1ARCSUB(1) NorduGrid Users Manual ARCSUB(1)
2
6 arcsub - ARC Submission
7
10 The arcsub command is used for submitting jobs to Grid enabled comput‐
11 ing resources.
12
15 arcsub [options] [filename ...]
16
19 -c, --cluster=name
20 select one or more computing elements: name can be an alias for
21 a single CE, a group of CEs or a URL
22 -g, --index=name
24 select one or more registries: name can be an alias for a single
25 registry, a group of registries or a URL
26 -R, --rejectdiscovery=URL
28 skip the service with the given URL during service discovery
29 -S, --submissioninterface=InterfaceName
31 only use this interface for submitting (e.g. org.nor‐
32 dugrid.gridftpjob, org.ogf.glue.emies.activitycreation,
33 org.ogf.bes)
34 -I, --infointerface=InterfaceName
36 the computing element specified by URL at the command line
37 should be queried using this information interface (possible op‐
38 tions: org.nordugrid.ldapng, org.nordugrid.ldapglue2, org.nor‐
39 dugrid.wsrfglue2, org.ogf.glue.emies.resourceinfo)
40 -e, --jobdescrstring=String
42 jobdescription string describing the job to be submitted
43 -f, --jobdescrfile=filename
45 jobdescription file describing the job to be submitted
46 -j, --joblist=filename
48 the file storing information about active jobs (default
49 ~/.arc/jobs.xml)
50 -o, --jobids-to-file=filename
52 the IDs of the submitted jobs will be appended to this file
53 -D, --dryrun
55 submit jobs as dry run (no submission to batch system)
56 --direct
58 submit directly - no resource discovery or matchmaking
59 -x, --dumpdescription
61 do not submit - dump job description in the language accepted by
62 the target
63 -P, --listplugins
65 list the available plugins
66 -t, --timeout=seconds
68 timeout in seconds (default 20)
69 -z, --conffile=filename
71 configuration file (default ~/.arc/client.conf)
72 -d, --debug=debuglevel
74 FATAL, ERROR, WARNING, INFO, VERBOSE or DEBUG
75 -b, --broker=broker
77 selected broker: Random (default), FastestQueue or custom. Use
78 -P to find possible options.
79 -v, --version
81 print version information
82 -?, --help
84 print help
85
87 filename ...
88 job description files describing the jobs to be submitted
89
91 arcsub is the key command when submitting jobs to Grid enabled comput‐
92 ing resources with the ARC client. As default arcsub is able to submit
93 jobs to A-REX, CREAM and EMI ES enabled computing elements (CEs), and
94 as always for successful submission you need to be authenticated at the
95 targeted computing services. Since arcsub is build on a modular li‐
96 brary, modules can be installed which enables submission to other tar‐
97 gets, e.g. the classic ARC CE Grid-Manager.
98 Job submission can be accomplished by specifying a job description file
100 to submit as an argument. arcsub will then by default perform resource
101 discovery on the Grid and then the discovered resources will be matched
102 to the job description and ranked according to the chosen broker
103 (--broker option). If no Grid environment has been configured, please
104 contact your system administrator, or setup one yourself in the client
105 configuration file (see files section). Another option is to explicitly
106 specify a registry service (or multiple) to arcsub using the --index
107 option, which accepts an URL, alias or group. Alternatively a specific
108 CE (or multiple) can be targeted by using the --cluster option. If such
109 a scenario is the most common, it is worthwhile to specify those CEs in
110 the client configuration as default services, which makes it superflu‐
111 ous to specify them as argument. In the same manner aliases and groups,
112 defined in the configuration file, can be utilized, and can be used as
113 argument to the --cluster or --index options. In all of the above sce‐
114 narios arcsub obtains resource information from the services which is
115 then used for matchmaking against the job description, however that
116 step can be avoided by specifying the --direct option, in which case
117 the job description is submitted directly to first specified endpoint.
118 The format of a classic GRIDFTP-based cluster URLs:
120 [ldap://]<hostname>[:2135/nordugrid-cluster-name=<hostname>,Mds-Vo-
121 name=local,o=grid]
122 Only the hostname part has to be specified, the rest of the URL is au‐
123 tomatically generated.
