production improvements: configurable embeddings, v1.1, O(1) ownership, retries

- Make Ollama URL configurable via OLLAMA_BASE_URL env var
- Add version: v1.1 to Mem0 config (required for latest features)
- Make embedding model and dimensions configurable
- Fix ownership check: O(1) lookup instead of fetching 10k records
- Add tenacity retry logic for database operations

backend/config.py

@@ -11,39 +11,87 @@ class Settings(BaseSettings):
     """Application settings loaded from environment variables."""
 
     model_config = SettingsConfigDict(
-        env_file=".env",
+        env_file=".env", case_sensitive=False, extra="ignore"
-        case_sensitive=False,
-        extra='ignore'
     )
 
     # API Configuration
     # Accept both OPENAI_API_KEY (from docker-compose) and OPENAI_COMPAT_API_KEY (from direct .env)
-    openai_api_key: str = Field(validation_alias=AliasChoices('OPENAI_API_KEY', 'OPENAI_COMPAT_API_KEY', 'openai_api_key'))
+    openai_api_key: str = Field(
-    openai_base_url: str = Field(validation_alias=AliasChoices('OPENAI_BASE_URL', 'OPENAI_COMPAT_BASE_URL', 'openai_base_url'))
+        validation_alias=AliasChoices(
-    cohere_api_key: str = Field(validation_alias=AliasChoices('COHERE_API_KEY', 'cohere_api_key'))
+            "OPENAI_API_KEY", "OPENAI_COMPAT_API_KEY", "openai_api_key"
+        )
+    )
+    openai_base_url: str = Field(
+        validation_alias=AliasChoices(
+            "OPENAI_BASE_URL", "OPENAI_COMPAT_BASE_URL", "openai_base_url"
+        )
+    )
+    cohere_api_key: str = Field(
+        validation_alias=AliasChoices("COHERE_API_KEY", "cohere_api_key")
+    )
 
     # Database Configuration
-    qdrant_host: str = Field(default="localhost", validation_alias=AliasChoices('QDRANT_HOST', 'qdrant_host'))
+    qdrant_host: str = Field(
-    qdrant_port: int = Field(default=6333, validation_alias=AliasChoices('QDRANT_PORT', 'qdrant_port'))
+        default="localhost", validation_alias=AliasChoices("QDRANT_HOST", "qdrant_host")
-    qdrant_collection_name: str = Field(default="mem0", validation_alias=AliasChoices('QDRANT_COLLECTION_NAME', 'qdrant_collection_name'))
+    )
+    qdrant_port: int = Field(
+        default=6333, validation_alias=AliasChoices("QDRANT_PORT", "qdrant_port")
+    )
+    qdrant_collection_name: str = Field(
+        default="mem0",
+        validation_alias=AliasChoices(
+            "QDRANT_COLLECTION_NAME", "qdrant_collection_name"
+        ),
+    )
 
     # Neo4j Configuration
-    neo4j_uri: str = Field(default="bolt://localhost:7687", validation_alias=AliasChoices('NEO4J_URI', 'neo4j_uri'))
+    neo4j_uri: str = Field(
-    neo4j_username: str = Field(default="neo4j", validation_alias=AliasChoices('NEO4J_USERNAME', 'neo4j_username'))
+        default="bolt://localhost:7687",
-    neo4j_password: str = Field(default="mem0_neo4j_password", validation_alias=AliasChoices('NEO4J_PASSWORD', 'neo4j_password'))
+        validation_alias=AliasChoices("NEO4J_URI", "neo4j_uri"),
+    )
+    neo4j_username: str = Field(
+        default="neo4j",
+        validation_alias=AliasChoices("NEO4J_USERNAME", "neo4j_username"),
+    )
+    neo4j_password: str = Field(
+        default="mem0_neo4j_password",
+        validation_alias=AliasChoices("NEO4J_PASSWORD", "neo4j_password"),
+    )
 
     # Application Configuration
-    log_level: str = Field(default="INFO", validation_alias=AliasChoices('LOG_LEVEL', 'log_level'))
+    log_level: str = Field(
-    cors_origins: str = Field(default="http://localhost:3000", validation_alias=AliasChoices('CORS_ORIGINS', 'cors_origins'))
+        default="INFO", validation_alias=AliasChoices("LOG_LEVEL", "log_level")
+    )
+    cors_origins: str = Field(
+        default="http://localhost:3000",
+        validation_alias=AliasChoices("CORS_ORIGINS", "cors_origins"),
+    )
 
