From 8fba52b4a2045c67edde7520730cf0bafbb3f3f2 Mon Sep 17 00:00:00 2001
From: Martin Fenner <martin.fenner@datacite.org>
Date: Fri, 4 Jan 2019 02:37:55 +0100
Subject: [PATCH] index in batches of 500

---
 app/models/doi.rb               | 18 +++++++++---------
 config/initializers/_version.rb |  2 +-
 lib/tasks/doi.rake              |  2 +-
 3 files changed, 11 insertions(+), 11 deletions(-)

diff --git a/app/models/doi.rb b/app/models/doi.rb
index 90d736f92..e69ec7eec 100644
--- a/app/models/doi.rb
+++ b/app/models/doi.rb
@@ -523,12 +523,12 @@ def self.index_by_day(options={})
 
   def self.index_by_ids(options={})
     from_id = (options[:from_id] || 1).to_i
-    until_id = (options[:until_id] || from_id + 249).to_i
+    until_id = (options[:until_id] || from_id + 499).to_i
 
     # get every id between from_id and end_id
-    (from_id..until_id).step(250).each do |id|
+    (from_id..until_id).step(500).each do |id|
       DoiIndexByIdJob.perform_later(id: id)
-      puts "Queued indexing for DOIs with IDs #{from_id} - #{(until_id)}."
+      puts "Queued indexing for DOIs with IDs starting with #{id}."
     end
   end
 
@@ -541,7 +541,7 @@ def self.index_by_id(options={})
 
     logger = Logger.new(STDOUT)
 
-    Doi.where(id: id..(id + 249)).find_in_batches(batch_size: 250) do |dois|
+    Doi.where(id: id..(id + 499)).find_in_batches(batch_size: 500) do |dois|
       response = Doi.__elasticsearch__.client.bulk \
         index:   Doi.index_name,
         type:    Doi.document_type,
@@ -557,21 +557,21 @@ def self.index_by_id(options={})
     end
 
     if errors > 1
-      logger.error "[Elasticsearch] #{errors} errors indexing #{count} DOIs with IDs #{id} - #{(id + 249)}."
+      logger.error "[Elasticsearch] #{errors} errors indexing #{count} DOIs with IDs #{id} - #{(id + 499)}."
     elsif count > 1
-      logger.info "[Elasticsearch] Indexed #{count} DOIs with IDs #{id} - #{(id + 249)}."
+      logger.info "[Elasticsearch] Indexed #{count} DOIs with IDs #{id} - #{(id + 499)}."
     end
   rescue Elasticsearch::Transport::Transport::Errors::RequestEntityTooLarge, Faraday::ConnectionFailed, ActiveRecord::LockWaitTimeout => error
-    logger.info "[Elasticsearch] Error #{error.message} indexing DOIs with IDs #{id} - #{(id + 249)}."
+    logger.info "[Elasticsearch] Error #{error.message} indexing DOIs with IDs #{id} - #{(id + 499)}."
 
     count = 0
 
-    Doi.where(id: id..(id + 249)).find_each do |doi|
+    Doi.where(id: id..(id + 499)).find_each do |doi|
       IndexJob.perform_later(doi)
       count += 1
     end
 
-    logger.info "[Elasticsearch] Indexed #{count} DOIs with IDs #{id} - #{(id + 249)}."
+    logger.info "[Elasticsearch] Indexed #{count} DOIs with IDs #{id} - #{(id + 499)}."
   end
 
   def uid
diff --git a/config/initializers/_version.rb b/config/initializers/_version.rb
index cb765f20b..4836da4c5 100644
--- a/config/initializers/_version.rb
+++ b/config/initializers/_version.rb
@@ -1,5 +1,5 @@
 module Lupo
   class Application
-    VERSION = "2.2.4"
+    VERSION = "2.2.5"
   end
 end
\ No newline at end of file
diff --git a/lib/tasks/doi.rake b/lib/tasks/doi.rake
index 5e6d05698..b9c9e6f9d 100644
--- a/lib/tasks/doi.rake
+++ b/lib/tasks/doi.rake
@@ -77,7 +77,7 @@ namespace :doi do
   desc 'Index DOIs by ID'
   task :index_by_ids => :environment do
     from_id = (ENV['FROM_ID'] || 1).to_i
-    until_id = (ENV['UNTIL_ID'] || from_id + 249).to_i
+    until_id = (ENV['UNTIL_ID'] || from_id + 499).to_i
 
     Doi.index_by_ids(from_id: from_id, until_id: until_id)
   end