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

   1 ---
   2 layout: default
   3 navsection: userguide
   4 title: "Concurrent Crunch tasks"
   5 ...
   6
   7 {% include 'pipeline_deprecation_notice' %}
   8
   9 In the previous tutorials, we used @arvados.job_setup.one_task_per_input_file()@ to automatically create concurrent jobs by creating a separate task per file.  For some types of jobs, you may need to split the work up differently, for example creating tasks to process different segments of a single large file.  This tutorial will demonstrate how to create Crunch tasks directly.
  10
  11 Start by entering the @crunch_scripts@ directory of your Git repository:
  12
  13 <notextile>
  14 <pre><code>~$ <span class="userinput">cd $USER/crunch_scripts</span>
  15 </code></pre>
  16 </notextile>
  17
  18 Next, using @nano@ or your favorite Unix text editor, create a new file called @concurrent-hash.py@ in the @crunch_scripts@ directory.
  19
  20 notextile. <pre>~/$USER/crunch_scripts$ <code class="userinput">nano concurrent-hash.py</code></pre>
  21
  22 Add the following code to compute the MD5 hash of each file in a collection:
  23
  24 <notextile> {% code 'concurrent_hash_script_py' as python %} </notextile>
  25
  26 Make the file executable:
  27
  28 notextile. <pre><code>~/$USER/crunch_scripts$ <span class="userinput">chmod +x concurrent-hash.py</span></code></pre>
  29
  30 Add the file to the Git staging area, commit, and push:
  31
  32 <notextile>
  33 <pre><code>~/$USER/crunch_scripts$ <span class="userinput">git add concurrent-hash.py</span>
  34 ~/$USER/crunch_scripts$ <span class="userinput">git commit -m"concurrent hash"</span>
  35 ~/$USER/crunch_scripts$ <span class="userinput">git push origin master</span>
  36 </code></pre>
  37 </notextile>
  38
  39 You should now be able to run your new script using Crunch, with "script" referring to our new "concurrent-hash.py" script.  We will use a different input from our previous examples.  We will use @887cd41e9c613463eab2f0d885c6dd96+83@ which consists of three files, "alice.txt", "bob.txt" and "carol.txt" (the example collection used previously in "fetching data from Arvados using Keep":{{site.baseurl}}/user/tutorials/tutorial-keep.html#dir).
  40
  41 <notextile>
  42 <pre><code>~/$USER/crunch_scripts$ <span class="userinput">cat &gt;~/the_job &lt;&lt;EOF
  43 {
  44  "script": "concurrent-hash.py",
  45  "repository": "$USER/$USER",
  46  "script_version": "master",
  47  "script_parameters":
  48  {
  49   "input": "887cd41e9c613463eab2f0d885c6dd96+83"
  50  }
  51 }
  52 EOF</span>
  53 ~/$USER/crunch_scripts$ <span class="userinput">arv job create --job "$(cat ~/the_job)"</span>
  54 {
  55  ...
  56  "uuid":"qr1hi-xxxxx-xxxxxxxxxxxxxxx"
  57  ...
  58 }
  59 ~/$USER/crunch_scripts$ <span class="userinput">arv job get --uuid qr1hi-xxxxx-xxxxxxxxxxxxxxx</span>
  60 {
  61  ...
  62  "output":"e2ccd204bca37c77c0ba59fc470cd0f7+162",
  63  ...
  64 }
  65 </code></pre>
  66 </notextile>
  67
  68 (Your shell should automatically fill in @$USER@ with your login name.  The job JSON that gets saved should have @"repository"@ pointed at your personal Git repository.)
  69
  70 Because the job ran in concurrent, each instance of concurrent-hash creates a separate @md5sum.txt@ as output.  Arvados automatically collates theses files into a single collection, which is the output of the job:
  71
  72 <notextile>
  73 <pre><code>~/$USER/crunch_scripts$ <span class="userinput">arv keep ls e2ccd204bca37c77c0ba59fc470cd0f7+162</span>
  74 ./md5sum.txt
  75 ~/$USER/crunch_scripts$ <span class="userinput">arv keep get e2ccd204bca37c77c0ba59fc470cd0f7+162/md5sum.txt</span>
  76 0f1d6bcf55c34bed7f92a805d2d89bbf alice.txt
  77 504938460ef369cd275e4ef58994cffe bob.txt
  78 8f3b36aff310e06f3c5b9e95678ff77a carol.txt
  79 </code></pre>
  80 </notextile>