ComplexType objects incorrectly require @odata.type property

[openUBMC-PLUSONE]

When validating a Certificate resource (e.g., Certificate.v1_3_0.Certificate), the validator reports the following error for the embedded Issuer property:

Required Property Error: The property '@odata.type' is mandatory, but not present in the payload.
Path: Issuer/@odata.type
Schema Definition
In the CSDL schema (certificate_v1.xml), the Issuer property is defined as:

<Property Name="Issuer" Type="Certificate.v1_0_0.Identifier" Nullable="false">
  <Annotation Term="OData.Description" String="The issuer of the certificate."/>
</Property>

The Identifier type is a ComplexType, not an EntityType:

  <ComplexType Name="Identifier">
    <Annotation Term="OData.AdditionalProperties" Bool="false"/>
    <Property Name="CommonName" Type="Edm.String" Nullable="false"/>
    <Property Name="Organization" Type="Edm.String" Nullable="false"/>
    ...
  </ComplexType>

In metadata.py (line 74-75), both EntityType and ComplexType are processed identically through the _add_object() method:

  if (child.tag == ODATA_TAG_ENTITY) or (child.tag == ODATA_TAG_COMPLEX):
      self._add_object(child)

The validator does not distinguish between EntityType and ComplexType. Does this cause it incorrectly requires @odata.type for ComplexType objects, then causing validation failures for conformant Redfish services?

[mraineri]

Does this cause it incorrectly requires @odata.type for ComplexType objects, then causing validation failures for conformant Redfish services?

No, we add an @odata.type requirement based on the base type hierarchy. Since Issuer in Certificate does not inherit from Resource.v1_0_0.Resource. See here: https://github.com/DMTF/Redfish-Service-Validator/blob/main/redfish_service_validator/metadata.py#L354

Please share your log file showing the error; this has not been observed on other systems implementing the same object.

[HanCJiang]

I have the similar problem.

The XML form nvidia
https://github.com/NVIDIA/bmcweb/blob/develop/redfish-core/schema/oem/nvidia/csdl/NvidiaMemoryMetrics_v1.xml

RedfishServiceValidatorDebug_04_16_2026_145352.log

RedfishServiceValidatorReport_04_16_2026_145352.html

[mraineri]

I think the issue with the NVIDIA one is different. From the logs and the schema file for the OEM extension, I see the 1_1_0 namespace is completely redefining objects and properties from the 1_0_0 namespace. The lookup logic to try to find the appropriate object definition starts from the "highest" namespace and works backwards.

Since the object inheritance tree for the RowRemapping object stops in 1_1_0 (since there's no basetype to point back to 1_0_0), the tool thinks it's an invalid usage. I'll see how we can accommodate this since it's technically valid (just not in good practice with other Redfish schemas).

[mraineri]

Here are fixes I made to the schema file to align it with Redfish modeling best practices

NvidiaMemoryMetrics_v1.xml