Uploaded image for project: 'Mongoid'
  1. Mongoid
  2. MONGOID-2331

Saving duplicates on 2 levels embedded nested attributes

XMLWordPrintableJSON

    • Type: Icon: Task Task
    • Resolution: Done
    • 3.0.7
    • Affects Version/s: None
    • Component/s: None

      Hi,

      I've got a strange behavior on 2 levels nested attributes with mongoid-3.0.5 (forms are handled by simple_form and cocoon).

      I have 3 models : Node, Volume and Filesystem.

      Node has_many :volumes and Volume embeds_many :filesystems.

      When adding new filesystems to an existing volume, those filesystems are persisted twice, with the same ID.

      When destroying one of those filesystems, both are destroyed (since they have the same id).

      The problem does not occurs when :

      _ Adding filesystems on a _new* volume (both created on the same time)

      • Using Volume has_many :filesystems instead of Volume embeds_many :filesystems (+ belongs_to instead of embedded_in in Filesystem)

      Code samples :

      (I just skip validations, which are just presence, inclusion and numericality, and custom instance methods)

      volume.rb header :
      class Volume
  
  # INCLUSIONS

    include Mongoid::Document
    include Mongoid::Timestamps

  # RELATIONS
  
    belongs_to :node
    embeds_many :filesystems
    accepts_nested_attributes_for :filesystems, allow_destroy: true, reject_if: proc { |attributes| attributes.all? {|k,v| v.blank? || ['size_unit'].include?(k)} }

  # FIELDS
  
    field :name
    field :size
    field :size_unit, default: :gb

[...]
      filesystem.rb header :
      class Filesystem
  
  # INCLUSIONS

    include Mongoid::Document
    include Mongoid::Timestamps

  # RELATIONS
  
    embedded_in :volume

  # FIELDS
  
    field :mount_point
    field :name
    field :size
    field :size_unit, default: :gb

[...]
      nodes_controller.rb update method :
          def update
      @node= platform.nodes.find_by name: params[:id]

      if @node.update_attributes params[:node]
        flash[:success] = t_flash :update_success, @node        redirect_to platforms_platform_node_path(organization, platform, @node)
      else
        flash.now[:error] = t_flash :update_error, @node, count: @node.errors.count
        render :form
      end
    end

      params hash :

      (I only paste params[:node][:volume_attributes] part since Node has many other nested attributes)

      When creating a new Volume with a Filesystem :
      volumes_attributes: !ruby/hash:ActiveSupport::HashWithIndifferentAccess
    '0': !ruby/hash:ActiveSupport::HashWithIndifferentAccess
      name: rootvg
      size: '100'
      size_unit: gb
      filesystems_attributes: !ruby/hash:ActiveSupport::HashWithIndifferentAccess
        '0': !ruby/hash:ActiveSupport::HashWithIndifferentAccess
          mount_point: /
          name: rootlv
          size: '10'
          size_unit: gb

      Gives :

      irb(main):004:0> Node.find_by(name: 'Server1').volumes.first.filesystems.each{|f| puts f.inspect}
#<Filesystem _id: 503f78226fad9df0220000f2, _type: nil, created_at: nil, updated_at: nil, mount_point: "/", name: "rootlv", size: "10", size_unit: "gb">
      When adding a new Filesystem to the same Volume :
        volumes_attributes: !ruby/hash:ActiveSupport::HashWithIndifferentAccess
    '0': !ruby/hash:ActiveSupport::HashWithIndifferentAccess
      name: rootvg
      size: '100'
      size_unit: gb
      filesystems_attributes: !ruby/hash:ActiveSupport::HashWithIndifferentAccess
        '0': !ruby/hash:ActiveSupport::HashWithIndifferentAccess
          mount_point: /
          name: rootlv
          size: '10'
          size_unit: gb
        '1346337168506': !ruby/hash:ActiveSupport::HashWithIndifferentAccess
          mount_point: /home
          name: homelv
          size: '20'
          size_unit: gb

      (Volume and first Filesystem IDs do not appear here since it's the filtered params hash but they are sent in the post request)

      Gives :

      irb(main):008:0> Node.find_by(name: 'Server1').volumes.first.filesystems.each{|f| puts f.inspect}
#<Filesystem _id: 503f78226fad9df0220000f2, _type: nil, created_at: nil, updated_at: nil, mount_point: "/", name: "rootlv", size: "10", size_unit: "gb">
#<Filesystem _id: 503f79a16fad9df0220000fa, _type: nil, created_at: nil, updated_at: nil, mount_point: "/home", name: "homelv", size: "20", size_unit: "gb">
#<Filesystem _id: 503f79a16fad9df0220000fa, _type: nil, created_at: nil, updated_at: nil, mount_point: "/home", name: "homelv", size: "20", size_unit: "gb">

      Note : I've just noted when pasting this log that filesystems timestamps fields are created with nil value...which is not the case of the Volume :

      `irb(main):024:0> Node.find_by(name: 'Server1').volumes.first
=> #<Volume _id: 503f78226fad9df0220000ef, _type: nil, created_at: 2012-08-30 14:26:42 UTC, updated_at: 2012-08-30 14:37:03 UTC, node_id: "503f78066fad9df0220000e9", name: "rootvg", size: "100", size_unit: "gb">

            Assignee:
            durran Durran Jordan
            Reporter:
            gauthier-delacroix gauthier-delacroix
            Votes:
            0 Vote for this issue
            Watchers:
            0 Start watching this issue

              Created:
              Updated:
              Resolved: