mobilize-hdfs 1.36 → 1.291

checksums.yaml

@@ -0,0 +1,7 @@
+---
+SHA1:
+  metadata.gz: 7206b82c26abc7e437dd9fdcc801f89313d746db
+  data.tar.gz: 7aca4e3c7e1d6c18cdc41dff0f8f4bcfd7877a8c
+SHA512:
+  metadata.gz: 42f739d78375dc2822132236d7c1a4f08b462274703603672f91a1fec384aeea355ef00430d4c9ce6493c0b0019503afb49356a20aa6a61114ef7bc99fc1a09f
+  data.tar.gz: cbb1ca0e3f25fe1d685259fa92fc4e779fcaaa237f5d1e679edc581f0f86b1a8997ea0b445c49a4af4302d686ca9917ae8dd0471aab69902caf8ebc8cf41ed0e

data/README.md

@@ -1,4 +1,234 @@
-Mobilize
-========
+Mobilize-Hdfs
+===============
 
-Please refer to the mobilize-server wiki: https://github.com/DeNA/mobilize-server/wiki
+Mobilize-Hdfs adds the power of hdfs to [mobilize-ssh][mobilize-ssh].
+* read, write, and copy hdfs files through Google
+Spreadsheets.
+
+Table Of Contents
+-----------------
+* [Overview](#section_Overview)
+* [Install](#section_Install)
+  * [Mobilize-Hdfs](#section_Install_Mobilize-Hdfs)
+  * [Install Dirs and Files](#section_Install_Dirs_and_Files)
+* [Configure](#section_Configure)
+  * [Hadoop](#section_Configure_Hadoop)
+* [Start](#section_Start)
+  * [Create Job](#section_Start_Create_Job)
+  * [Run Test](#section_Start_Run_Test)
+* [Meta](#section_Meta)
+* [Author](#section_Author)
+
+<a name='section_Overview'></a>
+Overview
+-----------
+
+* Mobilize-hdfs adds Hdfs methods to mobilize-ssh.
+
+<a name='section_Install'></a>
+Install
+------------
+
+Make sure you go through all the steps in the
+[mobilize-base][mobilize-base] and [mobilize-ssh][mobilize-ssh]
+install sections first.
+
+<a name='section_Install_Mobilize-Hdfs'></a>
+### Mobilize-Hdfs
+
+add this to your Gemfile:
+
+``` ruby
+gem "mobilize-hdfs"
+```
+
+or do
+
+  $ gem install mobilize-hdfs
+
+for a ruby-wide install.
+
+<a name='section_Install_Dirs_and_Files'></a>
+### Dirs and Files
+
+### Rakefile
+
+Inside the Rakefile in your project's root dir, make sure you have:
+
+``` ruby
+require 'mobilize-base/tasks'
+require 'mobilize-ssh/tasks'
+require 'mobilize-hdfs/tasks'
+```
+
+This defines rake tasks essential to run the environment.
+
+### Config Dir
+
+run
+
+  $ rake mobilize_hdfs:setup
+
+This will copy over a sample hadoop.yml to your config dir.
+
+<a name='section_Configure'></a>
+Configure
+------------
+
+<a name='section_Configure_Hadoop'></a>
+### Configure Hadoop
+
+* Hadoop is big data. That means we need to be careful when reading from
+the cluster as it could easily fill up our mongodb instance, RAM, local disk
+space, etc.
+* To achieve this, all hadoop operations, stage outputs, etc. are
+executed and stored on the cluster only. 
+  * The exceptions are:
+    * writing to the cluster from an external source, such as a google
+sheet. Here there
+is no risk as the external source has much more strict size limits than
+hdfs.
+    * reading from the cluster, such as for posting to google sheet. In
+this case, the read_limit parameter dictates the maximum amount that can
+be read. If the data is bigger than the read limit, an exception will be
+raised.
+
+The Hadoop configuration consists of:
+* output_dir, which is the absolute path to the directory in HDFS that will store stage
+outputs. Directory names should end with a slash (/). It will choose the
+first cluster as the default cluster to write to.
+* read_limit, which is the maximum size data that can be read from the
+cluster. Default is 1GB.
+* clusters - this defines aliases for clusters, which are used as
+parameters for Hdfs stages. Cluster aliases contain 5 parameters:
+  * namenode - defines the name and port for accessing the namenode
+    * name - namenode full name, as in namenode1.host.com
+    * port - namenode port, by default 50070
+  * gateway_node - defines the node that executes the cluster commands.
+  * exec_path - defines the path to the hadoop 
+This node must be defined in ssh.yml according to the specs in
+[mobilize-ssh][mobilize-ssh]. The gateway node can be the same for
+multiple clusters, depending on your cluster setup.
