Merge branch '15093-move-collection-update-out-of-band' into 1.4-dev

[arvados.git] / services / api / db / migrate / 20190322174136_add_file_info_to_collection.rb

diff --git a/services/api/db/migrate/20190322174136_add_file_info_to_collection.rb b/services/api/db/migrate/20190322174136_add_file_info_to_collection.rb
old mode 100755 (executable)
new mode 100644 (file)
index 61f9b2d..c0cd40d
--- a/services/api/db/migrate/20190322174136_add_file_info_to_collection.rb
+++ b/services/api/db/migrate/20190322174136_add_file_info_to_collection.rb
@@ -2,58 +2,16 @@
 #
 # SPDX-License-Identifier: AGPL-3.0
 
-require "arvados/keep"
-require "group_pdhs"
-
 class AddFileInfoToCollection < ActiveRecord::Migration[4.2]
-  def do_batch(pdhs)
-    pdhs_str = ''
-    pdhs.each do |pdh|
-      pdhs_str << "'" << pdh << "'" << ","
-    end
-
-    collections = ActiveRecord::Base.connection.exec_query(
-      "SELECT DISTINCT portable_data_hash, manifest_text FROM collections "\
-      "WHERE portable_data_hash IN (#{pdhs_str[0..-2]}) "
-    )
-
-    collections.rows.each do |row|
-      manifest = Keep::Manifest.new(row[1])
-      ActiveRecord::Base.connection.exec_query("BEGIN")
-      ActiveRecord::Base.connection.exec_query("UPDATE collections SET file_count=#{manifest.files_count}, "\
-                                               "file_size_total=#{manifest.files_size} "\
-                                               "WHERE portable_data_hash='#{row[0]}'")
-      ActiveRecord::Base.connection.exec_query("COMMIT")
-    end
-  end
-
   def up
     add_column :collections, :file_count, :integer, default: 0, null: false
     add_column :collections, :file_size_total, :integer, limit: 8, default: 0, null: false
 
-    distinct_pdh_count = ActiveRecord::Base.connection.exec_query(
-      "SELECT DISTINCT portable_data_hash FROM collections"
-    ).rows.count
-
-    # Generator that queries for all the distinct pdhs greater than last_pdh
-    ordered_pdh_query = lambda { |last_pdh, &block|
-      pdhs = ActiveRecord::Base.connection.exec_query(
-        "SELECT DISTINCT portable_data_hash FROM collections "\
-        "WHERE portable_data_hash > '#{last_pdh}' "\
-        "ORDER BY portable_data_hash LIMIT 1000"
-      )
-      pdhs.rows.each do |row|
-        block.call(row[0])
-      end
-    }
-
-    batch_size_max = 1 << 28 # 256 MiB
-    GroupPdhs.group_pdhs_for_multiple_transactions(ordered_pdh_query,
-                                                   distinct_pdh_count,
-                                                   batch_size_max,
-                                                   "AddFileInfoToCollection") do |pdhs|
-      do_batch(pdhs)
-    end
+    puts "Collections now have two new columns, file_count and file_size_total."
+    puts "They were initialized with a zero value. If you are upgrading an Arvados"
+    puts "installation, please run the populate-file-info-columns-in-collections.rb"
+    puts "script to populate the columns. If this is a new installation, that is not"
+    puts "necessary."
   end
 
   def down