Merge 4b5c471 into 1cdd043

.rubocop.yml

@@ -179,7 +179,7 @@ RSpec/NamedSubject:
 RSpec/NestedGroups:
   Max: 4 # default: 3
   Exclude:
-    - 'spec/lib/audit/catalog_to_moab_instance_spec.rb'
+    - 'spec/lib/audit/catalog_to_moab_spec.rb'
     - 'spec/lib/audit/moab_to_catalog_spec.rb'
     - 'spec/services/checksum_validator_spec.rb'
     - 'spec/services/preserved_object_handler_check_exist_spec.rb'

app/jobs/catalog_to_moab_job.rb

@@ -0,0 +1,16 @@
+# Check filesystem based on catalog, updating database
+# @see Audit::CatalogToMoab
+class CatalogToMoabJob < ApplicationJob
+  queue_as :c2m
+
+  before_enqueue do |job|
+    raise ArgumentError, 'CompleteMoab param required' unless job.arguments.first.is_a?(CompleteMoab)
+  end
+
+  # @param [CompleteMoab] complete_moab object to checksum
+  # @param [String] storage_dir
+  # @see Audit::CatalogToMoab#initialize
+  def perform(complete_moab, storage_dir)
+    Audit::CatalogToMoab.new(complete_moab, storage_dir).check_catalog_version
+  end
+end

app/lib/audit/catalog_to_moab.rb

@@ -1,49 +1,7 @@
 module Audit
   # Catalog to Moab existence check code
   class CatalogToMoab
-
-    def self.logger
-      @logger ||= Logger.new(STDOUT)
-                        .extend(ActiveSupport::Logger.broadcast(Logger.new(Rails.root.join('log', 'c2m.log'))))
-    end
-
-    def self.check_version_on_dir(last_checked_b4_date, storage_dir, limit=Settings.c2m_sql_limit)
-      logger.info "#{Time.now.utc.iso8601} C2M check_version starting for #{storage_dir}"
-
-      # cms_to_audit_relation is an AR Relation; it could return a lot of results, so we want to process in batches.
-      # We can't use ActiveRecord's .find_each, because that'll disregard the order .least_recent_version_audit
-      # specified.  so we use our own batch processing method, which does respect Relation order.
-      cms_to_audit_relation =
-        CompleteMoab.least_recent_version_audit(last_checked_b4_date).by_storage_location(storage_dir)
-      ActiveRecordUtils.process_in_batches(cms_to_audit_relation, limit) do |cm|
-        c2m = CatalogToMoab.new(cm, storage_dir)
-        c2m.check_catalog_version
-      end
-    ensure
-      logger.info "#{Time.now.utc.iso8601} C2M check_version ended for #{storage_dir}"
-    end
-
-    def self.check_version_on_dir_profiled(last_checked_b4_date, storage_dir)
-      Profiler.print_profile('C2M_check_version_on_dir') { check_version_on_dir(last_checked_b4_date, storage_dir) }
-    end
-
-    def self.check_version_all_dirs(last_checked_b4_date)
-      logger.info "#{Time.now.utc.iso8601} C2M check_version_all_dirs starting"
-      HostSettings.storage_roots.to_h.each_value do |strg_root_location|
-        check_version_on_dir(last_checked_b4_date, "#{strg_root_location}/#{Settings.moab.storage_trunk}")
-      end
-    ensure
-      logger.info "#{Time.now.utc.iso8601} C2M check_version_all_dirs ended"
-    end
-
-    def self.check_version_all_dirs_profiled(last_checked_b4_date)
-      Profiler.print_profile('C2M_check_version_all_dirs') { check_version_all_dirs(last_checked_b4_date) }
-    end
-
-    # ----  INSTANCE code below this line ---------------------------
-
     include ::MoabValidationHandler
-
     attr_reader :complete_moab, :storage_dir, :druid, :results
 
     def initialize(complete_moab, storage_dir)
@@ -53,14 +11,18 @@ def initialize(complete_moab, storage_dir)
       @results = AuditResults.new(druid, nil, complete_moab.moab_storage_root)
     end
 
