feat: Add new cloudformation primary id attribute for all supported AWS resources

Author: yiyuan-he

aws-opentelemetry-distro/src/amazon/opentelemetry/distro/_aws_metric_attribute_generator.py

@@ -378,52 +378,56 @@ def _set_remote_type_and_identifier(span: ReadableSpan, attributes: BoundedAttri
     """
     remote_resource_type: Optional[str] = None
     remote_resource_identifier: Optional[str] = None
+    # In most cases, this will be the same as remote_resource_identifier.
+    # However, there are a few edge cases we have to handle such as compound CFN identifiers.
     cloudformation_primary_identifier: Optional[str] = None
 
     if is_aws_sdk_span(span):
         # Only extract the table name when _AWS_TABLE_NAMES has size equals to one
         if is_key_present(span, _AWS_TABLE_NAMES) and len(span.attributes.get(_AWS_TABLE_NAMES)) == 1:
             remote_resource_type = _NORMALIZED_DYNAMO_DB_SERVICE_NAME + "::Table"
             remote_resource_identifier = _escape_delimiters(span.attributes.get(_AWS_TABLE_NAMES)[0])
-            cloudformation_primary_identifier = ""
+            cloudformation_primary_identifier = remote_resource_identifier
         elif is_key_present(span, AWS_KINESIS_STREAM_NAME):
             remote_resource_type = _NORMALIZED_KINESIS_SERVICE_NAME + "::Stream"
             remote_resource_identifier = _escape_delimiters(span.attributes.get(AWS_KINESIS_STREAM_NAME))
-            cloudformation_primary_identifier = ""
+            cloudformation_primary_identifier = remote_resource_identifier
         elif is_key_present(span, _AWS_BUCKET_NAME):
             remote_resource_type = _NORMALIZED_S3_SERVICE_NAME + "::Bucket"
             remote_resource_identifier = _escape_delimiters(span.attributes.get(_AWS_BUCKET_NAME))
-            cloudformation_primary_identifier = ""
+            cloudformation_primary_identifier = remote_resource_identifier
         elif is_key_present(span, AWS_SQS_QUEUE_NAME):
             remote_resource_type = _NORMALIZED_SQS_SERVICE_NAME + "::Queue"
             remote_resource_identifier = _escape_delimiters(span.attributes.get(AWS_SQS_QUEUE_NAME))
-            cloudformation_primary_identifier = ""
+            cloudformation_primary_identifier = _escape_delimiters(span.attributes.get(AWS_SQS_QUEUE_URL))
         elif is_key_present(span, AWS_SQS_QUEUE_URL):
             remote_resource_type = _NORMALIZED_SQS_SERVICE_NAME + "::Queue"
             remote_resource_identifier = _escape_delimiters(
                 SqsUrlParser.get_queue_name(span.attributes.get(AWS_SQS_QUEUE_URL))
             )
-            cloudformation_primary_identifier = ""
+            cloudformation_primary_identifier = _escape_delimiters(span.attributes.get(AWS_SQS_QUEUE_URL))
         elif is_key_present(span, AWS_BEDROCK_AGENT_ID):
             remote_resource_type = _NORMALIZED_BEDROCK_SERVICE_NAME + "::Agent"
             remote_resource_identifier = _escape_delimiters(span.attributes.get(AWS_BEDROCK_AGENT_ID))
-            cloudformation_primary_identifier = ""
+            cloudformation_primary_identifier = remote_resource_identifier
         elif is_key_present(span, AWS_BEDROCK_DATA_SOURCE_ID):
             remote_resource_type = _NORMALIZED_BEDROCK_SERVICE_NAME + "::DataSource"
             remote_resource_identifier = _escape_delimiters(span.attributes.get(AWS_BEDROCK_DATA_SOURCE_ID))
-            cloudformation_primary_identifier = ""
+            cloudformation_primary_identifier = _escape_delimiters(span.attributes.get(AWS_BEDROCK_KNOWLEDGE_BASE_ID)) + "|" + remote_resource_identifier
         elif is_key_present(span, AWS_BEDROCK_GUARDRAIL_ID):
             remote_resource_type = _NORMALIZED_BEDROCK_SERVICE_NAME + "::Guardrail"
             remote_resource_identifier = _escape_delimiters(span.attributes.get(AWS_BEDROCK_GUARDRAIL_ID))
-            cloudformation_primary_identifier = ""
+            cloudformation_primary_identifier = remote_resource_identifier
         elif is_key_present(span, AWS_BEDROCK_KNOWLEDGE_BASE_ID):
             remote_resource_type = _NORMALIZED_BEDROCK_SERVICE_NAME + "::KnowledgeBase"
             remote_resource_identifier = _escape_delimiters(span.attributes.get(AWS_BEDROCK_KNOWLEDGE_BASE_ID))
-            cloudformation_primary_identifier = ""
+            cloudformation_primary_identifier = remote_resource_identifier
         elif is_key_present(span, GEN_AI_REQUEST_MODEL):
             remote_resource_type = _NORMALIZED_BEDROCK_SERVICE_NAME + "::Model"
             remote_resource_identifier = _escape_delimiters(span.attributes.get(GEN_AI_REQUEST_MODEL))
-            cloudformation_primary_identifier = ""
+            # TODO: AWS::Bedrock::Model is not currently listed in CloudControl API
+            # Going with this assumption for now. May need to change later.
+            cloudformation_primary_identifier = remote_resource_identifier
         elif is_key_present(span, AWS_SNS_TOPIC_ARN):
             remote_resource_type = _NORMALIZED_SNS_SERVICE_NAME + "::Topic"
             remote_resource_identifier = _escape_delimiters(span.attributes.get(AWS_SNS_TOPIC_ARN))

aws-opentelemetry-distro/src/amazon/opentelemetry/distro/patches/_bedrock_patches.py

@@ -126,6 +126,7 @@ class _DataSourceOperation(_BedrockAgentOperation):
 
     request_attributes = {
         AWS_BEDROCK_DATA_SOURCE_ID: _DATA_SOURCE_ID,
+        AWS_BEDROCK_KNOWLEDGE_BASE_ID: _KNOWLEDGE_BASE_ID
     }
     response_attributes = {
         AWS_BEDROCK_DATA_SOURCE_ID: _DATA_SOURCE_ID,