ManageIQ/manageiq

View on GitHub
app/models/service/aggregation.rb

Summary

Maintainability
A
0 mins
Test Coverage
B
88%
module Service::Aggregation
  extend ActiveSupport::Concern

  included do
    virtual_column :aggregate_direct_vm_cpus,                 :type => :integer
    virtual_column :aggregate_direct_vm_memory,               :type => :integer
    virtual_column :aggregate_direct_vm_disk_count,           :type => :integer
    virtual_column :aggregate_direct_vm_disk_space_allocated, :type => :integer
    virtual_column :aggregate_direct_vm_disk_space_used,      :type => :integer
    virtual_column :aggregate_direct_vm_memory_on_disk,       :type => :integer

    virtual_attribute :aggregate_all_vm_cpus,   :integer,
                      :arel => aggregate_service_hardware_arel('cpus', :cpu_total_cores)
    virtual_attribute :aggregate_all_vm_memory, :integer,
                      :arel => aggregate_service_hardware_arel('memory', :memory_mb)
    virtual_attribute :aggregate_all_vm_disk_count, :integer,
                      :arel => aggregate_all_vms_disk_count_arel
    virtual_attribute :aggregate_all_vm_disk_space_allocated, :integer,
                      :arel => aggregate_all_vms_disk_space_allocated_arel
    virtual_attribute :aggregate_all_vm_disk_space_used, :integer,
                      :arel => aggregate_all_vms_disk_space_used_arel
    virtual_attribute :aggregate_all_vm_memory_on_disk, :integer,
                      :arel => aggregate_service_hardware_arel('memory_on_disk', :memory_mb, :on_vms_only => true)

    AGGREGATE_ALL_VM_ATTRS = [
      :aggregate_all_vm_cpus,
      :aggregate_all_vm_memory,
      :aggregate_all_vm_disk_count,
      :aggregate_all_vm_disk_space_allocated,
      :aggregate_all_vm_disk_space_used,
      :aggregate_all_vm_memory_on_disk
    ].freeze

    scope :with_aggregates, -> { select(Arel.star, *AGGREGATE_ALL_VM_ATTRS) }
  end

  module ClassMethods
    # Sends the following `aggregation_sql` for the given `hardware_column` to
    # `aggregate_hardware_arel`:
    #
    #     SUM("hardwares"."<<<hardware_column>>>")
    #
    # @return (see #aggregate_hardware_arel)
    def aggregate_service_hardware_arel(aggregate_alias, hardware_column, options = {})
      virtual_column_name = "aggregate_all_vm_#{aggregate_alias}"
      aggregation_sql = hardwares_tbl[hardware_column].sum
      aggregate_hardware_arel(virtual_column_name, aggregation_sql, options)
    end

    # Sends the following `aggregation_sql` to `aggregate_hardware_arel`:
    #
    #     COUNT("disks"."id")
    #
    # @return (see #aggregate_hardware_arel)
    def aggregate_all_vms_disk_count_arel
      virtual_column_name = "aggregate_all_vm_disk_count"
      aggregation_sql = disks_tbl[:id].count
      aggregate_hardware_arel(virtual_column_name, aggregation_sql, :include_disks => true)
    end

    # Sends the following `aggregation_sql` to `aggregate_hardware_arel`:
    #
    #     SUM(COALESCE("disks"."size", 0))
    #
    # @return (see #aggregate_hardware_arel)
    def aggregate_all_vms_disk_space_allocated_arel
      column_name     = "aggregate_all_vm_disk_space_allocated"
      coalesce_values = [disks_tbl[:size], Arel.sql("0")]
      aggregation_sql = Arel::Nodes::NamedFunction.new('SUM', [disks_tbl.coalesce(coalesce_values)])
      aggregate_hardware_arel(column_name, aggregation_sql, :include_disks => true)
    end

    # Sends the following `aggregation_sql` to `aggregate_hardware_arel`:
    #
    #     SUM(COALESCE("disks"."size_on_disk", "disks"."size", 0))
    #
    # @return (see #aggregate_hardware_arel)
    def aggregate_all_vms_disk_space_used_arel
      column_name     = "aggregate_all_vm_disk_space_used"
      coalesce_values = [disks_tbl[:size_on_disk], disks_tbl[:size], Arel.sql("0")]
      aggregation_sql = Arel::Nodes::NamedFunction.new('SUM', [disks_tbl.coalesce(coalesce_values)])
      aggregate_hardware_arel(column_name, aggregation_sql, :include_disks => true)
    end

