Return reason in error message as to why LoRA adapter could not be loaded

The reason why a LoRA adapter could not be loaded may include information
from model validation, such as that model signature verification did not
succeed because unsigned files were found.

Signed-off-by: Stefan Berger <[email protected]>

Author: stefanberger

vllm/entrypoints/openai/serving_models.py

@@ -249,6 +249,7 @@ async def resolve_lora(self, lora_name: str) -> Union[LoRARequest, ErrorResponse
             base_model_name = self.model_config.model
             unique_id = self.lora_id_counter.inc(1)
             found_adapter = False
+            reason = ""
 
             # Try to resolve using available resolvers
             for resolver in self.lora_resolvers:
@@ -275,13 +276,15 @@ async def resolve_lora(self, lora_name: str) -> Union[LoRARequest, ErrorResponse
                             resolver.__class__.__name__,
                             e,
                         )
+                        reason = str(e)
                         continue
 
             if found_adapter:
                 # An adapter was found, but all attempts to load it failed.
                 return create_error_response(
                     message=(
-                        f"LoRA adapter '{lora_name}' was found but could not be loaded."
+                        f"LoRA adapter '{lora_name}' was found "
+                        f"but could not be loaded: {reason}"
                     ),
                     err_type="BadRequestError",
                     status_code=HTTPStatus.BAD_REQUEST,