doc/user/topics/tutorial-job1.html.textile.liquid

   1 ---
   2 layout: default
   3 navsection: userguide
   4 title: "Running a Crunch job on the command line"
   5 ...
   6
   7 This tutorial introduces how to run individual Crunch jobs using the @arv@ command line tool.
   8
   9 *This tutorial assumes that you are "logged into an Arvados VM instance":{{site.baseurl}}/user/getting_started/ssh-access.html#login, and have a "working environment.":{{site.baseurl}}/user/getting_started/check-environment.html*
  10
  11 You will create a job to run the "hash" Crunch script.  The "hash" script computes the md5 hash of each file in a collection.
  12
  13 h2. Jobs
  14
  15 Crunch pipelines consist of one or more jobs.  A "job" is a single run of a specific version of a Crunch script with a specific input.  You can also run jobs individually.
  16
  17 A request to run a Crunch job are is described using a JSON object.  For example:
  18
  19 <notextile>
  20 <pre><code>~$ <span class="userinput">cat &gt;~/the_job &lt;&lt;EOF
  21 {
  22  "script": "hash",
  23  "repository": "arvados",
  24  "script_version": "master",
  25  "script_parameters": {
  26   "input": "c1bad4b39ca5a924e481008009d94e32+210"
  27  },
  28  "no_reuse": "true"
  29 }
  30 EOF
  31 </code></pre>
  32 </notextile>
  33
  34 * @cat@ is a standard Unix utility that writes a sequence of input to standard output.
  35 * @<<EOF@ tells the shell to direct the following lines into the standard input for @cat@ up until it sees the line @EOF@.
  36 * @>~/the_job@ redirects standard output to a file called @~/the_job@.
  37 * @"repository"@ is the name of a git repository to search for the script version.  You can access a list of available git repositories on the Arvados Workbench under "*Compute* %(rarr)&rarr;% *Code repositories*":https://{{site.arvados_workbench_host}}/repositories.
  38 * @"script_version"@ specifies the version of the script that you wish to run.  This can be in the form of an explicit git revision hash, a tag, or a branch (in which case it will use the most recent commit on the specified branch).  Arvados logs the script version that was used in the run, enabling you to go back and re-run any past job with the guarantee that the exact same code will be used as was used in the previous run.
  39 * @"script"@ specifies the name of the script to run.  The script is searched for in the @crunch_scripts/@ subdirectory of the git repository.
  40 * @"script_parameters"@ are provided to the script.  In this case, the input is the PGP data Collection that we "put in Keep earlier":{{site.baseurl}}/user/tutorials/tutorial-keep.html.
  41 * Setting the @"no_reuse"@ flag tells Crunch not to reuse work from past jobs.  This helps ensure that you can watch a new Job process for the rest of this tutorial, without reusing output from a past run that you made, or somebody else marked as public.  (If you want to experiment, after the first run below finishes, feel free to edit this job to remove the @"no_reuse"@ line and resubmit it.  See what happens!)
  42
  43 Use @arv job create@ to actually submit the job.  It should print out a JSON object which describes the newly created job:
  44
  45 <notextile>
  46 <pre><code>~$ <span class="userinput">arv job create --job "$(cat ~/the_job)"</span>
  47 {
  48  "href":"https://qr1hi.arvadosapi.com/arvados/v1/jobs/qr1hi-8i9sb-1pm1t02dezhupss",
  49  "kind":"arvados#job",
  50  "etag":"ax3cn7w9whq2hdh983yxvq09p",
  51  "uuid":"qr1hi-8i9sb-1pm1t02dezhupss",
  52  "owner_uuid":"qr1hi-tpzed-9zdpkpni2yddge6",
  53  "created_at":"2013-12-16T20:44:32Z",
  54  "modified_by_client_uuid":"qr1hi-ozdt8-obw7foaks3qjyej",
  55  "modified_by_user_uuid":"qr1hi-tpzed-9zdpkpni2yddge6",
  56  "modified_at":"2013-12-16T20:44:32Z",
  57  "updated_at":"2013-12-16T20:44:33Z",
  58  "submit_id":null,
  59  "priority":null,
  60  "script":"hash",
  61  "script_parameters":{
  62   "input":"c1bad4b39ca5a924e481008009d94e32+210"
  63  },
  64  "script_version":"d9cd657b733d578ac0d2167dd75967aa4f22e0ac",
  65  "cancelled_at":null,
  66  "cancelled_by_client_uuid":null,
  67  "cancelled_by_user_uuid":null,
  68  "started_at":null,
  69  "finished_at":null,
  70  "output":null,
  71  "success":null,
  72  "running":null,
  73  "is_locked_by_uuid":null,
  74  "log":null,
  75  "runtime_constraints":{},
  76  "tasks_summary":{},
  77  "dependencies":[
  78   "c1bad4b39ca5a924e481008009d94e32+210"
  79  ],
  80  "log_stream_href":"https://qr1hi.arvadosapi.com/arvados/v1/jobs/qr1hi-8i9sb-1pm1t02dezhupss/log_tail_follow"
  81 }
  82 </code></pre>
  83 </notextile>
  84
  85 The job is now queued and will start running as soon as it reaches the front of the queue.  Fields to pay attention to include:
  86
  87  * @"uuid"@ is the unique identifier for this specific job.
