ci: Distribute static-analysis tasks more intelligently

Signed-off-by: Rémy Coutable <remy@rymai.me>

scripts/static-analysis

@@ -19,18 +19,18 @@ class StaticAnalysis
   # file that is generated by an EE installation, which can
   # contain values that a FOSS installation won't find. To work
   # around this we will only enable this task on EE installations.
-  TASKS_BY_DURATIONS_SECONDS_DESC = {
+  TASKS_WITH_DURATIONS_SECONDS = {
     %w[bin/rake lint:haml] => 800,
     # We need to disable the cache for this cop since it creates files under tmp/feature_flags/*.used,
     # the cache would prevent these files from being created.
     %w[bundle exec rubocop --only Gitlab/MarkUsedFeatureFlags --cache false] => 600,
     (Gitlab.ee? ? %w[bin/rake gettext:updated_check] : nil) => 360,
     %w[yarn run lint:eslint:all] => 312,
+    %w[bundle exec rubocop --parallel] => 300,
     %w[yarn run lint:prettier] => 162,
     %w[bin/rake gettext:lint] => 65,
     %w[bundle exec license_finder] => 61,
     %w[bin/rake lint:static_verification] => 45,
-    %w[bundle exec rubocop --parallel] => 40,
     %w[bin/rake config_lint] => 26,
     %w[bin/rake gitlab:sidekiq:all_queues_yml:check] => 15,
     (Gitlab.ee? ? %w[bin/rake gitlab:sidekiq:sidekiq_queues_yml:check] : nil) => 11,
@@ -40,14 +40,21 @@ class StaticAnalysis
     %w[scripts/lint-rugged] => 1,
     %w[scripts/gemfile_lock_changed.sh] => 1,
     %w[scripts/frontend/check_no_partial_karma_jest.sh] => 1
-  }.reject { |k| k.nil? }.sort_by { |a| -a[1] }.to_h.keys.freeze
+  }.reject { |k| k.nil? }.freeze
 
-  def run_tasks!
-    tasks = tasks_to_run((ENV['CI_NODE_INDEX'] || 1).to_i, (ENV['CI_NODE_TOTAL'] || 1).to_i)
+  StaticAnalysisTasks = Struct.new(:tasks, :duration)
+
+  def run_tasks!(options = {})
+    node_tasks = tasks_to_run((ENV['CI_NODE_TOTAL'] || 1).to_i, debug: options[:debug])[(ENV['CI_NODE_INDEX'] || 1).to_i - 1]
+
+    if options[:dry_run]
+      puts "Dry-run mode!"
+      return
+    end
 
     static_analysis = Gitlab::Popen::Runner.new
 
-    static_analysis.run(tasks) do |cmd, &run|
+    static_analysis.run(node_tasks.tasks) do |cmd, &run|
       puts
       puts "$ #{cmd.join(' ')}"
 
@@ -107,16 +114,62 @@ class StaticAnalysis
       .count { |result| !ALLOWED_WARNINGS.include?(result.stderr.strip) }
   end
 
-  def tasks_to_run(node_index, node_total)
-    tasks = []
-    TASKS_BY_DURATIONS_SECONDS_DESC.each_with_index do |task, i|
-      tasks << task if i % node_total == (node_index - 1)
+  def tasks_to_run(node_total, debug: false)
+    tasks_per_node = Array.new(node_total) { StaticAnalysisTasks.new([], 0) }
+
+    total_time = TASKS_WITH_DURATIONS_SECONDS.values.sum.to_f
+    ideal_time_per_job = total_time / node_total
+    tasks_by_duration_desc = TASKS_WITH_DURATIONS_SECONDS.sort_by { |a| -a[1] }.to_h
+
+    p "total_time: #{total_time}" if debug
+    p "ideal_time_per_job: #{ideal_time_per_job}" if debug
+
+    tasks_by_duration_desc.each_with_index do |(task, duration), i|
+      puts "Assigning #{task}..." if debug
+      (0...node_total).each do |node_index|
+        puts "Current node: #{node_index}..." if debug
+        # Task is already longer than the ideal time
+        if duration >= ideal_time_per_job && tasks_per_node[node_index].tasks.empty?
+          puts "Assigning #{task} to node #{node_index} (#{duration}s)." if debug
+          assign_task_to_node(tasks_by_duration_desc, tasks_per_node[node_index], task, duration)
+          break
+        elsif tasks_per_node[node_index].duration + duration <= ideal_time_per_job
+          puts "Assigning #{task} to node #{node_index} (#{duration}s)." if debug
+          assign_task_to_node(tasks_by_duration_desc, tasks_per_node[node_index], task, duration)
+          break
+        else
+          puts "Node #{node_index} is already full (#{tasks_per_node[node_index]})" if debug
+        end
+      end
     end
 
-    tasks
+    raise "There are unassigned tasks: #{tasks_by_duration_desc}" unless tasks_by_duration_desc.empty?
+
+    tasks_per_node.each_with_index do |node, i|
+      puts "\nExpected duration for node #{i + 1}: #{node.duration}"
+      node.tasks.each { |task| puts "- #{task.join(' ')}" }
+    end
+
+    tasks_per_node
+  end
+
+  def assign_task_to_node(remaining_tasks, node, task_name, duration)
+    node.tasks << task_name
+    node.duration += duration
+    remaining_tasks.delete(task_name)
   end
 end
 
 if $0 == __FILE__
-  StaticAnalysis.new.run_tasks!
+  options = {}
+
+  if ARGV.include?('--dry-run')
+    options[:dry_run] = true
+  end
+
+  if ARGV.include?('--debug')
+    options[:debug] = true
+  end
+
+  StaticAnalysis.new.run_tasks!(options)
 end