Implement `failure_reason` on `ci_builds`

app/models/ci/build.rb

@@ -103,7 +103,7 @@ def retry(build, current_user)
         end
       end
 
-      before_transition any => [:failed] do |build|
+      before_transition any => [:failed] do |build, transition|
         next if build.retries_max.zero?
 
         if build.retries_count < build.retries_max

app/models/commit_status.rb

@@ -38,6 +38,19 @@ class CommitStatus < ActiveRecord::Base
   scope :retried_ordered, -> { retried.ordered.includes(project: :namespace) }
   scope :after_stage, -> (index) { where('stage_idx > ?', index) }
 
+  enum failure_reason: {
+    no_error: nil,
+    failed_by_script: 1, # TODO: Not used. Should we expand pipeline as well?
+    failed_by_missing_dependency: 2, # This will be done in the next MR.
+    failed_by_system: 3, # TODO: Not used. What's this state?
+    failed_by_failed_job_state: 4,
+    failed_by_out_of_quota: 5, # TODO: Only EE. How can we detect?
+    failed_by_stuck_and_timeout: 6,
+    failed_by_no_runner: 7, # TODO: Not used. How can we detect?
+    failed_by_api: 8,
+    failed_by_page: 9
+  }
+
   state_machine :status do
     event :process do
       transition [:skipped, :manual] => :created
@@ -79,6 +92,11 @@ class CommitStatus < ActiveRecord::Base
       commit_status.finished_at = Time.now
     end
 
+    before_transition any => :failed do |commit_status, transition|
+      failure_reason = transition.args.first
+      commit_status.failure_reason = failure_reason
+    end
+
     after_transition do |commit_status, transition|
       next if transition.loopback?
 

app/services/projects/update_pages_service.rb

@@ -53,7 +53,7 @@ def error(message, http_status = nil)
       log_error("Projects::UpdatePagesService: #{message}")
       @status.allow_failure = !latest?
       @status.description = message
-      @status.drop
+      @status.drop(:page)
       super
     end
 

app/workers/stuck_ci_jobs_worker.rb

@@ -53,7 +53,7 @@ def search(status, timeout)
   def drop_build(type, build, status, timeout)
     Rails.logger.info "#{self.class}: Dropping #{type} build #{build.id} for runner #{build.runner_id} (status: #{status}, timeout: #{timeout})"
     Gitlab::OptimisticLocking.retry_lock(build, 3) do |b|
-      b.drop
+      b.drop(:stuck_and_timeout)
     end
   end
 end

db/migrate/20170830125940_add_failure_reason_to_ci_builds.rb

+class AddFailureReasonToCiBuilds < ActiveRecord::Migration
+  include Gitlab::Database::MigrationHelpers
+
+  DOWNTIME = false
+
+  def change
+    add_column :ci_builds, :failure_reason, :integer
+  end
+end

db/schema.rb

@@ -11,7 +11,7 @@
 #
 # It's strongly recommended that you check this file into your version control system.
 
-ActiveRecord::Schema.define(version: 20170824162758) do
+ActiveRecord::Schema.define(version: 20170830125940) do
 
   # These are extensions that must be enabled in order to support this database
   enable_extension "plpgsql"
@@ -247,6 +247,7 @@
     t.boolean "retried"
     t.integer "stage_id"
     t.boolean "protected"
+    t.integer "failure_reason"
   end
 
   add_index "ci_builds", ["auto_canceled_by_id"], name: "index_ci_builds_on_auto_canceled_by_id", using: :btree

lib/api/commit_statuses.rb

@@ -103,7 +103,7 @@ class CommitStatuses < Grape::API
           when 'success'
             status.success!
           when 'failed'
-            status.drop!
+            status.drop!(:api)
           when 'canceled'
             status.cancel!
           else

lib/api/runner.rb

@@ -127,7 +127,7 @@ class Runner < Grape::API
         when 'success'
           job.success
         when 'failed'
-          job.drop
+          job.drop(:failed_job_state)
         end
       end
 

spec/models/ci/build_spec.rb

@@ -1710,4 +1710,24 @@ def create_mr(build, pipeline, factory: :merge_request, created_at: Time.now)
       end
     end
   end
+
+  describe 'set failure_reason when drop' do
+    let(:build) { create(:ci_build, :created) }
+
+    before do
+      build.drop!(reason)
+    end
+
+    context 'when failure_reason is nil' do
+      let(:reason) { }
+
+      it { expect(build).to be_no_error }
+    end
+
+    context 'when failure_reason is script_error' do
+      let(:reason) { :script_error }
+
+      it { expect(build).to be_failed_by_missing_dependency }
+    end
+  end
 end

spec/requests/api/commit_statuses_spec.rb

@@ -142,6 +142,9 @@ def create_status(commit, opts = {})
               expect(json_response['ref']).not_to be_empty
               expect(json_response['target_url']).to be_nil
               expect(json_response['description']).to be_nil
+              if status == 'failed'
+                expect(CommitStatus.find(json_response['id'])).to be_failed_by_api
+              end
             end
           end
         end

spec/requests/api/runner_spec.rb

@@ -627,12 +627,14 @@ def request_job(token = runner.token, **params)
           update_job(state: 'success')
 
           expect(job.reload.status).to eq 'success'
+          expect(job).to be_no_error
         end
 
         it 'mark job as failed' do
           update_job(state: 'failed')
 
           expect(job.reload.status).to eq 'failed'
+          expect(job).to be_failed_by_failed_job_state
         end
       end
 

spec/services/projects/update_pages_service_spec.rb

@@ -116,6 +116,7 @@
 
         expect(deploy_status.description)
           .to match(/artifacts for pages are too large/)
+        expect(deploy_status).to be_failed_by_page
       end
     end
 

spec/workers/stuck_ci_jobs_worker_spec.rb

@@ -20,6 +20,7 @@
     it 'changes status' do
       worker.perform
       is_expected.to eq('failed')
+      expect(job).to be_failed_by_stuck_and_timeout
     end
   end