+
+Sample hadoop.yml:
+
+``` yml
+---
+development:
+  output_dir: /user/mobilize/development/
+  read_limit: 1000000000
+  clusters:
+    dev_cluster:
+      namenode:
+        name: dev_namenode.host.com
+        port: 50070
+      gateway_node: dev_hadoop_host
+      exec_path: /path/to/hadoop
+    dev_cluster_2:
+      namenode:
+        name: dev_namenode_2.host.com
+        port: 50070
+      gateway_node: dev_hadoop_host
+      exec_path: /path/to/hadoop
+test:
+  output_dir: /user/mobilize/test/
+  read_limit: 1000000000
+  clusters:
+    test_cluster:
+      namenode:
+        name: test_namenode.host.com
+        port: 50070
+      gateway_node: test_hadoop_host
+      exec_path: /path/to/hadoop
+    test_cluster_2:
+      namenode:
+        name: test_namenode_2.host.com
+        port: 50070
+      gateway_node: test_hadoop_host
+      exec_path: /path/to/hadoop
+production:
+  output_dir: /user/mobilize/production/
+  read_limit: 1000000000
+  clusters:
+    prod_cluster:
+      namenode:
+        name: prod_namenode.host.com
+        port: 50070
+      gateway_node: prod_hadoop_host
+      exec_path: /path/to/hadoop
+    prod_cluster_2:
+      namenode:
+        name: prod_namenode_2.host.com
+        port: 50070
+      gateway_node: prod_hadoop_host
+      exec_path: /path/to/hadoop
+```
+
+<a name='section_Start'></a>
+Start
+-----
+
+<a name='section_Start_Create_Job'></a>
+### Create Job
+
+* For mobilize-hdfs, the following stages are available. 
+  * cluster and user are optional for all of the below.
+    * cluster defaults to output_cluster;
+    * user is treated the same way as in [mobilize-ssh][mobilize-ssh].
+  * hdfs.write `source:<full_path>, target:<hdfs_full_path>, user:<user>` 
+  * The full_path can use `<gsheet_path>` or `<hdfs_path>`. The test uses "test_hdfs_1.in".
+  * `<hdfs_path>` is the cluster alias followed by absolute path on the cluster.
+    * if a full path is supplied without a preceding cluster alias (e.g. "/user/mobilize/test/test_hdfs_1.in"), 
+      the first listed cluster will be used as the default.
+    * The test uses "/user/mobilize/test/test_hdfs_1.in" for the initial
+write, then "test_cluster_2/user/mobilize/test/test_hdfs_copy.out" for
+the cross-cluster write.
+  * both cluster arguments and user are optional. If writing from
+one cluster to another, your source_cluster gateway_node must be able to
+access both clusters.
+
+<a name='section_Start_Run_Test'></a>
+### Run Test
+
+To run tests, you will need to 
+
+1) go through the [mobilize-base][mobilize-base] and [mobilize-ssh][mobilize-ssh] tests first
+
+2) clone the mobilize-hdfs repository 
+
+From the project folder, run
+
+3) $ rake mobilize_hdfs:setup
+
+Copy over the config files from the mobilize-base and mobilize-ssh
+projects into the config dir, and populate the values in the hadoop.yml file.
+
+If you don't have two clusters, you can populate test_cluster_2 with the
+same cluster as your first.
+
+3) $ rake test
+
+* The test runs a 3 stage job:
+  * test_hdfs_1:
+    * `hdfs.write target:"/user/mobilize/test/test_hdfs_1.out", source:"test_hdfs_1.in"`
+    * `hdfs.write source:"/user/mobilize/test/test_hdfs_1.out",target:"test_cluster_2/user/mobilize/test/test_hdfs_1_copy.out"`
+    * `gsheet.write source:"hdfs://test_cluster_2/user/mobilize/test/test_hdfs_1_copy.out", target:"test_hdfs_1_copy.out"`
+  * at the end of the test, there should be a sheet named "test_hdfs_1_copy.out" with the same data as test_hdfs_1.in
+
+<a name='section_Meta'></a>
+Meta
+----
+
+* Code: `git clone git://github.com/dena/mobilize-hdfs.git`
+* Home: <https://github.com/dena/mobilize-hdfs>
+* Bugs: <https://github.com/dena/mobilize-hdfs/issues>
+* Gems: <http://rubygems.org/gems/mobilize-hdfs>
+
+<a name='section_Author'></a>
+Author
+------
+
+Cassio Paes-Leme :: cpaesleme@dena.com :: @cpaesleme
+
+[mobilize-base]: https://github.com/dena/mobilize-base
+[mobilize-ssh]: https://github.com/dena/mobilize-ssh

