attachment gc improvements + support ContentMigration

fixes CORE-58

test plan:
- successfully deletes old ContentMigrations, including those tied to
old ContentExports

Change-Id: Ib9ca545da8a6929f495317bba08d0103acbaf1cc
Reviewed-on: https://gerrit.instructure.com/143330
Reviewed-by: Cody Cutrer <cody@instructure.com>
Tested-by: Jenkins
Product-Review: Simon Williams <simon@instructure.com>
QA-Review: Simon Williams <simon@instructure.com>

app/models/attachments/garbage_collector.rb

@@ -30,8 +30,8 @@ class Attachments::GarbageCollector
       to_delete_scope.where(root_attachment_id: nil).find_ids_in_batches(batch_size: 500) do |ids_batch|
         non_type_children = Attachment.where(root_attachment_id: ids_batch).
           not_deleted.
-          where.not(root_attachment_id: nil). # postgres is being weird
           where.not(context_type: context_type).
+          where.not(root_attachment_id: nil). # postgres is being weird
           order([:root_attachment_id, :id]).
           select("distinct on (attachments.root_attachment_id) attachments.*").
           group_by(&:root_attachment_id)
@@ -53,22 +53,30 @@ class Attachments::GarbageCollector
           end
 
           if non_type_children[att.id].present?
+            if context_type == 'ContentExport' &&
+                non_type_children[att.id].detect{ |x| x.context_type == 'ContentMigration' }.present?
+              stats[:cm_skipped] += 1
+              next
+            end
+
             stats[:reparent] += 1
             # make_childless separates this object and copies the content to
             # a new root attachment, so we still want to delete the content here.
             att.make_childless(non_type_children[att.id].first) unless dry_run
-            att.destroy_content unless dry_run
+            destroy_att_with_retries(att)
           elsif att.filename.present?
             stats[:destroyed] += 1
-            att.destroy_content unless dry_run
+            destroy_att_with_retries(att)
           end
 
           to_delete_ids.concat([att.id, same_type_children_ids].flatten)
         end
 
-        stats[:marked_deleted] += to_delete_ids.count
-        updates = { workflow_state: 'deleted', file_state: 'deleted', deleted_at: Time.now.utc }
-        Attachment.where(id: to_delete_ids).update_all(updates) unless dry_run
+        if to_delete_ids.present?
+          stats[:marked_deleted] += to_delete_ids.count
+          updates = { workflow_state: 'deleted', file_state: 'deleted', deleted_at: Time.now.utc }
+          Attachment.where(id: to_delete_ids).update_all(updates) unless dry_run
+        end
       end
     end
 
@@ -124,6 +132,13 @@ class Attachments::GarbageCollector
       return false unless children_max_created_at
       children_max_created_at >= older_than
     end
+
+    def destroy_att_with_retries(att, tries = 3)
+      att.destroy_content unless dry_run
+    rescue Aws::S3::Errors::InternalError
+      tries -= 1
+      tries.zero? ? raise : (sleep(10) && retry)
+    end
   end
 
   # context_type: 'Folder' is no longer generated by the code.
@@ -145,7 +160,7 @@ class Attachments::GarbageCollector
   # - context_type='User' (in the case of user data exports)
   # which is why we use the join conditions below
   class ContentExportContextType < ByContextType
-    def initialize(older_than:, dry_run: false)
+    def initialize(older_than: ContentExport.expire_days.days.ago, dry_run: false)
       super(context_type: 'ContentExport', older_than: older_than, dry_run: dry_run)
     end
 
@@ -161,4 +176,37 @@ SQL
       super
     end
   end
+
+  # We do lump exports and migrations together here because they are often
+  # intertwined.
+  #
+  # NOTE: content_migration.attachment are always either
+  # - context_type='ContentMigration', or
+  # - context_type='ContentExport'
+  class ContentExportAndMigrationContextType < ByContextType
+    def initialize(older_than: ContentMigration.expire_days.days.ago, dry_run: false)
+      super(context_type: ['ContentExport', 'ContentMigration'], older_than: older_than, dry_run: dry_run)
+    end
+
+    def delete_rows
+      raise "Cannot delete rows in dry_run mode" if dry_run
+      ce_null_scope = ContentExport.joins(<<-SQL).
+INNER JOIN #{Attachment.quoted_table_name}
+ON attachments.context_type = 'ContentExport'
+AND content_exports.attachment_id = attachments.id
+SQL
+        where(attachments: { workflow_state: 'deleted', file_state: 'deleted' })
+      while ce_null_scope.limit(1000).update_all(attachment_id: nil) > 0; end
+
+      cm_null_scope = ContentMigration.joins(<<-SQL).
+INNER JOIN #{Attachment.quoted_table_name}
+ON attachments.context_type IN ('ContentMigration', 'ContentExport')
+AND content_migrations.attachment_id = attachments.id
+SQL
+        where(attachments: { workflow_state: 'deleted', file_state: 'deleted' })
+      while cm_null_scope.limit(1000).update_all(attachment_id: nil) > 0; end
+
+      super
+    end
+  end
 end