     # Model Configuration - Ultra-minimal (single model)
-    default_model: str = Field(default="claude-sonnet-4", validation_alias=AliasChoices('DEFAULT_MODEL', 'default_model'))
+    default_model: str = Field(
+        default="claude-sonnet-4",
+        validation_alias=AliasChoices("DEFAULT_MODEL", "default_model"),
+    )
+
+    # Embedder Configuration
+    ollama_base_url: str = Field(
+        default="http://host.docker.internal:11434",
+        validation_alias=AliasChoices("OLLAMA_BASE_URL", "ollama_base_url"),
+    )
+    embedding_model: str = Field(
+        default="qwen3-embedding:4b-q8_0",
+        validation_alias=AliasChoices("EMBEDDING_MODEL", "embedding_model"),
+    )
+    embedding_dims: int = Field(
+        default=2560, validation_alias=AliasChoices("EMBEDDING_DIMS", "embedding_dims")
+    )
 
     # Authentication Configuration
     # Format: JSON string mapping API keys to user IDs
     # Example: {"api_key_123": "alice", "api_key_456": "bob"}
-    api_keys: str = Field(default="{}", validation_alias=AliasChoices('API_KEYS', 'api_keys'))
+    api_keys: str = Field(
-
+        default="{}", validation_alias=AliasChoices("API_KEYS", "api_keys")
+    )
 
     @property
     def cors_origins_list(self) -> List[str]:

backend/main.py

@@ -464,13 +464,10 @@ async def update_memory(
                 detail=f"Access denied: You can only update your own memories (authenticated as '{authenticated_user}')",
             )
 
-        # Verify the memory actually belongs to the authenticated user
+        # Verify memory ownership with O(1) lookup instead of fetching all memories
-        user_memories = await mem0_manager.get_user_memories(
+        if not await mem0_manager.verify_memory_ownership(
-            user_id=authenticated_user, limit=10000
+            update_request.memory_id, authenticated_user
-        )
+        ):
-        memory_ids = {m.get("id") for m in user_memories if m.get("id")}
-
-        if update_request.memory_id not in memory_ids:
             raise HTTPException(
                 status_code=404,
                 detail=f"Memory '{update_request.memory_id}' not found or access denied",
@@ -506,13 +503,10 @@ async def delete_memory(
 ):
     """Delete a specific memory - verifies ownership before deletion."""
     try:
-        # Verify the memory actually belongs to the authenticated user
+        # Verify memory ownership with O(1) lookup instead of fetching all memories
-        user_memories = await mem0_manager.get_user_memories(
+        if not await mem0_manager.verify_memory_ownership(
-            user_id=authenticated_user, limit=10000
+            memory_id, authenticated_user
-        )
+        ):
-        memory_ids = {m.get("id") for m in user_memories if m.get("id")}
-
-        if memory_id not in memory_ids:
             raise HTTPException(
                 status_code=404,
                 detail=f"Memory '{memory_id}' not found or access denied",
@@ -614,13 +608,8 @@ async def get_memory_history(
                 detail=f"Access denied: You can only view your own memory history (authenticated as '{authenticated_user}')",
             )
 
-        # Verify the memory belongs to this user before returning history
+        # Verify memory ownership with O(1) lookup instead of fetching all memories
-        user_memories = await mem0_manager.get_user_memories(
+        if not await mem0_manager.verify_memory_ownership(memory_id, user_id):
-            user_id=user_id, limit=10000
-        )
-        memory_ids = {m.get("id") for m in user_memories if m.get("id")}
-
-        if memory_id not in memory_ids:
             raise HTTPException(
                 status_code=404,
                 detail=f"Memory '{memory_id}' not found or access denied",

backend/mem0_manager.py

@@ -5,12 +5,28 @@ from typing import Dict, List, Optional, Any
 from datetime import datetime
 from mem0 import Memory
 from openai import OpenAI
+from tenacity import (
+    retry,
+    stop_after_attempt,
+    wait_exponential,
+    retry_if_exception_type,
+    before_sleep_log,
+)
 
 from config import settings
 from monitoring import timed
 
 logger = logging.getLogger(__name__)
 
