compliant with OGSA BES specifications Thttp www spec org cpu2000 CINT2000 16 2 2 2 arcstat CHAPTER 2 COMMANDS Options all j joblist filename i jobids from file filename status statusstr 1 long S criterion R criterion u show unavailable print jobids P listplugins t timeout time d debug verbosity z conffile filename version h 1 Options in ARC 11 05 c cluster Options in ARC 12 05 c cluster designator rejectmanagement URL Arguments job all jobs the file storing information about active jobs on Linux default SHOME arc jobs dat file containing a list of job IDs only select jobs whose status is statusstr long format extended information sort jobs according to job ID criterion jobid sub mission time submissiontime or job name job name reverse sorting of jobs according to job ID submis sion time or job name show jobs where status information is unavailable instead of the status only the IDs of the selected jobs will be printed list the available plugins timeout for qu
55. new proxy to the job This can be done while a job is still running thus preventing it from failing If a job has failed in file upload due to expired proxy arcrenew can be issued whithin 24 hours or whatever is the expiration time set by the site after the job end which must be followed by arcresume The Grid Manager or A REX will then attempt to finalize the job by uploading the output files to the desired location Options a all all jobs j joblist filename the file storing information about active jobs default arc jobs dat i jobids from file filename file containing a list of job IDs status statusstr only select jobs whose status is statusstr P listplugins list the available plugins t timeout time timeout for queries default 20 sec d debug verbosity verbosity level is one of FATAL ERROR WARN ING INFO VERBOSE or DEBUG z conffile filename configuration file on Linux default HOME arc client conf version print version information h 1 print help page Options in ARC 11 05 c cluster explicitly select or reject a specific site cluster Options in ARC 12 05 cluster designator select one or more computing elements by an alias for a single CE a group of CEs or a URL rejectmanagement URL skip jobs which are on a computing element with a given URL Arguments job list of job IDs and or job names Prior to using arcrenew be sur
56. nloaded Just like in arcstat and arccat cases the job can be referred to either by the job ID that was returned by arcsub at submission time or by its name if the job description contained a job name attribute By default the job is downloaded into a newly created directory in the current path with the name typically being a large random string In order to instruct arcget to use another path use option note the capital D e g arcget D tmp myjobs Test job nr 1 After downloading your jobs will be erased from the execution site Use command line option to keep finished jobs in the Grid Other command line options are identical to those of e g arcstat When several of the lt cluster gt 1 lt jobidfile gt and job command line options are spec ified the command retrieves all jobs listed on the command line plus all jobs on the specified clusters plus all jobs from the specified jobidfile However the c lt cluster gt for ARC 11 05 r lt URL gt for ARC 12 05 and s lt status gt options will filter the jobs selected by the above mentioned options or if none of those are specified then these will filter all the jobs For example arcget s Finished c mycluster lt jobid gt will retrieve the fin ished jobs on mycluster plus lt jobid gt but only if it is finished Or arcget i jobidfile r mycluster example org will retrieve jobs which are in the jobidfile but not on mycluster exampl
57. oxy validity in seconds e validityLeft duration of remaining proxy validity in seconds vomsVO VO name represented by the VOMS attribute e vomsSubject subject of the certificate for which the VOMS attribute was issued e vomsIssuer subject of the service which issued the VOMS certificate e vomsACvalidityStart timestamp when the VOMS attribute validity starts e vomsACvalidityEnd timestamp when the VOMS attribute validity ends e vomsACvalidityPeriod duration of the VOMS attribute validity in seconds vomsACvalidityLeft duration of the remaining VOMS attribute validity in seconds proxyPolicy policy associated with the proxy keybits size of the proxy certificate key in bits e signingAlgorithm algorithm used to sign the proxy certificate arcproxy makes use of the following configuration files etc vomses common for all users file containing a list of selected VO contact points one VO per line for example gin kuiken nikhef nl 15050 O dutchgrid O hosts OU nikhef nl CN kuiken nikhef nl gin ggf org nordugrid org voms uninett no 15015 0 Grid 0 NorduGrid CN host voms ndgf org nordugrid org 2 2 JOB SUBMISSION AND MANAGEMENT 11 voms vomses same as etc vomses but located in user s home area If exists has precedence over etc vomses The order of the parsing of vomses location is 1 command line options 2 client configuration file arc client