    # Generates a nested select statement that builds off of the `id` of the
    # record being queried against (or multiple records, though this will not
    # scale well for that). The generated SQL will look something like the
    # following when used with a query limiting by a single ID, `my_service_id`
    # in this case:
    #
    #     SELECT id,
    #           (SELECT <<<aggregation_sql>>>
    #            FROM services aggregate_all_vm_cpus_services
    #            JOIN service_resources ON service_resources.service_id = aggregate_all_vm_cpus_services.id
    #                                  AND service_resources.resource_type = 'VmOrTemplate'
    #            JOIN vms               ON vms.id = service_resources.resource_id
    #            JOIN hardwares         ON hardwares.vm_or_template_id = vms.id
    #            WHERE (("aggregate_all_vm_cpus_services"."id" = services.id
    #                    OR "aggregate_all_vm_cpus_services"."ancestry" ILIKE CONCAT(services.id, '/%'))
    #                   OR "aggregate_all_vm_cpus_services"."ancestry" = CAST(services.id AS VARCHAR))) AS <<<virtual_column_name>>>
    #     FROM services
    #     WHERE id = <<<some_service_id>>>;
    #
    # @param virtual_column_name [String] Name of the virtual_column method
    # @param aggregation_sql [String] Arel that handles the returned SELECT
    #
    # @param options  [Hash] Refines the joins in the query (default: {})
    # @option options [Boolean] :include_disks Include the disks table
    # @option options [Boolean] :on_vms_only   Only join on powered "on" VMs
    #
    # @return [Proc] A proc with the sql to generate a nested SELECT for
    #   the aggregate column, which returns an Arel statement when called
    def aggregate_hardware_arel(virtual_column_name, aggregation_sql, options = {})
      lambda do |t|
        subtree_services             = Arel::Table.new(:services)
        subtree_services.table_alias = "#{virtual_column_name}_services"

        subselect = subtree_services.project(aggregation_sql)
        subselect = base_service_aggregation_join(subselect, subtree_services, options)
        join_on_disks(subselect) if options[:include_disks]

        t.grouping(subselect.where(aggregation_where_clause(t, subtree_services)))
      end
    end

    # Generates the following WHERE clause in arel
    #
    #     WHERE (("aggregate_services"."id" = services.id
    #            OR "aggregate_services"."ancestry" ILIKE CONCAT(services.id, '/%'))
    #           OR "aggregate_services"."ancestry" = CAST(services.id AS VARCHAR))
    #
    # Where "aggregate_services" is the services associated with the id passed
    # by the main query that calls `aggregate_hardware_arel`.  This generated
    # WHERE clause finds all of the descendant services for the service in the
    # "main query" (`services.id` in the above SQL output) by finding:
    #
    # * The services that with an `id` of `services.id`
    # * The services with an `ancestry` under `services.id` (ancestory)
    # * The services with an `ancestry` equaling `services.id` (direct child)
    #
    # @param arel [Arel::Table] The table for the outer query, or basically
    #   Service.arel_table
    # @param subtree_services [Arel::Table] The aliased table for services
    #   inside the nested SELECT, usually named off of the virtual_column
    # @return [Arel::Nodes::Grouping] A conditional statement  wrapped in
    #   parthenses
    def aggregation_where_clause(arel, subtree_services)
      arel.grouping(
        subtree_services[:id].eq(arel[:id])
         .or(subtree_services[:ancestry].matches(ancestry_ilike, nil, true))
      ).or(subtree_services[:ancestry].eq(service_id_cast))
    end

    # Generates the following JOIN statement in arel
    #
    #    JOIN service_resources ON service_resources.service_id = aggregate_all_vm_cpus_services.id
    #                          AND service_resources.resource_type = 'VmOrTemplate'
    #    JOIN vms               ON vms.id = service_resources.resource_id
    #    JOIN hardwares         ON hardwares.vm_or_template_id = vms.id
    #
    # @param arel [Arel::Table] The aliased table for services inside the
    #   nested SELECT, usually named off of the virtual_column
    # @option (see #vm_join_clause)
    # @return [Arel::SelectManager] Arel with joins statements for
    #   service_resources, vms, and hardwares
    def base_service_aggregation_join(arel, services_tbl, options = {})
      arel.join(service_resources_tbl).on(service_resources_tbl[:service_id].eq(services_tbl[:id])
                                  .and(service_resources_tbl[:resource_type].eq(Arel.sql("'VmOrTemplate'"))))
          .join(vms_tbl).on(vm_join_clause(options)).tap do |arel_query|
            unless options[:skip_hardware]
              arel_query.join(hardwares_tbl).on(hardwares_tbl[:vm_or_template_id].eq(vms_tbl[:id]))
            end
          end
    end

