Merge branch 'blazer' into panel.sutty.nl

app/jobs/concerns/recursive_rollup.rb

@@ -0,0 +1,38 @@
+# frozen_string_literal: true
+
+module RecursiveRollup
+  extend ActiveSupport::Concern
+
+  included do
+    private
+
+    # Genera un rollup recursivo en base al período anterior y aplica una
+    # operación.
+    #
+    # @param :name [String]
+    # @param :interval_previous [String]
+    # @param :interval [String]
+    # @param :operation [Symbol]
+    # @param :dimensions [Hash]
+    # @param :new_name [String]
+    # @param :beginning [Time]
+    # @return [Rollup]
+    def recursive_rollup(name:, interval_previous:, interval:, dimensions:, beginning:, operation: :sum, new_name: nil)
+      Rollup.where(name: name, interval: interval_previous)
+            .where('time >= ?', beginning.try(:"beginning_of_#{interval}"))
+            .where_dimensions(**dimensions)
+            .group(*dimensions_to_jsonb_query(dimensions))
+            .rollup(new_name || name, interval: interval, update: true) do |rollup|
+              rollup.try(operation, :value)
+            end
+    end
+
+    # @param :dimensions [Hash]
+    # @return [Array]
+    def dimensions_to_jsonb_query(dimensions)
+      dimensions.keys.map do |key|
+        "dimensions->'#{key}'"
+      end
+    end
+  end
+end

app/jobs/stat_collection_job.rb

@@ -3,11 +3,17 @@
 # Genera resúmenes de información para poder mostrar estadísticas y se
 # corre regularmente a sí misma.
 class StatCollectionJob < PeriodicJob
+  include RecursiveRollup
+
   STAT_NAME = 'stat_collection_job'
 
   def perform(site_id:, once: true)
     @site = Site.find site_id
 
+    # Registrar que se hicieron todas las recolecciones
+    stat = site.stats.create! name: STAT_NAME
+    beginning = beginning_of_interval
+
     scope.rollup('builds', **options)
 
     scope.rollup('space_used', **options) do |rollup|
@@ -20,34 +26,21 @@ class StatCollectionJob < PeriodicJob
 
     # XXX: Es correcto promediar promedios?
     Stat::INTERVALS.reduce do |previous, current|
-      rollup(name: 'builds', interval_previous: previous, interval: current)
-      rollup(name: 'space_used', interval_previous: previous, interval: current, operation: :average)
-      rollup(name: 'build_time', interval_previous: previous, interval: current, operation: :average)
+      opts = { interval_previous: previous, interval: current, beginning: beginning, dimensions: { site_id: site.id } }
+
+      recursive_rollup(name: 'builds', **opts)
+      recursive_rollup(name: 'space_used', operation: :average, **opts)
+      recursive_rollup(name: 'build_time', operation: :average, **opts)
 
       current
     end
 
-    # Registrar que se hicieron todas las recolecciones
-    site.stats.create! name: STAT_NAME
-
+    stat.touch
     run_again! unless once
   end
 
   private
 
-  # Genera un rollup recursivo en base al período anterior y aplica una
-  # operación.
-  #
-  # @return [NilClass]
-  def rollup(name:, interval_previous:, interval:, operation: :sum)
-    Rollup.where(name: name, interval: interval_previous)
-          .where_dimensions(site_id: site.id)
-          .group("dimensions->'site_id'")
-          .rollup(name, interval: interval, update: true) do |rollup|
-            rollup.try(operation, :value)
-          end
-  end
-
   # Los registros a procesar
   #
   # @return [ActiveRecord::Relation]

app/jobs/uri_collection_job.rb

@@ -11,6 +11,8 @@
 # Los hostnames de un sitio van a poder obtenerse a partir de
 # Site#hostnames con la garantía de que son únicos.
 class UriCollectionJob < PeriodicJob
+  include RecursiveRollup
+
   # Ignoramos imágenes porque suelen ser demasiadas y no aportan a las
   # estadísticas.
   IMAGES = %w[.png .jpg .jpeg .gif .webp .jfif].freeze
@@ -21,35 +23,19 @@ class UriCollectionJob < PeriodicJob
 
     # Recordar la última vez que se corrió la tarea
     stat = site.stats.create! name: STAT_NAME
+    name = 'host|uri'
+    beginning = beginning_of_interval
+
+    hostnames.each do |host|
+      break if stop?
 
-    hostnames.each do |hostname|
       uris.each do |uri|
-        next if stop?
+        break if stop?
 
-        AccessLog.where(host: hostname, uri: uri)
-                 .where('created_at >= ?', beginning_of_interval)
-                 .completed_requests
-                 .non_robots
-                 .group(:host, :uri)
-                 .rollup('host|uri', interval: starting_interval, update: true)
-
-        # Reducir las estadísticas calculadas aplicando un rollup sobre el
-        # intervalo más amplio.
-        Stat::INTERVALS.reduce do |previous, current|
-          beginning_of_this_interval = beginning_of_interval.try(:"beginning_of_#{current}")
-
-          Rollup.where(name: 'host|uri', interval: previous)
-                .where('time >= ?', beginning_of_this_interval)
-                .where_dimensions(host: hostname, uri: uri)
-                .group("dimensions->'host'", "dimensions->'uri'")
-                .rollup('host|uri', interval: current, update: true) do |rollup|
-                  rollup.sum(:value)
-                end
-
-          # Devolver el intervalo actual
-          current
-        end
+        rollup_uri(uri, host, name, beginning)
       end
+
+      rollup_host(host, name, beginning)
     end
 
     stat.touch
@@ -59,6 +45,52 @@ class UriCollectionJob < PeriodicJob
 
   private
 
+  def rollup_uri(uri, host, name, beginning)
+    dimensions = { host: host, uri: uri }
+
+    AccessLog.where(**dimensions)
+             .where('created_at >= ?', beginning)
+             .completed_requests
+             .non_robots
+             .group(*dimensions.keys)
+             .rollup(name, interval: starting_interval, update: true)
+
+    # Reducir las estadísticas calculadas aplicando un rollup sobre el
+    # intervalo más amplio.
+    Stat::INTERVALS.reduce do |previous, current|
+      recursive_rollup(name: name,
+                       interval_previous: previous,
+                       interval: current,
+                       dimensions: dimensions,
+                       beginning: beginning)
+
+      # Devolver el intervalo actual
+      current
+    end
+  end
+
+  def rollup_host(host, name, beginning)
+    dimensions = { host: host }
+    new_name = 'host'
+
+    recursive_rollup(name: name,
+                     new_name: new_name,
+                     interval_previous: starting_interval,
+                     interval: starting_interval,
+                     dimensions: dimensions,
+                     beginning: beginning)
+
+    Stat::INTERVALS.reduce do |previous, current|
+      recursive_rollup(name: new_name,
+                       interval_previous: previous,
+                       interval: current,
+                       dimensions: dimensions,
+                       beginning: beginning)
+
+      current
+    end
+  end
+
   def stat_name
     STAT_NAME
   end

bin/access_logs

@@ -1,8 +1,8 @@
 #!/bin/sh
 set -e
 
-# Volcar y eliminar todos los access logs del día anterior
-date=`date +%F`
+# Volcar y eliminar todos los access logs de dos días atrás
+date=`bundle exec rails runner "puts (Date.today - 2.days)"`
 
 psql -h postgresql "${DATABASE:-sutty}" sutty <<SQL | gzip > "/srv/_storage/${date}.psql.gz"
 begin;