+# Retry decorator for database operations (Qdrant, Neo4j)
+db_retry = retry(
+    stop=stop_after_attempt(3),
+    wait=wait_exponential(multiplier=1, min=1, max=10),
+    retry=retry_if_exception_type((ConnectionError, TimeoutError, OSError)),
+    before_sleep=before_sleep_log(logger, logging.WARNING),
+    reraise=True,
+)
+
 # Monkey-patch Mem0's OpenAI LLM to remove the 'store' parameter for LiteLLM compatibility
 from mem0.llms.openai import OpenAILLM
 
@@ -48,6 +64,7 @@ class Mem0Manager:
             settings,
         )
         config = {
+            "version": "v1.1",
             "enable_graph": True,
             "llm": {
                 "provider": "openai",
@@ -56,16 +73,15 @@ class Mem0Manager:
                     "api_key": settings.openai_api_key,
                     "openai_base_url": settings.openai_base_url,
                     "temperature": 0.1,
-                    "top_p": None,  # Don't use top_p with Claude models
+                    "top_p": None,
                 },
             },
             "embedder": {
                 "provider": "ollama",
                 "config": {
-                    "model": "qwen3-embedding:4b-q8_0",
+                    "model": settings.embedding_model,
-                    # "api_key": settings.embedder_api_key,
+                    "ollama_base_url": settings.ollama_base_url,
-                    "ollama_base_url": "http://172.17.0.1:11434",
+                    "embedding_dims": settings.embedding_dims,
-                    "embedding_dims": 2560,
                 },
             },
             "vector_store": {
@@ -74,7 +90,7 @@ class Mem0Manager:
                     "collection_name": settings.qdrant_collection_name,
                     "host": settings.qdrant_host,
                     "port": settings.qdrant_port,
-                    "embedding_model_dims": 2560,
+                    "embedding_model_dims": settings.embedding_dims,
                     "on_disk": True,
                 },
             },
@@ -106,6 +122,7 @@ class Mem0Manager:
         logger.info("Initialized ultra-minimal Mem0Manager with custom endpoint")
 
     # Pure passthrough methods - no custom logic
+    @db_retry
     @timed("add_memories")
     async def add_memories(
         self,
@@ -161,6 +178,7 @@ class Mem0Manager:
             logger.error(f"Error adding memories: {e}")
             raise
 
+    @db_retry
     @timed("search_memories")
     async def search_memories(
         self,
@@ -204,6 +222,7 @@ class Mem0Manager:
             logger.error(f"Error searching memories: {e}")
             raise
 
+    @db_retry
     async def get_user_memories(
         self,
         user_id: str,
@@ -227,6 +246,24 @@ class Mem0Manager:
             logger.error(f"Error getting user memories: {e}")
             raise
 
+    @db_retry
+    async def get_memory(self, memory_id: str) -> Optional[Dict[str, Any]]:
+        """Get a single memory by ID. Returns None if not found."""
+        try:
+            result = self.memory.get(memory_id=memory_id)
+            return result
+        except Exception as e:
+            logger.debug(f"Memory {memory_id} not found or error: {e}")
+            return None
+
+    async def verify_memory_ownership(self, memory_id: str, user_id: str) -> bool:
+        """Check if a memory belongs to a user. O(1) instead of O(n)."""
+        memory = await self.get_memory(memory_id)
+        if memory is None:
+            return False
+        return memory.get("user_id") == user_id
+
+    @db_retry
     @timed("update_memory")
     async def update_memory(
         self,
@@ -241,6 +278,7 @@ class Mem0Manager:
             logger.error(f"Error updating memory: {e}")
             raise
 
+    @db_retry
     @timed("delete_memory")
     async def delete_memory(self, memory_id: str) -> Dict[str, Any]:
         """Delete memory - pure Mem0 passthrough."""

backend/requirements.txt

@@ -19,6 +19,7 @@ ollama
 # Utilities
 pydantic
 pydantic-settings
+tenacity
 python-dotenv
 httpx
 aiofiles