124 The format of an A-REX URL is:
126 [https://]<hostname>[:<port>][/<path>]
127 Here the port is 443 by default, but the path cannot be guessed, so if
128 it is not specified, then the service is assumed to live on the root
129 path.
130 Job descriptions can also be specified using the --jobdescrfile option
132 which expect the file name of the description as argument, or the
133 --jobdescrstring option which expect as argument the job description as
134 a string, and both options can be specified multiple times and one does
135 not exclude the other. The default supported job description languages
136 are xRSL and EMIES ADL.
137 If the job description is successfully submitted a job-ID is returned
139 and printed. This job-ID uniquely identifies the job while it is being
140 executed. On the other hand it is also possible that no CEs matches
141 the constraints defined in the description in which case no submission
142 will be done. Upon successful submission, the job-ID along with more
143 technical job information is stored in the job-list file (described be‐
144 low). The stored information enables the job management commands of the
145 ARC client to manage jobs easily, and thus the job-ID need not to be
146 saved manually. By default the job-list file is stored in the .arc di‐
147 rectory in the home directory of the user, however another location can
148 be specified using the --joblist option taking the location of this
149 file as argument. If the --joblist option was used during submission,
150 it should also be specified in the consecutive commands when managing
151 the job. If a Computing Element has multiple job submission interfaces
152 (e.g. gridftp, EMI-ES, BES), then the brokering algorithm will choose
153 one of them. With the --submissioninterface option the requested in‐
154 terface can be specified, and in that case only those Computing Ele‐
155 ments will be considered which has that specific interface, and only
156 that interface will be used to submit the jobs.
157 As mentioned above registry or index services can be specified with the
159 --index option. Specifying one or multiple index servers instructs the
160 arcsub command to query the servers for registered CEs, the returned
161 CEs will then be matched against the job description and those matching
162 will be ranked by the chosen broker (see below) and submission will be
163 tried in order until successful or reaching the end. From the returned
164 list of CEs it might happen that a troublesome or undesirable CE is se‐
165 lected for submission, in that case it possible to reject that cluster
166 using the --rejectdiscovery option and providing the URL (or just the
167 hostname) of the CE, which will disregard that CE as a target for sub‐
168 mission.
169 When multiple CEs are targeted for submission, the resource broker will
171 be used to filter out CEs which do not match the job description re‐
172 quirements and then rank the remaining CEs. The broker used by default
173 will rank the CEs randomly, however a different broker can be chosen by
174 using the --broker option, which takes the name of the broker as argu‐
175 ment. The broker type can also be specified in client.conf. The brokers
176 available can be seen using arcsub -P. By default the following bro‐
177 kers are available:
178 Random (default)
180 Chooses a random CE matching the job requirements.
181 FastestQueue
183 Ranks matching CEs according to the length of the job queue at
184 the CEs, ranking those with shortest queue first/highest.
185 Benchmark
187 Ranks matching CEs according to a specified benchmark, which
188 should be specified by appending the broker name with ':' and
189 then the name of the benchmark. If no option is given to the
190 Benchmark broker then CEs will be ranked according to the
191 'specint2000' benchmark.
192 Data Ranks matching CEs according to the amount of input data cached
194 by each CE, by querying the CE. Only CEs with the A-REX BES in‐
195 terface support this operation.
196 Null Choose a random CE with no filtering at all of CEs.
198 PythonBroker
200 User-defined custom brokers can be created in Python. See the
201 example broker SampleBroker.py or ACIXBroker.py (like Data bro‐
202 ker but uses the ARC Cache Index) that come installed with ARC
203 for more details of how to write your own broker. A PythonBroker
204 is specified by --broker PythonBroker:Filename.Class:args, where
205 Filename is the file containing the class Class which implements
206 the broker interface. The directory containing this file must be
207 in the PYTHONPATH. args is optional and allows specifying argu‐
208 ments to the broker.
209 Before submission, arcsub performs an intelligent modification of the
211 job description (adding or modifying attributes, even converting the
212 description language to fit the needs of the CE) ensuring that it is
213 valid. The modified job description can be printed by specifying the
214 --dumpdescription option. The format, i.e. job description language, of
215 the printed job description cannot be specified, and will be that which
216 will be sent to and accepted by the chosen target. Further information
217 from arcsub can be obtained by increasing the verbosity, which is done
218 with the --debug option where the default verbosity level is WARNING.