    # Generates the following SQL conditional to be used in
    # `base_service_aggrigation_join`:
    #
    #     "vms"."id" = "service_resources"."resource_id"
    #
    # But will append an "AND" if `on_vms_only` is passed in as a option:
    #
    #     "vms"."id" = "service_resources"."resource_id" AND LOWER("vms"."power_state") == "on"
    #
    #
    # @option options [Boolean] :on_vms_only Filter "on" VMs only in the JOIN
    # @return [Arel::Nodes::And] The Arel conditional for JOINing the VMs table
    def vm_join_clause(options = {})
      clause = vms_tbl[:id].eq(service_resources_tbl[:resource_id])
      if options[:on_vms_only] # meaning VMs that are powered "on"
        clause = clause.and(vms_tbl[:power_state].lower.eq('on'))
      end
      clause
    end

    # Generates the following JOIN statement in arel
    #
    #    JOIN disks ON disks.hardware_id = hardwares.id
    #
    # @param (see #base_service_aggrigation_join)
    # @return (see #base_service_aggrigation_join)
    def join_on_disks(arel)
      arel.join(disks_tbl).on(disks_tbl[:hardware_id].eq(hardwares_tbl[:id]))
    end

    # NOTE: The following class methods are technically "constants", but to
    # make sure they are not triggered when the class is first loaded and cause
    # requests to the DB (can break building the appliance), we are making them
    # memoized class methods.

    def service_resources_tbl
      @service_resources_tbl ||= ServiceResource.arel_table
    end

    def vms_tbl
      @vms_tbl ||= Vm.arel_table
    end

    def hardwares_tbl
      @hardwares_tbl ||= Hardware.arel_table
    end

    def disks_tbl
      @disks_tbl ||= Disk.arel_table
    end

    def partitions_tbl
      @partitions_tbl ||= Partition.arel_table
    end

    def ancestry_ilike
      @ancestry_ilike ||= Arel::Nodes::NamedFunction.new("CONCAT", [Service.arel_table[:id], Arel.sql("'/%'")])
    end

    def service_id_cast
      @service_id_cast ||= Arel::Nodes::NamedFunction.new("CAST", [Service.arel_table[:id].as("VARCHAR")])
    end
  end

  def aggregate_direct_vm_cpus
    direct_vms.inject(0) { |aggregate, vm| aggregate + vm.cpu_total_cores.to_i }
  end

  def aggregate_direct_vm_memory
    direct_vms.inject(0) { |aggregate, vm| aggregate + vm.ram_size.to_i }
  end

  def aggregate_direct_vm_disk_count
    direct_vms.inject(0) { |aggregate, vm| aggregate + vm.num_disks.to_i }
  end

  def aggregate_direct_vm_disk_space_allocated
    direct_vms.inject(0) { |aggregate, vm| aggregate + vm.allocated_disk_storage.to_i }
  end

  def aggregate_direct_vm_disk_space_used
    direct_vms.inject(0) { |aggregate, vm| aggregate + vm.used_disk_storage.to_i }
  end

  def aggregate_direct_vm_memory_on_disk
    direct_vms.inject(0) { |aggregate, vm| aggregate + vm.ram_size_in_bytes_by_state.to_i }
  end

  def aggregate_all_vm_cpus
    if has_attribute?("aggregate_all_vm_cpus")
      self["aggregate_all_vm_cpus"]
    else
      all_vms.inject(0) { |aggregate, vm| aggregate + vm.cpu_total_cores.to_i }
    end
  end

  def aggregate_all_vm_memory
    if has_attribute?("aggregate_all_vm_memory")
      self["aggregate_all_vm_memory"]
    else
      all_vms.inject(0) { |aggregate, vm| aggregate + vm.ram_size.to_i }
    end
  end

  def aggregate_all_vm_disk_count
    if has_attribute?("aggregate_all_vm_disk_count")
      self["aggregate_all_vm_disk_count"]
    else
      all_vms.inject(0) { |aggregate, vm| aggregate + vm.num_disks.to_i }
    end
  end

  def aggregate_all_vm_disk_space_allocated
    if has_attribute?("aggregate_all_vm_disk_space_allocated")
      self["aggregate_all_vm_disk_space_allocated"]
    else
      all_vms.inject(0) { |aggregate, vm| aggregate + vm.allocated_disk_storage.to_i }
    end
  end

  def aggregate_all_vm_disk_space_used
    if has_attribute?("aggregate_all_vm_disk_space_used")
      self["aggregate_all_vm_disk_space_used"]
    else
      all_vms.inject(0) { |aggregate, vm| aggregate + vm.used_disk_storage.to_i }
    end
  end

  def aggregate_all_vm_memory_on_disk
    if has_attribute?("aggregate_all_vm_memory_on_disk")
      # Avoids (poorly) "ERROR:  integer out of range" in postgres
      self["aggregate_all_vm_memory_on_disk"].try(:*, 1.megabyte)
    else
      all_vms.inject(0) { |aggregate, vm| aggregate + vm.ram_size_in_bytes_by_state.to_i }
    end
  end
end