+    def logger
+      @logger ||= Logger.new(Rails.root.join('log', 'c2m.log'))
+    end
+
     # shameless green implementation
     def check_catalog_version
       results.check_name = 'check_catalog_version'
       unless complete_moab.matches_po_current_version?
         results.add_result(AuditResults::CM_PO_VERSION_MISMATCH,
                            cm_version: complete_moab.version,
                            po_version: complete_moab.preserved_object.current_version)
-        return results.report_results(Audit::CatalogToMoab.logger)
+        return results.report_results(logger)
       end
 
       unless online_moab_found?
@@ -73,10 +35,10 @@ def check_catalog_version
         results.add_result(AuditResults::MOAB_NOT_FOUND,
                            db_created_at: complete_moab.created_at.iso8601,
                            db_updated_at: complete_moab.updated_at.iso8601)
-        return results.report_results(Audit::CatalogToMoab.logger)
+        return results.report_results(logger)
       end
 
-      return results.report_results(Audit::CatalogToMoab.logger) unless can_validate_current_comp_moab_status?
+      return results.report_results(logger) unless can_validate_current_comp_moab_status?
 
       compare_version_and_take_action
     end
@@ -100,7 +62,7 @@ def compare_version_and_take_action
         if catalog_version == moab_version
           set_status_as_seen_on_disk(true) unless complete_moab.ok?
           results.add_result(AuditResults::VERSION_MATCHES, 'CompleteMoab')
-          results.report_results(Audit::CatalogToMoab.logger)
+          results.report_results(logger)
         elsif catalog_version < moab_version
           set_status_as_seen_on_disk(true)
           pohandler = PreservedObjectHandler.new(druid, moab_version, moab.size, complete_moab.moab_storage_root)
@@ -110,7 +72,7 @@ def compare_version_and_take_action
           results.add_result(
             AuditResults::UNEXPECTED_VERSION, db_obj_name: 'CompleteMoab', db_obj_version: complete_moab.version
           )
-          results.report_results(Audit::CatalogToMoab.logger)
+          results.report_results(logger)
         end
 
         complete_moab.update_audit_timestamps(ran_moab_validation?, true)

app/models/complete_moab.rb

@@ -43,9 +43,6 @@ class CompleteMoab < ApplicationRecord
 
   scope :least_recent_version_audit, lambda { |last_checked_b4_date|
     where('last_version_audit IS NULL or last_version_audit < ?', normalize_date(last_checked_b4_date))
-
-    # possibly counter-intuitive: the .order sorts so that null values come first (because IS NOT NULL evaluates
-    # to 0 for nulls, which sorts before 1 for non-nulls, which are then sorted by last_version_audit)
   }
 
   scope :fixity_check_expired, lambda {
@@ -56,8 +53,6 @@ class CompleteMoab < ApplicationRecord
         ' AND (last_checksum_validation + (fixity_ttl * INTERVAL \'1 SECOND\')) < CURRENT_TIMESTAMP'\
         ' OR last_checksum_validation IS NULL'
       )
-    # possibly counter-intuitive: the .order sorts so that null values come first (because IS NOT NULL evaluates
-    # to 0 for nulls, which sorts before 1 for non-nulls, which are then sorted by last_checksum_validation)
   }
 
   # This is where we make sure we have ZMV rows for all needed ZipEndpoints and versions.

app/models/moab_storage_root.rb

@@ -14,6 +14,13 @@ def validate_expired_checksums!
     cms.find_each { |cm| ChecksumValidationJob.perform_later(cm) }
   end
 
+  # Use a queue to check all associated CompleteMoab objects for C2M
+  def c2m_check!(last_checked_b4_date = Time.current)
+    complete_moabs.least_recent_version_audit(last_checked_b4_date).find_each do |cm|
+      CatalogToMoabJob.perform_later(cm, storage_location)
+    end
+  end
+
   # Iterates over the storage roots enumerated in settings, creating a MoabStorageRoot for
   #   each if it doesn't already exist.
   # @param preservation_policies [Enumerable<PreservationPolicy>] the list of preservation policies

