Merge pull request #3148 from betagouv/frederic/fix_2180/speed_up_migration

#2180 speed up migration

lib/active_storage/service/cellar_service.rb

@@ -37,7 +37,7 @@ module ActiveStorage
     def delete_prefixed(prefix)
       instrument :delete_prefixed, prefix: prefix do
         @adapter.session do |s|
-          keys = s.list_prefixed(prefix)
+          keys = s.list_prefixed(prefix).map(&:first)
           s.delete_keys(keys)
         end
       end

lib/cellar/cellar_adapter.rb

@@ -91,7 +91,7 @@ module Cellar
           if response.is_a?(Net::HTTPSuccess)
             (listing, truncated) = parse_bucket_listing(response.body)
             result += listing
-            marker = listing.last
+            marker = listing.last.first
           else
             # TODO: error handling
             return nil
@@ -139,8 +139,13 @@ module Cellar
       def parse_bucket_listing(bucket_listing_xml)
         doc = Nokogiri::XML(bucket_listing_xml)
         listing = doc
-          .xpath('//xmlns:Contents/xmlns:Key')
-          .map(&:text)
+          .xpath('//xmlns:Contents')
+          .map do |node|
+          [
+            node.xpath('xmlns:Key').text,
+            DateTime.iso8601(node.xpath('xmlns:LastModified').text)
+          ]
+        end
         truncated = doc.xpath('//xmlns:IsTruncated').text == 'true'
         [listing, truncated]
       end

lib/tasks/2018_12_03_finish_piece_jointe_transfer.rake

@@ -67,23 +67,32 @@ namespace :'2018_12_03_finish_piece_jointe_transfer' do
       # This task ports them to the new storage after the switch, while being careful not to
       # overwrite attachments that may have changed in the new storage after the switch.
       def refresh_outdated_files
-        rake_puts "Refresh outdated attachments"
-
         refreshed_keys = []
-        missing_keys = []
         old_pj_adapter.session do |old_pjs|
-          keys = old_pjs.list_prefixed('')
-          progress = ProgressReport.new(keys.count)
-          keys.each do |key|
-            new_pj_metadata = new_pjs.files.head(key)
+          rake_puts "List old PJs"
+          old_pj_listing = old_pjs.list_prefixed('')
 
-            refresh_needed = new_pj_metadata.nil?
-            if !refresh_needed
-              new_pj_last_modified = new_pj_metadata.last_modified.in_time_zone
-              old_pj_last_modified = old_pjs.last_modified(key)
-              if old_pj_last_modified.nil?
-                missing_keys.push(key)
-              else
+          rake_puts "List new PJs"
+          new_pj_listing = {}
+          progress = ProgressReport.new(new_pjs.count.to_i)
+          new_pjs.files.each do |f|
+            new_pj_listing[f.key] = f.last_modified.in_time_zone
+            progress.inc
+          end
+          progress.finish
+
+          rake_puts "Refresh outdated attachments"
+          progress = ProgressReport.new(old_pj_listing.count)
+          old_pj_listing.each do |key, old_pj_last_modified|
+            new_pj_last_modified = new_pj_listing[key]
+
+            if new_pj_last_modified.nil? || new_pj_last_modified < old_pj_last_modified
+              # Looks like we need to refresh  this PJ.
+              # Fetch fresh metadata to avoid overwriting a last-minute change
+              new_pj_metadata = new_pjs.files.head(key)
+              refresh_needed = new_pj_metadata.nil?
+              if !refresh_needed
+                new_pj_last_modified = new_pj_metadata.last_modified.in_time_zone
                 refresh_needed = new_pj_last_modified < old_pj_last_modified
               end
             end
@@ -114,9 +123,6 @@ namespace :'2018_12_03_finish_piece_jointe_transfer' do
         if verbose?
           rake_puts "Refreshed #{refreshed_keys.count} attachments\n#{refreshed_keys.join(', ')}"
         end
-        if missing_keys.present?
-          rake_puts "Failed to refresh #{missing_keys.count} attachments\n#{missing_keys.join(', ')}"
-        end
       end
 
       # For OpenStack, the content type cannot be forced dynamically from a direct download URL.

spec/lib/cellar/cellar_adapter_spec.rb

@@ -43,7 +43,7 @@ describe 'CellarAdapter' do
           </Contents>
           <Contents>
             <Key>sample2.jpg</Key>
-            <LastModified>2011-02-26T01:56:20.000Z</LastModified>
+            <LastModified>2014-03-21T17:44:07.000Z</LastModified>
             <ETag>&quot;bf1d737a4d46a19f3bced6905cc8b902&quot;</ETag>
             <Size>142863</Size>
             <StorageClass>STANDARD</StorageClass>
@@ -54,7 +54,17 @@ describe 'CellarAdapter' do
 
     subject { session.send(:parse_bucket_listing, response) }
 
-    it { is_expected.to eq([["sample1.jpg", "sample2.jpg"], false]) }
+    it do
+      is_expected.to eq(
+        [
+          [
+            ["sample1.jpg", DateTime.new(2011, 2, 26, 1, 56, 20, 0)],
+            ["sample2.jpg", DateTime.new(2014, 3, 21, 17, 44, 7, 0)]
+          ],
+          false
+        ]
+      )
+    end
   end
 
   describe 'bulk_deletion_request_body' do