data/lib/mobilize-hdfs/handlers/hdfs.rb

@@ -127,7 +127,17 @@ module Mobilize
         path = path.starts_with?("/") ? path : "/#{path}"
       end
       url = "hdfs://#{cluster}#{path}"
-      return url
+      hdfs_url = Hdfs.hdfs_url(url)
+      begin
+        response = Hadoop.run(cluster, "fs -tail '#{hdfs_url}'", user_name)
+        if response['exit_code']==0 or is_target
+          return "hdfs://#{cluster}#{path}"
+        else
+          raise "Unable to find #{url} with error: #{response['stderr']}"
+        end
+      rescue => exc
+        raise Exception, "Unable to find #{url} with error: #{exc.to_s}", exc.backtrace
+      end
     end
 
     def Hdfs.user_name_by_stage_path(stage_path,cluster=nil)

data/lib/mobilize-hdfs/tasks.rb

@@ -1,4 +1,3 @@
-require 'yaml'
 namespace :mobilize_hdfs do
   desc "Set up config and log folders and files"
   task :setup do

data/lib/mobilize-hdfs/version.rb

@@ -1,5 +1,5 @@
 module Mobilize
   module Hdfs
-    VERSION = "1.36"
+    VERSION = "1.291"
   end
 end

data/lib/mobilize-hdfs.rb

@@ -3,9 +3,6 @@ require "mobilize-ssh"
 
 module Mobilize
   module Hdfs
-    def Hdfs.home_dir
-      File.expand_path('..',File.dirname(__FILE__))
-    end
   end
 end
 require "mobilize-hdfs/handlers/hadoop"

data/mobilize-hdfs.gemspec

@@ -16,5 +16,5 @@ Gem::Specification.new do |gem|
   gem.executables   = gem.files.grep(%r{^bin/}).map{ |f| File.basename(f) }
   gem.test_files    = gem.files.grep(%r{^(test|spec|features)/})
   gem.require_paths = ["lib"]
-  gem.add_runtime_dependency "mobilize-ssh","1.36"
+  gem.add_runtime_dependency "mobilize-ssh","1.291"
 end