  88  * @"script_version"@ is the actual revision of the script used.  This is useful if the version was described using the "repository:branch" format.
  89
  90 h2. Monitor job progress
  91
  92 Go to the "Workbench dashboard":https://{{site.arvados_workbench_host}} and visit *Activity* %(rarr)&rarr;% *Recent jobs*.  Your job should be near the top of the table.  This table refreshes automatically.  When the job has completed successfully, it will show <span class="label label-success">finished</span> in the *Status* column.
  93
  94 On the command line, you can access log messages while the job runs using @arv job log_tail_follow@:
  95
  96 notextile. <pre><code>~$ <span class="userinput">arv job log_tail_follow --uuid qr1hi-8i9sb-xxxxxxxxxxxxxxx</span></code></pre>
  97
  98 This will print out the last several lines of the log for that job.
  99
 100 h2. Inspect the job output
 101
 102 On the "Workbench dashboard":https://{{site.arvados_workbench_host}}, look for the *Output* column of the *Recent jobs* table.  Click on the link under *Output* for your job to go to the files page with the job output.  The files page lists all the files that were output by the job.  Click on the link under the *file* column to view a file, or click on the download icon <span class="glyphicon glyphicon-download-alt"></span> to download the output file.
 103
 104 On the command line, you can use @arv job get@ to access a JSON object describing the output:
 105
 106 <notextile>
 107 <pre><code>~$ <span class="userinput">arv job get --uuid qr1hi-8i9sb-xxxxxxxxxxxxxxx</span>
 108 {
 109  "href":"https://qr1hi.arvadosapi.com/arvados/v1/jobs/qr1hi-8i9sb-1pm1t02dezhupss",
 110  "kind":"arvados#job",
 111  "etag":"1bk98tdj0qipjy0rvrj03ta5r",
 112  "uuid":"qr1hi-8i9sb-1pm1t02dezhupss",
 113  "owner_uuid":"qr1hi-tpzed-9zdpkpni2yddge6",
 114  "created_at":"2013-12-16T20:44:32Z",
 115  "modified_by_client_uuid":null,
 116  "modified_by_user_uuid":"qr1hi-tpzed-9zdpkpni2yddge6",
 117  "modified_at":"2013-12-16T20:44:55Z",
 118  "updated_at":"2013-12-16T20:44:55Z",
 119  "submit_id":null,
 120  "priority":null,
 121  "script":"hash",
 122  "script_parameters":{
 123   "input":"c1bad4b39ca5a924e481008009d94e32+210"
 124  },
 125  "script_version":"d9cd657b733d578ac0d2167dd75967aa4f22e0ac",
 126  "cancelled_at":null,
 127  "cancelled_by_client_uuid":null,
 128  "cancelled_by_user_uuid":null,
 129  "started_at":"2013-12-16T20:44:36Z",
 130  "finished_at":"2013-12-16T20:44:53Z",
 131  "output":"dd755dbc8d49a67f4fe7dc843e4f10a6+54",
 132  "success":true,
 133  "running":false,
 134  "is_locked_by_uuid":"qr1hi-tpzed-9zdpkpni2yddge6",
 135  "log":"2afdc6c8b67372ffd22d8ce89d35411f+91",
 136  "runtime_constraints":{},
 137  "tasks_summary":{
 138   "done":2,
 139   "running":0,
 140   "failed":0,
 141   "todo":0
 142  },
 143  "dependencies":[
 144   "c1bad4b39ca5a924e481008009d94e32+210"
 145  ],
 146  "log_stream_href":null
 147 }
 148 </code></pre>
 149 </notextile>
 150
 151 * @"output"@ is the unique identifier for this specific job's output.  This is a Keep collection.  Because the output of Arvados jobs should be deterministic, the known expected output is <code>dd755dbc8d49a67f4fe7dc843e4f10a6+54</code>.
