Add support for verifying remote uploads, artifacts, and LFS objects in check rake tasks

5 files changed, 79 insertions, 16 deletions

diff --git a/lib/gitlab/verify/batch_verifier.rb b/lib/gitlab/verify/batch_verifier.rb
index 1ef369a4b67..167ba1b3149 100644
--- a/lib/gitlab/verify/batch_verifier.rb
+++ b/lib/gitlab/verify/batch_verifier.rb
@@ -7,13 +7,15 @@ module Gitlab
         @batch_size = batch_size
         @start = start
         @finish = finish
+
+        fix_google_api_logger
       end
 
       # Yields a Range of IDs and a Hash of failed verifications (object => error)
       def run_batches(&blk)
-        relation.in_batches(of: batch_size, start: start, finish: finish) do |relation| # rubocop: disable Cop/InBatches
-          range = relation.first.id..relation.last.id
-          failures = run_batch(relation)
+        all_relation.in_batches(of: batch_size, start: start, finish: finish) do |batch| # rubocop: disable Cop/InBatches
+          range = batch.first.id..batch.last.id
+          failures = run_batch_for(batch)
 
           yield(range, failures)
         end
@@ -29,24 +31,56 @@ module Gitlab
 
       private
 
-      def run_batch(relation)
-        relation.map { |upload| verify(upload) }.compact.to_h
+      def run_batch_for(batch)
+        batch.map { |upload| verify(upload) }.compact.to_h
       end
 
       def verify(object)
+        local?(object) ? verify_local(object) : verify_remote(object)
+      rescue => err
+        failure(object, err.inspect)
+      end
+
+      def verify_local(object)
         expected = expected_checksum(object)
         actual = actual_checksum(object)
 
-        raise 'Checksum missing' unless expected.present?
-        raise 'Checksum mismatch' unless expected == actual
+        return failure(object, 'Checksum missing') unless expected.present?
+        return failure(object, 'Checksum mismatch') unless expected == actual
+
+        success
+      end
 
+      # We don't calculate checksum for remote objects, so just check existence
+      def verify_remote(object)
+        return failure(object, 'Remote object does not exist') unless remote_object_exists?(object)
+
+        success
+      end
+
+      def success
         nil
-      rescue => err
-        [object, err]
+      end
+
+      def failure(object, message)
+        [object, message]
+      end
+
+      # It's already set to Logger::INFO, but acts as if it is set to
+      # Logger::DEBUG, and this fixes it...
+      def fix_google_api_logger
+        if Object.const_defined?('Google::Apis')
+          Google::Apis.logger.level = Logger::INFO
+        end
       end
 
       # This should return an ActiveRecord::Relation suitable for calling #in_batches on
-      def relation
+      def all_relation
+        raise NotImplementedError.new
+      end
+
+      # Should return true if the object is stored locally
+      def local?(_object)
         raise NotImplementedError.new
       end
 
@@ -59,6 +93,11 @@ module Gitlab
       def actual_checksum(_object)
         raise NotImplementedError.new
       end
+
+      # Be sure to perform a hard check of the remote object (don't just check DB value)
+      def remote_object_exists?(object)
+        raise NotImplementedError.new
+      end
     end
   end
 end
diff --git a/lib/gitlab/verify/job_artifacts.rb b/lib/gitlab/verify/job_artifacts.rb
index 03500a61074..dbadfbde9e3 100644
--- a/lib/gitlab/verify/job_artifacts.rb
+++ b/lib/gitlab/verify/job_artifacts.rb
@@ -11,10 +11,14 @@ module Gitlab
 
       private
 
-      def relation
+      def all_relation
         ::Ci::JobArtifact.all
       end
 
+      def local?(artifact)
+        artifact.local_store?
+      end
+
       def expected_checksum(artifact)
         artifact.file_sha256
       end
@@ -22,6 +26,10 @@ module Gitlab
       def actual_checksum(artifact)
         Digest::SHA256.file(artifact.file.path).hexdigest
       end
+
+      def remote_object_exists?(artifact)
+        artifact.file.file.exists?
+      end
     end
   end
 end
diff --git a/lib/gitlab/verify/lfs_objects.rb b/lib/gitlab/verify/lfs_objects.rb
index 970e2a7b718..d3f58a73ac7 100644
--- a/lib/gitlab/verify/lfs_objects.rb
+++ b/lib/gitlab/verify/lfs_objects.rb
@@ -11,8 +11,12 @@ module Gitlab
 
       private
 
-      def relation
-        LfsObject.with_files_stored_locally
+      def all_relation
+        LfsObject.all
+      end
+
+      def local?(lfs_object)
+        lfs_object.local_store?
       end
 
       def expected_checksum(lfs_object)
@@ -22,6 +26,10 @@ module Gitlab
       def actual_checksum(lfs_object)
         LfsObject.calculate_oid(lfs_object.file.path)
       end
+
+      def remote_object_exists?(lfs_object)
+        lfs_object.file.file.exists?
+      end
     end
   end
 end
diff --git a/lib/gitlab/verify/rake_task.rb b/lib/gitlab/verify/rake_task.rb
index dd138e6b92b..e190eaddc79 100644
--- a/lib/gitlab/verify/rake_task.rb
+++ b/lib/gitlab/verify/rake_task.rb
@@ -45,7 +45,7 @@ module Gitlab
         return unless verbose?
 
         failures.each do |object, error|
-          say "  - #{verifier.describe(object)}: #{error.inspect}".color(:red)
+          say "  - #{verifier.describe(object)}: #{error}".color(:red)
         end
       end
     end
diff --git a/lib/gitlab/verify/uploads.rb b/lib/gitlab/verify/uploads.rb
index 0ffa71a6d72..01f09ab8df7 100644
--- a/lib/gitlab/verify/uploads.rb
+++ b/lib/gitlab/verify/uploads.rb
@@ -11,8 +11,12 @@ module Gitlab
 
       private
 
-      def relation
-        Upload.with_files_stored_locally
+      def all_relation
+        Upload.all
+      end
+
+      def local?(upload)
+        upload.local?
       end
 
       def expected_checksum(upload)
@@ -22,6 +26,10 @@ module Gitlab
       def actual_checksum(upload)
         Upload.hexdigest(upload.absolute_path)
       end
+
+      def remote_object_exists?(upload)
+        upload.build_uploader.file.exists?
+      end
     end
   end
 end