data/test/hdfs_job_rows.yml

@@ -0,0 +1,10 @@
+- name: test_hdfs_1
+  active: true
+  trigger: once
+  status: ""
+  stage1: hdfs.write target:"/user/mobilize/test/test_hdfs_1.out",
+            source:"test_hdfs_1.in"
+  stage2: hdfs.write source:"/user/mobilize/test/test_hdfs_1.out",
+            target:"test_cluster_2/user/mobilize/test/test_hdfs_1_copy.out",
+  stage3: gsheet.write source:"hdfs://test_cluster_2/user/mobilize/test/test_hdfs_1_copy.out",
+            target:"Runner_mobilize(test)/test_hdfs_1_copy.out"

data/test/mobilize-hdfs_test.rb

@@ -0,0 +1,70 @@
+require 'test_helper'
+
+describe "Mobilize" do
+
+  def before
+    puts 'nothing before'
+  end
+
+  # enqueues 4 workers on Resque
+  it "runs integration test" do
+
+    puts "restart workers"
+    Mobilize::Jobtracker.restart_workers!
+
+    gdrive_slot = Mobilize::Gdrive.owner_email
+    puts "create user 'mobilize'"
+    user_name = gdrive_slot.split("@").first
+    u = Mobilize::User.where(:name=>user_name).first
+    r = u.runner
+    hdfs_1_sheet = Mobilize::Gsheet.find_by_path("#{r.path.split("/")[0..-2].join("/")}/test_hdfs_1.in",gdrive_slot)
+    [hdfs_1_sheet].each {|s| s.delete if s}
+
+    puts "add test_source data"
+    hdfs_1_sheet = Mobilize::Gsheet.find_or_create_by_path("#{r.path.split("/")[0..-2].join("/")}/test_hdfs_1.in",gdrive_slot)
+    hdfs_1_tsv = ([%w{test0 test1 test2 test3 test4 test5 test6 test7 test8 test9}.join("\t")]*10).join("\n")
+    hdfs_1_sheet.write(hdfs_1_tsv,u.name)
+
+    jobs_sheet = r.gsheet(gdrive_slot)
+
+    test_job_rows = ::YAML.load_file("#{Mobilize::Base.root}/test/hdfs_job_rows.yml")
+    test_job_rows.map{|j| r.jobs(j['name'])}.each{|j| j.delete if j}
+    jobs_sheet.add_or_update_rows(test_job_rows)
+
+    hdfs_1_target_sheet = Mobilize::Gsheet.find_by_path("#{r.path.split("/")[0..-2].join("/")}/test_hdfs_1_copy.out",gdrive_slot)
+    [hdfs_1_target_sheet].each {|s| s.delete if s}
+
+    puts "job row added, force enqueued requestor, wait for stages"
+    r.enqueue!
+    wait_for_stages
+
+    puts "jobtracker posted data to test sheet"
+    test_destination_sheet = Mobilize::Gsheet.find_by_path("#{r.path.split("/")[0..-2].join("/")}/test_hdfs_1_copy.out",gdrive_slot)
+
+    assert test_destination_sheet.read(u.name).length == 599
+  end
+
+  def wait_for_stages(time_limit=600,stage_limit=120,wait_length=10)
+    time = 0
+    time_since_stage = 0
+    #check for 10 min
+    while time < time_limit and time_since_stage < stage_limit
+      sleep wait_length
+      job_classes = Mobilize::Resque.jobs.map{|j| j['class']}
+      if job_classes.include?("Mobilize::Stage")
+        time_since_stage = 0
+        puts "saw stage at #{time.to_s} seconds"
+      else
+        time_since_stage += wait_length
+        puts "#{time_since_stage.to_s} seconds since stage seen"
+      end
+      time += wait_length
+      puts "total wait time #{time.to_s} seconds"
+    end
+
+    if time >= time_limit
+      raise "Timed out before stage completion"
+    end
+  end  
+
+end