lib/tasks/c2m_tasks.rake

@@ -1,43 +1,23 @@
 namespace :c2m do
-
   desc "Run C2M version checks on a single storage root"
-  task :one_root, [:last_checked_b4_date, :storage_root, :profile] => [:environment] do |_t, args|
-    unless args[:profile] == 'profile' || args[:profile].nil?
-      p "usage: rake c2m:one_root[last_checked_b4_date, fixture_sr1] || rake c2m:one_root[last_checked_b4_date,fixture_sr1,profile]"
-      exit 1
-    end
-    root = args[:storage_root].to_sym
-    storage_dir = "#{HostSettings.storage_roots[root]}/#{Settings.moab.storage_trunk}"
+  task :one_root, [:last_checked_b4_date, :storage_root] => [:environment] do |_t, args|
+    root_key = args[:storage_root].to_sym
     last_checked = args[:last_checked_b4_date].to_s
+    storage_dir = "#{HostSettings.storage_roots[root_key]}/#{Settings.moab.storage_trunk}"
+    root = MoabStorageRoot.find_by!(storage_location: storage_dir)
     begin
-      if args[:profile] == 'profile'
-        puts "When done, check log/profile_C2M_check_version_on_dir[TIMESTAMP].txt for profiling details"
-        Audit::CatalogToMoab.check_version_on_dir_profiled(last_checked, storage_dir)
-      elsif args[:profile].nil?
-        Audit::CatalogToMoab.check_version_on_dir(last_checked, storage_dir)
-      end
-      puts "#{Time.now.utc.iso8601} Catalog to Moab version check on #{storage_dir} is done."
+      root.c2m_check!(last_checked_b4_date)
     rescue TypeError, ArgumentError
       p "You've entered an incorrect timestamp format #{last_checked}."
       p "Please enter correct timestamp format (UTC) (2018-02-01T18:54:48Z)"
     end
   end
 
   desc "Run C2M version checks on all storage roots"
-  task :all_roots, [:last_checked_b4_date, :profile] => [:environment] do |_t, args|
-    unless args[:profile] == 'profile' || args[:profile].nil?
-      p "usage: rake c2m:all_roots[last_checked_b4_date] || rake c2m:all_roots[last_checked_b4_date,profile]"
-      exit 1
-    end
+  task :all_roots, [:last_checked_b4_date] => [:environment] do |_t, args|
     last_checked = args[:last_checked_b4_date].to_s
     begin
-      if args[:profile] == 'profile'
-        puts "When done, check log/profile_C2M_check_version_all_roots[TIMESTAMP].txt for profiling details"
-        Audit::CatalogToMoab.check_version_all_dirs_profiled(last_checked)
-      elsif args[:profile].nil?
-        Audit::CatalogToMoab.check_version_all_dirs(last_checked)
-      end
-      puts "#{Time.now.utc.iso8601} Catalog to Moab version check on all roots is done."
+      MoabStorageRoot.find_each { |root| root.c2m_check!(last_checked) }
     rescue TypeError, ArgumentError
       p "You've entered an incorrect timestamp format #{last_checked}."
       p "Please enter correct timestamp format (UTC) (2018-02-01T18:54:48Z)"

spec/factories/preserved_object.rb

@@ -4,4 +4,26 @@
     current_version 1
     preservation_policy { PreservationPolicy.default_policy }
   end
+
+  # searches through fixture dirs to find the druid, creates a complete moab for the PO
+  factory :preserved_object_fixture, parent: :preserved_object do
+    current_version { Stanford::StorageServices.current_version(druid) }
+
+    after(:create) do |po|
+      locations = Settings.storage_root_map['default'].to_h.values.map { |x| File.join(x, Settings.moab.storage_trunk) }
+      root_dir = locations.find do |root|
+        found = false
+        Stanford::MoabStorageDirectory.find_moab_paths(root) do |druid, _path, _match|
+          found = true if druid && druid == po.druid
+        end
+        found
+      end
+      create_list(:complete_moab, 1,
+                  preserved_object: po,
+                  moab_storage_root: MoabStorageRoot.find_by!(storage_location: root_dir),
+                  version: po.current_version,
+                  size: Stanford::StorageServices.object_size(po.druid),
+                  status: 'validity_unknown')
+    end
+  end
 end

spec/jobs/catalog_to_moab_job_spec.rb

@@ -0,0 +1,17 @@
+require 'rails_helper'
+
+describe CatalogToMoabJob, type: :job do
+  let(:job) { described_class.new(cm) }
+  let(:cm) { create :complete_moab }
+  let(:storage_dir) { 'foobar' }
+
+  describe '#perform' do
+    let(:validator) { instance_double(Audit::CatalogToMoab) }
+
+    it 'calls Audit::CatalogToMoab#check_catalog_version' do
+      expect(validator).to receive(:check_catalog_version)
+      expect(Audit::CatalogToMoab).to receive(:new).with(cm, storage_dir).and_return(validator)
+      job.perform(cm, storage_dir)
+    end
+  end
+end