58. proxy file is not specified with option the value of X509_USER_PROXY environment variable is used explicitly If no such value is provided the default location is used lt TEMPORARY DIRECTORY gt x509up_u lt USER ID gt Here TEMPORARY DIRECTORY is derived from environ ment variables TMPDIR TMP TEMP or default location tmp is used The public certificate file as specified with option C can be either pem der or pkcs12 formated If this option is not set then the path specified by the environment variable X509_USER_CERT will be searched If X509_USER_CERT is not set then the certificatepath attribute in the client configuration file client conf will be used If after all these attempts the certificate still is not found then file usercert pem wll be searched arc globus etc arc and If the certificate file as specified with option is in pkcs12 format then no need to specify private key with option K If option K is not set and the certificate is not pkcs12 then the path specified by the environment variable X509_USER_KEY will be searched If X509_USER_KEY is not set then the keypath attribute in the client configuration file client conf will be used If after all these attempts the key still is not found then file userkey pem will be searched in arc globus etc arc and If option T is not set then the path specified by the environment variable X509_CERT_DIR will be searched If X509_CERT_DIR
59. re In addition arctest can provide basic information about available user credentials proxy certificate Options J jobid integer test job number j joblist filename the file storing information about active jobs on Linux default arc jobs dat o jobids to file filename the IDs of the submitted jobs will be appended to this file D add dryrun option to the job description dumpdescription do not submit dump transformed job description to stdout E certificate prints information about available user credentials b broker string select broker method default is Random t timeout seconds timeout for network communication in seconds de fault 20 d debug verbosity verbosity level FATAL ERROR WARNING INFO VERBOSE or DEBUG default WARNING z conffile filename configuration file on Linux default HOME arc client conf 2 4 TEST SUITE version h help Options in ARC 11 05 c cluster g 1 Options in ARC 12 05 c cluster g 1 R rejectdiscovery S submissioninterface Options in ARC 13 02 I infointerface r runtime url url designator designator URL InterfaceName InterfaceName ant 33 print version information print help page explicitly select or reject a specific site explicitly select or reject a specific index server select one or more computin
60. rid gridftpjob Default verbosity debug level to use for the ARC clients Corresponds to the d command line option of the clients Default value is WARNING possible values are FATAL ERROR WARNING INFO VERBOSE or DEBUG Example verbosity INFO Sets the period of time the client should wait for a service information computing storage etc to respond when communicating with it The period should be given in seconds Default value is 20 seconds This attribute corresponds to the t command line option Example timeout 10 42 CHAPTER 4 ARC CLIENT CONFIGURATION Configures which brokering algorithm to use during job submission This attribute corresponds to the b command line option The default one is the Random broker that chooses targets randomly Another possibility is for example the FastestQueue broker that chooses the target with the shortest estimated queue waiting time For an overview of brokers please refer to Section 2 2 1 Example brokername Data This attribute is used in case a broker comes with arguments This corresponds to the parameter that follows column in the b command line option Example brokerarguments cow The file storing information about active jobs This file will be used by commands such as arcsub arcstat arcsync etc to read and write information about jobs This attribute corresponds to the j command line option The default location of the file on Linux platforms is in the SHOME arc dir
61. status matches the instruction This option must be given together with either a or ones e g arcstat as Finished Other command line options are identical to those of arcsub 2 2 3 arccat It is often useful to monitor the job progress by checking what it prints on the standard output or error The command arccat assists here extracting the corresponding information from the execution cluster and dumping it on the user s screen It works both for running tasks and for the finished ones This allows a user to check the output of the finished task without actually retreiving it Options a all all jobs joblist filename the file storing information about active jobs default arc jobs dat i jobids from file filename file containing a list of job IDs status statusstr only select jobs whose status is statusstr o stdout show the stdout of the job default e stderr show the stderr of the job 1 joblog show the CE s error log of the job P listplugins list the available plugins t timeout time timeout for queries default 20 sec d debug verbosity verbosity level is one of FATAL ERROR WARN ING INFO VERBOSE or DEBUG 18 CHAPTER 2 COMMANDS z conffile filename configuration file on Linux default HOME arc client conf v version print version information h 1 print help page Options in ARC 11 05 cluster url explicitly select or reje
62. t arcsub has to contact are slow to answer or are down altogether This will not prevent you from submitting a job but will slow down the submission To speed it up you may want to specify a shorter timeout default is 20 seconds with the t option arcsub t 5 myjob jsdl Default value for the timeout can be set in the user s configuration file see Section If you would like to get diagnostics of the process of resource discovery and requirements matching a very useful option is The following command arcsub d VERBOSE myjob xrsl will print out the steps taken by the ARC client to find the best cluster satisfying your job requirements Possible diagnostics levels in the order of increasing verbosity are FATAL ERROR WARNING INFO VERBOSE and DEBUG Default is WARNING and it can be set to another value in the user s configuration file Default configuration file on Linux platforms is HOME arc client conf However a user can choose any other pre defined configuration through option z Command line option v prints out version of the installed ARC client package and option h provides a short help text For certain advanced computational jobs which may need to communicate their status to some external services there may be a need for knowing internal job ID For jobs accepted by ARC computational services this information can be found in the local for the job executable environment variable GRID_GLOBAL_JOBID One
