services/api/db/migrate/20180917205609_recompute_file_names_index.rb

   1 # Copyright (C) The Arvados Authors. All rights reserved.
   2 #
   3 # SPDX-License-Identifier: AGPL-3.0
   4
   5 class RecomputeFileNamesIndex < ActiveRecord::Migration[4.2]
   6   def do_batch(pdhs:)
   7     ActiveRecord::Base.connection.exec_query('BEGIN')
   8     Collection.select(:portable_data_hash, :manifest_text).where(portable_data_hash: pdhs).distinct(:portable_data_hash).each do |c|
   9       ActiveRecord::Base.connection.exec_query("update collections set file_names=$1 where portable_data_hash=$2",
  10                                                "update file_names index",
  11                                                [[nil, c.manifest_files], [nil, c.portable_data_hash]])
  12     end
  13     ActiveRecord::Base.connection.exec_query('COMMIT')
  14   end
  15   def up
  16     # Process collections in multiple transactions, where the total
  17     # size of all manifest_texts processed in a transaction is no more
  18     # than batch_size_max.  Collections whose manifest_text is bigger
  19     # than batch_size_max are updated in their own individual
  20     # transactions.
  21     batch_size_max = 1 << 28    # 256 MiB
  22     batch_size = 0
  23     batch_pdhs = {}
  24     last_pdh = '0'
  25     total = Collection.distinct.count(:portable_data_hash)
  26     done = 0
  27     any = true
  28     while any
  29       any = false
  30       Collection.
  31         unscoped.
  32         select(:portable_data_hash).distinct.
  33         order(:portable_data_hash).
  34         where('portable_data_hash > ?', last_pdh).
  35         limit(1000).each do |c|
  36         any = true
  37         last_pdh = c.portable_data_hash
  38         manifest_size = c.portable_data_hash.split('+')[1].to_i
  39         if batch_size > 0 && batch_size + manifest_size > batch_size_max
  40           do_batch(pdhs: batch_pdhs.keys)
  41           done += batch_pdhs.size
  42           Rails.logger.info("RecomputeFileNamesIndex: #{done}/#{total}")
  43           batch_pdhs = {}
  44           batch_size = 0
  45         end
  46         batch_pdhs[c.portable_data_hash] = true
  47         batch_size += manifest_size
  48       end
  49     end
  50     do_batch(pdhs: batch_pdhs.keys)
  51     Rails.logger.info("RecomputeFileNamesIndex: finished")
  52   end
  53   def down
  54   end
  55 end