feat: Add configurable timeouts for Ollama and OpenRouter embedding models.

Author: rostilos

python-ecosystem/rag-pipeline/src/rag_pipeline/core/embedding_factory.py

@@ -29,11 +29,12 @@ def create_embedding_model(config: RAGConfig) -> BaseEmbedding:
     provider = config.embedding_provider.lower()
 
     if provider == "ollama":
-        logger.info(f"Creating Ollama embedding model: {config.ollama_model}")
+        timeout = float(os.getenv("OLLAMA_TIMEOUT", "120"))
+        logger.info(f"Creating Ollama embedding model: {config.ollama_model} (timeout={timeout}s)")
         return OllamaEmbedding(
             model=config.ollama_model,
             base_url=config.ollama_base_url,
-            timeout=120.0,
+            timeout=timeout,
             expected_dim=config.embedding_dim
         )
 
@@ -51,10 +52,11 @@ def create_embedding_model(config: RAGConfig) -> BaseEmbedding:
 
     else:
         logger.warning(f"Unknown embedding provider '{provider}', defaulting to Ollama")
+        timeout = float(os.getenv("OLLAMA_TIMEOUT", "120"))
         return OllamaEmbedding(
             model=config.ollama_model,
             base_url=config.ollama_base_url,
-            timeout=120.0,
+            timeout=timeout,
             expected_dim=config.embedding_dim
         )
 

python-ecosystem/rag-pipeline/src/rag_pipeline/core/openrouter_embedding.py

@@ -36,7 +36,7 @@ def __init__(
         api_key: str,
         model: str = "openai/text-embedding-3-small",
         api_base: str = "https://openrouter.ai/api/v1",
-        timeout: float = 60.0,
+        timeout: float = float(os.getenv("OPENROUTER_TIMEOUT", "300")),
         max_retries: int = 3,
         embed_batch_size: int = EMBEDDING_BATCH_SIZE,
         expected_dim: Optional[int] = None,