63. t to wait for the system to clean up the job leftovers but to use 2 2 JOB SUBMISSION AND MANAGEMENT 23 arcclean to release the disk space and to remove the job ID from the list of submitted jobs and from the Information System Options a all all jobs j joblist filename the file storing information about active jobs default arc jobs dat i jobids from file filename file containing a list of job IDs status statusstr only select jobs whose status is statusstr f force removes the job ID from the local list even if the job is not found on the Grid P listplugins list the available plugins t timeout time timeout for queries default 20 sec d debug verbosity verbosity level is one of FATAL ERROR WARN ING INFO VERBOSE or DEBUG z conffile filename configuration file on Linux default HOME arc client conf v version print version information h 1 print help page Options in ARC 11 05 c cluster name explicitly select or reject a specific site cluster Options in ARC 12 05 cluster designator select or more computing elements by an alias for a single CE a group of CEs or a URL rejectmanagement URL skip jobs which are on a computing element with a given URL Arguments job list of job IDs and or job names Only jobs that have finished or were cancelled can be cleaned It happens ever so often that the job is cleaned by the system or
64. the configuration notes in Section 4 for description of these Detailed information about queried computing services can be obtained by specifying the long flag When specifying the index flag the information about the computing services registered at the index server will be queried rather than the status of the index server itself 2 2 7 arckill It happens that a user may wish to cancel a job This is done by using the arckill command A job can be killed almost at any stage of processing through the Grid 22 CHAPTER 2 COMMANDS Options all all jobs j joblist filename the file storing information about active jobs default arc jobs dat i jobids from file filename file containing a list of job IDs status statusstr only select jobs whose status is statusstr k keep files in the Grid do not clean listplugins list the available plugins t timeout time timeout for queries default 20 sec d debug verbosity verbosity level is one of FATAL ERROR WARN ING INFO VERBOSE or DEBUG z conffile filename configuration file on Linux default HOME arc client conf v 1 print version information h 1 print help page Options in ARC 11 05 c cluster url explicitly select or reject a specific site Options in ARC 12 05 cluster designator select one or more computing elements by an alias for a single CE a group of CEs or a UR
65. this tool can be found in Section 3 2 3 2 arccp is a powerful tool to copy files over the Grid It is a part of the A REX but can be used by the User Interface as well Options use passive transfer off by default if secure is on on by default if secure is not requested n do not try to force passive transfer f force if the destination is an indexing service and not the same as the source and the destination is already registered then the copy is normally not done How ever if this option is specified the source is assumed to be a replica of the destination created in an un controlled way and the copy is done like in case of replication Using this option also skips validation of completed transfers i indicate show progress indicator If the transfer time is short then there may be no indicator T notransfer do not transfer file just register it destination must be non existing meta url u use secure transfer insecure by default y path path to local cache use to put file into cache See 4 for information on caching r recursive operate recursively if possible D depth recursion_level operate recursively if possible up to specified level 0 no recursion available in ARC 13 02 R retries number how many times to retry transfer of every file before failing L location URL physical file to write to
66. tp www nordugrid org documents a rex pdf NORDUGRID TECH 14 5 F Pacini and A Maraschini Job Description Language attributes specification 2007 URL edms cern ch document 590869 1 EGEE JRA1 TEC 590869 JDL Attributes v0 8 6 A Sim A Shoshani et al The Storage Resource Manager Interface SRM Specification v2 2 GFD R P 129 2008 URL http www org documents GFD 129 pdf 7 O Smirnova Extended Resource Specification Language The NorduGrid Collaboration URL http www nordugrid org documents xrsl pdf NORDUGRID MANUAL 4 8 M Smith and T A Howes LDAP Programming Directory Enabled Applications with Lightweigt Directory Access Protocol Macmillan 1997 Index Numbers written in italic refer to the page where the corresponding entry is described numbers underlined refer to the definition numbers in roman refer to the pages where the entry is used B arcs b h APCCAL os adaos mi ei faves broker arcclean arctest daro a aost configuration commands bartender ssi fh Pe BECO siren tinera brokerarguments arckill Bb Rae s brokername arcl eon eR ES ae cacertificatepath arcmkdir cacertificatesdirectory arcinfo ape edita eaan

ARC Clients User Manual

Contents

Download Pdf Manuals

Related Search

Related Contents