add the option of deleting an existing index during onboarding

src/api/settings.py

@@ -67,6 +67,7 @@ class OnboardingBody(BaseModel):
     llm_model: Optional[str] = Field(None, min_length=1)
     embedding_provider: Optional[str] = Field(None, pattern="^(openai|watsonx|ollama)$")
     embedding_model: Optional[str] = Field(None, min_length=1)
+    delete_existing_index: Optional[bool] = None
     openai_api_key: Optional[str] = Field(None, min_length=1)
     anthropic_api_key: Optional[str] = Field(None, min_length=1)
     watsonx_api_key: Optional[str] = Field(None, min_length=1)
@@ -1110,10 +1111,21 @@ async def onboarding(
                 # Import here to avoid circular imports
                 from main import init_index_when_ready
 
+                # Handle delete_existing_index
+                delete_existing_index = False
+                if body.delete_existing_index:
+                    delete_existing_index = True
+                    await TelemetryClient.send_event(
+                        Category.ONBOARDING,
+                        MessageId.ORB_ONBOARD_DELETE_EXISTING_INDEX
+                    )
+                    logger.info("Delete existing index requested during onboarding")
+
                 logger.info(
                     "Initializing OpenSearch index after onboarding configuration"
                 )
-                await init_index_when_ready()
+                await init_index_when_ready(delete_existing=delete_existing_index)
+
                 logger.info("OpenSearch index initialization completed successfully")
             except Exception as e:
                 logger.error(

src/main.py

@@ -213,7 +213,7 @@ async def _ensure_opensearch_index():
         # The service can still function, document operations might fail later
 
 
-async def init_index():
+async def init_index(delete_existing: bool = False):
     """Initialize OpenSearch index and security roles"""
     try:
         await wait_for_opensearch()
@@ -232,9 +232,17 @@ async def init_index():
             endpoint=getattr(embedding_provider_config, "endpoint", None),
         )
 
-        # Create documents index
         index_name = get_index_name()
-        if not await clients.opensearch.indices.exists(index=index_name):
+        index_exists = await clients.opensearch.indices.exists(index=index_name)
+        if index_exists and delete_existing:
+            # Asked to delete the existing index ..
+            logger.info(f"Deleting index '{index_name}'...")
+            resp = await clients.opensearch.indices.delete(index=index_name)
+            logger.info(f"Deleted index '{index_name}', response: {resp}")
+            index_exists = False
+
+        # Create documents index
+        if not index_exists:
             await clients.opensearch.indices.create(
                 index=index_name, body=dynamic_index_body
             )
@@ -322,10 +330,10 @@ async def init_index():
         raise e
 
 
-async def init_index_when_ready():
+async def init_index_when_ready(delete_existing: bool = False):
     """Wait for the OpenSearch service to be ready and then initialize the OpenSearch index."""
     await wait_for_opensearch()
-    await init_index()
+    await init_index(delete_existing)
 
 
 def generate_jwt_keys():

src/utils/telemetry/message_id.py

@@ -213,6 +213,8 @@ class MessageId:
     ORB_ONBOARD_EMBED_MODEL = "ORB_ONBOARD_EMBED_MODEL"
     # Message: Sample data ingestion requested
     ORB_ONBOARD_SAMPLE_DATA = "ORB_ONBOARD_SAMPLE_DATA"
+    # Message: Delete existing index requested
+    ORB_ONBOARD_DELETE_EXISTING_INDEX = "ORB_ONBOARD_DELETE_EXISTING_INDEX"
     # Message: Configuration marked as edited
     ORB_ONBOARD_CONFIG_EDITED = "ORB_ONBOARD_CONFIG_EDITED"
     # Message: Onboarding rolled back due to all files failing