Fix error 500 when sorting issues by milestone due date and filtering by labels

4 files changed, 30 insertions, 3 deletions

diff --git a/CHANGELOG b/CHANGELOG
index a0194a1ba25..a885e44f38b 100644
--- a/CHANGELOG
+++ b/CHANGELOG
@@ -10,6 +10,7 @@ v 8.9.0 (unreleased)
   - Changed the Slack build message to use the singular duration if necessary (Aran Koning)
   - Fix issues filter when ordering by milestone
   - Todos will display target state if issuable target is 'Closed' or 'Merged'
+  - Fix bug when sorting issues by milestone due date and filtering by two or more labels
   - Remove 'main language' feature
   - Measure queue duration between gitlab-workhorse and Rails
 
diff --git a/app/finders/issuable_finder.rb b/app/finders/issuable_finder.rb
index 8ed3ccf1c02..7d8c56f4c22 100644
--- a/app/finders/issuable_finder.rb
+++ b/app/finders/issuable_finder.rb
@@ -271,7 +271,7 @@ class IssuableFinder
       if filter_by_no_label?
         items = items.without_label
       else
-        items = items.with_label(label_names)
+        items = items.with_label(label_names, params[:sort])
         if projects
           items = items.where(labels: { project_id: projects })
         end
diff --git a/app/models/concerns/issuable.rb b/app/models/concerns/issuable.rb
index 91315b3459f..117d66131ae 100644
--- a/app/models/concerns/issuable.rb
+++ b/app/models/concerns/issuable.rb
@@ -126,13 +126,30 @@ module Issuable
       joins(join_clause).group(issuable_table[:id]).reorder("COUNT(notes.id) DESC")
     end
 
-    def with_label(title)
+    def with_label(title, sort = nil)
       if title.is_a?(Array) && title.size > 1
-        joins(:labels).where(labels: { title: title }).group(arel_table[:id]).having("COUNT(DISTINCT labels.title) = #{title.size}")
+        joins(:labels).where(labels: { title: title }).group(*get_grouping_columns(sort)).having("COUNT(DISTINCT labels.title) = #{title.size}")
       else
         joins(:labels).where(labels: { title: title })
       end
     end
+
+    # Includes table keys in group by clause when sorting
+    # preventing errors in postgres
+    #
+    # Returns an array of arel columns
+
+    def get_grouping_columns(sort)
+      default_columns = [arel_table[:id]]
+
+      if ["milestone_due_desc", "milestone_due_asc"].include?(sort)
+        milestone_table = Milestone.arel_table
+        default_columns << milestone_table[:id]
+        default_columns << milestone_table[:due_date]
+      end
+
+      default_columns
+    end
   end
 
   def today?
diff --git a/spec/features/merge_requests/user_lists_merge_requests_spec.rb b/spec/features/merge_requests/user_lists_merge_requests_spec.rb
index 2c7e1c748ad..1c130057c56 100644
--- a/spec/features/merge_requests/user_lists_merge_requests_spec.rb
+++ b/spec/features/merge_requests/user_lists_merge_requests_spec.rb
@@ -131,6 +131,15 @@ describe 'Projects > Merge requests > User lists merge requests', feature: true
         expect(first_merge_request).to include('fix')
         expect(count_merge_requests).to eq(1)
       end
+
+      it 'sorts by recently due milestone' do
+        visit namespace_project_merge_requests_path(project.namespace, project,
+          label_name: [label.name, label2.name],
+          assignee_id: user.id,
+          sort: sort_value_milestone_soon)
+
+        expect(first_merge_request).to include('fix')
+      end
     end
   end