data/test/test_helper.rb

@@ -8,4 +8,3 @@ $dir = File.dirname(File.expand_path(__FILE__))
 ENV['MOBILIZE_ENV'] = 'test'
 require 'mobilize-hdfs'
 $TESTING = true
-require "#{Mobilize::Ssh.home_dir}/test/test_helper"

metadata

@@ -1,32 +1,29 @@
 --- !ruby/object:Gem::Specification
 name: mobilize-hdfs
 version: !ruby/object:Gem::Version
-  version: '1.36'
-  prerelease: 
+  version: '1.291'
 platform: ruby
 authors:
 - Cassio Paes-Leme
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2013-05-21 00:00:00.000000000 Z
+date: 2013-03-27 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: mobilize-ssh
   requirement: !ruby/object:Gem::Requirement
-    none: false
     requirements:
     - - '='
       - !ruby/object:Gem::Version
-        version: '1.36'
+        version: '1.291'
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
-    none: false
     requirements:
     - - '='
       - !ruby/object:Gem::Version
-        version: '1.36'
+        version: '1.291'
 description: Adds hdfs read, write, and copy support to mobilize-ssh
 email:
 - cpaesleme@dena.com
@@ -46,46 +43,35 @@ files:
 - lib/mobilize-hdfs/version.rb
 - lib/samples/hadoop.yml
 - mobilize-hdfs.gemspec
-- test/fixtures/hdfs1.in.yml
-- test/fixtures/integration_expected.yml
-- test/fixtures/integration_jobs.yml
-- test/integration/mobilize-hdfs_test.rb
+- test/hdfs_job_rows.yml
+- test/mobilize-hdfs_test.rb
 - test/redis-test.conf
 - test/test_helper.rb
 homepage: http://github.com/dena/mobilize-hdfs
 licenses: []
+metadata: {}
 post_install_message: 
 rdoc_options: []
 require_paths:
 - lib
 required_ruby_version: !ruby/object:Gem::Requirement
-  none: false
   requirements:
-  - - ! '>='
+  - - '>='
     - !ruby/object:Gem::Version
       version: '0'
-      segments:
-      - 0
-      hash: 3270257259800653106
 required_rubygems_version: !ruby/object:Gem::Requirement
-  none: false
   requirements:
-  - - ! '>='
+  - - '>='
     - !ruby/object:Gem::Version
       version: '0'
-      segments:
-      - 0
-      hash: 3270257259800653106
 requirements: []
 rubyforge_project: 
-rubygems_version: 1.8.25
+rubygems_version: 2.0.3
 signing_key: 
-specification_version: 3
+specification_version: 4
 summary: Adds hdfs read, write, and copy support to mobilize-ssh
 test_files:
-- test/fixtures/hdfs1.in.yml
-- test/fixtures/integration_expected.yml
-- test/fixtures/integration_jobs.yml
-- test/integration/mobilize-hdfs_test.rb
+- test/hdfs_job_rows.yml
+- test/mobilize-hdfs_test.rb
 - test/redis-test.conf
 - test/test_helper.rb

