Explicitly pass transformation_mapping when generating mapping items

[djberg96]

In order to properly perform validations on a TransformationMappingItem, we need the parent to already be defined in some cases. This modifies the create_mapping method so that it accepts and uses the parent transformation mapping when creating a new transformation mapping item.

Tied to ManageIQ/manageiq#19204

[djberg96]

@miq-bot add_label transformation

[thearifismail]

change new to create on line 5 TransformationMapping.new(data.except("transformation_mapping_items")).tap do |mapping| to create a transformation map with a valid ID, which is needed for TransformationMappingItems.

[miq-bot]

Checked commits https://github.com/djberg96/manageiq-api/compare/b07d5c6b53e460c3670a092d84461a342cdeb512~...bcd4777c151a119809c79e206d1f7be390868d52 with ruby 2.4.6, rubocop 0.69.0, haml-lint 0.20.0, and yamllint 1.10.0
1 file checked, 0 offenses detected
Everything looks fine. 👍

[agrare]

Actually, if you do:

mapping = TransformationMapping.new(data.except("transformation_mapping_items"))
mapping.transformation_mapping_items = create_mapping_items(data["transformation_mapping_items"], mapping)
mapping.save!

Then if the transformation_mapping_item validation fails it will rollback the create of the top-level transformation_mapping which is what I believe @thearifismail was going for in ManageIQ/manageiq#19216 wrong link sorry, here it is ManageIQ/manageiq#19204

[agrare]

Up to you if that is the behavior you want though

[djberg96]

Right, but the mapping items still wouldn't have a parent at that point, which we need to get at datastore and lan information.

[djberg96]

@thearifismail, at least, I thought that was the issue, right?

[agrare]

Right, but the mapping items still wouldn't have a parent at that point, which we need to get at datastore and lan information.

If you pass the mapping in to TransformationMappingItem.new they will though, see ManageIQ/manageiq#19204 (comment) I tested that case where the parent TM isn't saved when the TMI validations are run

[thearifismail]

@djberg96 is right. We had to use create to get transformation_mapping_id generated for use with the subsequent mapping items calls. This is what I have:

      TransformationMapping.create(data.except("transformation_mapping_items")).tap do |mapping|
        mapping.transformation_mapping_items = create_mapping_items(data["transformation_mapping_items"], mapping)
        mapping.save!
      end

[agrare]

Did you try it though?

>> vc = ManageIQ::Providers::Vmware::InfraManager.first; rhv = ManageIQ::Providers::Redhat::InfraManager.first
>> source_storage = Storage.create!(:name => "unattached storage")
>> tm = TransformationMapping.new(:name => "my transformation")
>> tm.transformation_mapping_items = [
  TransformationMappingItem.new(:source => vc.ems_clusters.first, :destination => rhv.ems_clusters.first, :transformation_mapping => tm),
  TransformationMappingItem.new(:source => vc.lans.first, :destination => rhv.lans.first, :transformation_mapping => tm),
  TransformationMappingItem.new(:source => source_storage, :destination => rhv.storages.first, :transformation_mapping => tm)
]
>> tm.save!
   (0.3ms)  BEGIN
   (0.2ms)  ROLLBACK
ActiveRecord::RecordInvalid (Validation failed: TransformationMapping: Transformation mapping items is invalid)

And just to make sure:

   31:   def validate_source_datastore
   32:     byebug
   33:     tm                   = transformation_mapping
=> 34:     tmis                 = tm.transformation_mapping_items.where(:source_type => "EmsCluster")
   35:     src_cluster_storages = tmis.collect(&:source).flat_map(&:storages)
   36:     source_storage       = source
   37: 
   38:     unless src_cluster_storages.include?(source_storage)
(byebug) tm
#<TransformationMapping id: nil, name: "my transformation", description: nil, comments: nil, state: nil, options: {}, tenant_id: nil, created_at: nil, updated_at: nil>

and

   43:   def validate_destination_datastore
   44:     byebug
   45:     tm                  = transformation_mapping
=> 46:     destination_storage = destination
   47: 
   48:     if destination.kind_of?(Storage) # red hat
   49:       tmis                 = tm.transformation_mapping_items.where(:destination_type=> "EmsCluster")
   50:       dst_cluster_storages = tmis.collect(&:destination).flat_map(&:storages)
(byebug) tm
#<TransformationMapping id: nil, name: "my transformation", description: nil, comments: nil, state: nil, options: {}, tenant_id: nil, created_at: nil, updated_at: nil>

[agrare]

I should note that

>> tm.transformation_mapping_items = [
  TransformationMappingItem.new(:source => vc.ems_clusters.first, :destination => rhv.ems_clusters.first, :transformation_mapping => tm),
  TransformationMappingItem.new(:source => vc.lans.first, :destination => rhv.lans.first, :transformation_mapping => tm),
  TransformationMappingItem.new(:source => source_storage, :destination => rhv.storages.first, :transformation_mapping => tm)
]

Isn't how I'd do it, but it is functionally equivalent to what is being done here.

[thearifismail]

We had started off using new but it would not generate any mapping id and had observed exactly what @agrare observed using byebug, which produced transformation_mapping_id = nil. That's why we had to fall back to using create.

[agrare]

Wait I'm confused, I was saying that worked perfectly. Not having an :id yet isn't an issue, you can run your validations and if they pass tm.save! will commit the whole set to the database.

[djberg96]

Closed in favor of #666