219 Setting the level to DEBUG will show all messages, while setting it to
220 FATAL will only show fatal log messages.
221 To validate your job description without actually submitting a job, use
223 the --dryrun option: it will capture possible syntax or other errors,
224 but will instruct the site not to submit the job for execution. Only
225 the grid-manager (ARC0) and A-REX (ARC1) CEs support this feature.
226
229 Submission of a job description file "helloworld.adl" to the Grid
230 arcsub helloworld.adl
231 A information index server (registry) can also be queried for CEs to
233 submit to:
234 arcsub -g registry.example.com helloworld.adl
235 Submission of a job description file "helloworld.adl" to ce.exam‐
237 ple.com:
238 arcsub -c ce.example.com helloworld.adl
239 Direct submission to a CE is done as:
241 arcsub --direct -c cd.example.com helloworld.adl
242 The job description can also be specified directly on the command line
244 as shown in the example, using the XRSL job description language:
245 arcsub -c example.com/arex -e \
246 ´&(executable="/bin/echo")(arguments="Hello World!")´
247 When submitting against CEs retrieved from information index servers it
249 might be useful to do resource brokering:
250 arcsub -g registry.example.com -b FastestQueue helloworld.adl
251 If the job has a large input data set, it can be useful to send it to a
253 CE where those files are already cached. The ACIX broker can be used
254 for this:
255 arcsub -g registry.example.com -b PythonBroker:ACIXBroker.ACIXBro‐
256 ker:https://cacheindex.ndgf.org:6443/data/index helloworld.adl
257 Disregarding a specific CE for submission submitting against an infor‐
259 mation index server:
260 arcsub -g registry.example.com -R badcomputingelement.com/arex hel‐
261 loworld.adl
262 Dumping the job description is done as follows:
264 arcsub -c example.com/arex -x helloworld.adl
265
268 ~/.arc/client.conf
269 Some options can be given default values by specifying them in
270 the ARC client configuration file. Registry and computing ele‐
271 ment services can be specified in separate sections of the con‐
272 fig. The default services can be specified by adding 'de‐
273 fault=yes' attribute to the section of the service, thus when no
274 --cluster or --index options are given these will be used for
275 submission. Each service has an alias, and can be member of any
276 number of groups. Then specifying the alias or the name of the
277 group with the --cluster or --index options will select the
278 given services. By using the --conffile option a different con‐
279 figuration file can be used than the default. Note that some in‐
280 stallations also have a system client configuration file, how‐
281 ever attributes in the client one takes precedence, and then
282 command line options takes precedence over configuration file
283 attributes.
284 ~/.arc/jobs.xml
287 This a local list of the user's active jobs. When a job is suc‐
288 cessfully submitted it is added to this list and when it is re‐
289 moved from the remote cluster it is removed from this list. This
290 list is used as the list of all active jobs when the user speci‐
291 fies the --all option to the various NorduGrid ARC user inter‐
292 face commands. By using the --joblist option a different file
293 can be used than the default.
294
297 X509_USER_PROXY
298 The location of the user's Grid proxy file. Shouldn't be set un‐
299 less the proxy is in a non-standard location.
300 ARC_LOCATION
303 The location where ARC is installed can be specified by this
304 variable. If not specified the install location will be deter‐
305 mined from the path to the command being executed, and if this
306 fails a WARNING will be given stating the location which will be
307 used.
308 ARC_PLUGIN_PATH
311 The location of ARC plugins can be specified by this variable.
312 Multiple locations can be specified by separating them by : (;
313 in Windows). The default location is $ARC_LOCATION/lib/arc (\ in
314 Windows).
315
318 APACHE LICENSE Version 2.0
319
322 ARC software is developed by the NorduGrid Collaboration
323 (http://www.nordugrid.org), please consult the AUTHORS file distributed
324 with ARC. Please report bugs and feature requests to
325 http://bugzilla.nordugrid.org
326
329 arccat(1), arcclean(1), arccp(1), arcget(1), arcinfo(1), arckill(1),
330 arcls(1), arcmkdir(1), arcproxy(1), arcrenew(1), arcresub(1), arcre‐
331 sume(1), arcrm(1), arcstat(1), arcsync(1), arctest(1)
332NorduGrid ARC 6.15.1 2022-07-03 ARCSUB(1)