 152
 153 Now you can list the files in the collection:
 154
 155 <notextile>
 156 <pre><code>~$ <span class="userinput">arv keep ls dd755dbc8d49a67f4fe7dc843e4f10a6+54</span>
 157 ./md5sum.txt
 158 </code></pre>
 159 </notextile>
 160
 161 This collection consists of the @md5sum.txt@ file.  Use @arv keep get@ to show the contents of the @md5sum.txt@ file:
 162
 163 <notextile>
 164 <pre><code>~$ <span class="userinput">arv keep get dd755dbc8d49a67f4fe7dc843e4f10a6+54/md5sum.txt</span>
 165 44b8ae3fde7a8a88d2f7ebd237625b4f ./var-GS000016015-ASM.tsv.bz2
 166 </code></pre>
 167 </notextile>
 168
 169 This md5 hash matches the md5 hash which we "computed earlier":{{site.baseurl}}/user/tutorials/tutorial-keep.html.
 170
 171 h2. The job log
 172
 173 When the job completes, you can access the job log.  On the Workbench, visit *Activity* %(rarr)&rarr;% *Recent jobs* %(rarr)&rarr;% your job's UUID under the *uuid* column %(rarr)&rarr;% the collection link on the *log* row.
 174
 175 On the command line, the Keep identifier listed in the @"log"@ field from @arv job get@ specifies a collection.  You can list the files in the collection:
 176
 177 <notextile>
 178 <pre><code>~$ <span class="userinput">arv keep ls xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx+91</span>
 179 ./qr1hi-8i9sb-xxxxxxxxxxxxxxx.log.txt
 180 </code></pre>
 181 </notextile>
 182
 183 The log collection consists of one log file named with the job's UUID.  You can access it using @arv keep get@:
 184
 185 <notextile>
 186 <pre><code>~$ <span class="userinput">arv keep get xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx+91/qr1hi-8i9sb-xxxxxxxxxxxxxxx.log.txt</span>
 187 2013-12-16_20:44:35 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  check slurm allocation
 188 2013-12-16_20:44:35 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  node compute13 - 8 slots
 189 2013-12-16_20:44:36 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  start
 190 2013-12-16_20:44:36 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  Install revision d9cd657b733d578ac0d2167dd75967aa4f22e0ac
 191 2013-12-16_20:44:37 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  Clean-work-dir exited 0
 192 2013-12-16_20:44:37 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  Install exited 0
 193 2013-12-16_20:44:37 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  script hash
 194 2013-12-16_20:44:37 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  script_version d9cd657b733d578ac0d2167dd75967aa4f22e0ac
 195 2013-12-16_20:44:37 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  script_parameters {"input":"c1bad4b39ca5a924e481008009d94e32+210"}
 196 2013-12-16_20:44:37 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  runtime_constraints {"max_tasks_per_node":0}
 197 2013-12-16_20:44:37 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  start level 0
 198 2013-12-16_20:44:37 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  status: 0 done, 0 running, 1 todo
 199 2013-12-16_20:44:38 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575 0 job_task qr1hi-ot0gb-23c1k3kwrf8da62
 200 2013-12-16_20:44:38 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575 0 child 7681 started on compute13.1
 201 2013-12-16_20:44:38 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  status: 0 done, 1 running, 0 todo
 202 2013-12-16_20:44:39 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575 0 child 7681 on compute13.1 exit 0 signal 0 success=true
 203 2013-12-16_20:44:39 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575 0 success in 1 seconds
 204 2013-12-16_20:44:39 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575 0 output
 205 2013-12-16_20:44:39 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  wait for last 0 children to finish
 206 2013-12-16_20:44:39 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  status: 1 done, 0 running, 1 todo
 207 2013-12-16_20:44:39 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  start level 1
 208 2013-12-16_20:44:39 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  status: 1 done, 0 running, 1 todo
 209 2013-12-16_20:44:39 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575 1 job_task qr1hi-ot0gb-iwr0o3unqothg28
 210 2013-12-16_20:44:39 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575 1 child 7716 started on compute13.1
 211 2013-12-16_20:44:39 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  status: 1 done, 1 running, 0 todo
 212 2013-12-16_20:44:52 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575 1 child 7716 on compute13.1 exit 0 signal 0 success=true
 213 2013-12-16_20:44:52 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575 1 success in 13 seconds
 214 2013-12-16_20:44:52 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575 1 output dd755dbc8d49a67f4fe7dc843e4f10a6+54
 215 2013-12-16_20:44:52 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  wait for last 0 children to finish
 216 2013-12-16_20:44:52 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  status: 2 done, 0 running, 0 todo
 217 2013-12-16_20:44:52 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  release job allocation
 218 2013-12-16_20:44:52 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  Freeze not implemented
 219 2013-12-16_20:44:52 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  collate
 220 2013-12-16_20:44:53 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  output dd755dbc8d49a67f4fe7dc843e4f10a6+54+K@qr1hi
 221 2013-12-16_20:44:53 qr1hi-8i9sb-xxxxxxxxxxxxxxx 7575  finish
 222 </code></pre>
 223 </notextile>