data/test/fixtures/hdfs1.in.yml

@@ -1,91 +0,0 @@
----
-- test0: test0
-  test1: test1
-  test2: test2
-  test3: test3
-  test4: test4
-  test5: test5
-  test6: test6
-  test7: test7
-  test8: test8
-  test9: test9
-- test0: test0
-  test1: test1
-  test2: test2
-  test3: test3
-  test4: test4
-  test5: test5
-  test6: test6
-  test7: test7
-  test8: test8
-  test9: test9
-- test0: test0
-  test1: test1
-  test2: test2
-  test3: test3
-  test4: test4
-  test5: test5
-  test6: test6
-  test7: test7
-  test8: test8
-  test9: test9
-- test0: test0
-  test1: test1
-  test2: test2
-  test3: test3
-  test4: test4
-  test5: test5
-  test6: test6
-  test7: test7
-  test8: test8
-  test9: test9
-- test0: test0
-  test1: test1
-  test2: test2
-  test3: test3
-  test4: test4
-  test5: test5
-  test6: test6
-  test7: test7
-  test8: test8
-  test9: test9
-- test0: test0
-  test1: test1
-  test2: test2
-  test3: test3
-  test4: test4
-  test5: test5
-  test6: test6
-  test7: test7
-  test8: test8
-  test9: test9
-- test0: test0
-  test1: test1
-  test2: test2
-  test3: test3
-  test4: test4
-  test5: test5
-  test6: test6
-  test7: test7
-  test8: test8
-  test9: test9
-- test0: test0
-  test1: test1
-  test2: test2
-  test3: test3
-  test4: test4
-  test5: test5
-  test6: test6
-  test7: test7
-  test8: test8
-  test9: test9
-- test0: test0
-  test1: test1
-  test2: test2
-  test3: test3
-  test4: test4
-  test5: test5
-  test6: test6
-  test7: test7
-  test8: test8
-  test9: test9

data/test/fixtures/integration_expected.yml

@@ -1,17 +0,0 @@
----
-- path: "Runner_mobilize(test)/jobs"
-  state: working
-  count: 1
-  confirmed_ats: []
-- path: "Runner_mobilize(test)/jobs/hdfs1/stage1"
-  state: working
-  count: 1
-  confirmed_ats: []
-- path: "Runner_mobilize(test)/jobs/hdfs1/stage2"
-  state: working
-  count: 1
-  confirmed_ats: []
-- path: "Runner_mobilize(test)/jobs/hdfs1/stage3"
-  state: working
-  count: 1
-  confirmed_ats: []

data/test/fixtures/integration_jobs.yml

@@ -1,10 +0,0 @@
-- name: hdfs1
-  active: true
-  trigger: once
-  status: ""
-  stage1: hdfs.write target:"/user/mobilize/test/hdfs1.out",
-            source:"hdfs1.in"
-  stage2: hdfs.write source:"/user/mobilize/test/hdfs1.out",
-            target:"test_cluster_2/user/mobilize/test/hdfs1_copy.out",
-  stage3: gsheet.write source:"hdfs://test_cluster_2/user/mobilize/test/hdfs1_copy.out",
-            target:"Runner_mobilize(test)/hdfs1_copy.out"

data/test/integration/mobilize-hdfs_test.rb

@@ -1,42 +0,0 @@
-require 'test_helper'
-describe "Mobilize" do
-  # enqueues 4 workers on Resque
-  it "runs integration test" do
-
-    puts "restart workers"
-    Mobilize::Jobtracker.restart_workers!
-
-    u = TestHelper.owner_user
-    r = u.runner
-    user_name = u.name
-    gdrive_slot = u.email
-
-    puts "add test data"
-    ["hdfs1.in"].each do |fixture_name|
-      target_url = "gsheet://#{r.title}/#{fixture_name}"
-      TestHelper.write_fixture(fixture_name, target_url, 'replace')
-    end
-
-    puts "add/update jobs"
-    u.jobs.each{|j| j.delete}
-    jobs_fixture_name = "integration_jobs"
-    jobs_target_url = "gsheet://#{r.title}/jobs"
-    TestHelper.write_fixture(jobs_fixture_name, jobs_target_url, 'update')
-
-    puts "job rows added, force enqueue runner, wait for stages"
-    #wait for stages to complete
-    expected_fixture_name = "integration_expected"
-    Mobilize::Jobtracker.stop!
-    r.enqueue!
-    TestHelper.confirm_expected_jobs(expected_fixture_name)
-
-    puts "update job status and activity"
-    r.update_gsheet(gdrive_slot)
-
-    puts "check posted data"
-    ['hdfs1_copy.out'].each do |out_name|
-      url = "gsheet://#{r.title}/#{out_name}"
-      assert TestHelper.check_output(url, 'min_length' => 599) == true
-    end
-  end
-end