Probably Concurency Issues with the default SQLAlchemyDataLayer,

[rudolf-erdei]

Describe the bug
I'm starting out with Chainlit, learning how to create a langchain Agent with it. After setting up the login part, using the example implementation from the documentation (SQLAlchemyDataLayer with BaseStorageClient, using it with docker image postres:latest),

To Reproduce
Steps to reproduce the behavior:

1. Implement database saving with SQLAlchemyDataLayer and BaseStorageClient. Using any user and pass, as it will be just for saving conversations.
2. Define a LangChain agent (from langchain.agents import Tool, initialize_agent), with DuckDuckFoSearchRun (from langchain_community.tools import DuckDuckGoSearchRun)
3. Try a complex query, which will force the platform to create more than one step.

Expected behavior
The platform working smootly, without issues and returning a relevant result.

Screenshots
Not applicable. Log entries below.

Desktop (please complete the following information):

• OS: Windows 11
• Browser: Firefox (although not relevant, as problem is in the backend
• Version [e.g. 22]

Smartphone (please complete the following information):

• Device: [e.g. iPhone6]
• OS: [e.g. iOS8.1]
• Browser [e.g. stock browser, safari]
• Version: "aiohttp>=3.11.12",
  "asyncpg>=0.30.0",
  "chainlit>=2.2.1",
  "greenlet>=3.1.1",
  "langchain-google-genai>=2.0.9",
  "langchain>=0.3.18",
  "sqlalchemy>=2.0.38",
  "langchain-community>=0.3.17",
  "wikipedia>=1.4.0",
  "duckduckgo-search>=7.4.2",
  "numexpr>=2.10.2",

Additional context
Error logs extract:

025-02-17 14:46:11 - SQLAlchemy: create_step, step_id=fd6923c1-efb9-449a-a57a-e8a70ac83c53
Thought: The prompt asks for use for a 3D printer in my home. It also provides a format for the output, including an execution plan and rest of needed information. I need to create an execution plan to find the most creative use for a 3D printer and then provide any additional information.

Action:

{
  "action": "Web Search",
  "action_input": "uses for 3D printer at home"
}
```2025-02-17 14:46:11 - SQLAlchemy: create_step, step_id=bc940f63-791b-44b9-8166-c2982952db47
2025-02-17 14:46:11 - SQLAlchemy: create_step, step_id=260c53bc-bcdd-45a0-bea8-db23ce7b32a2
2025-02-17 14:46:11 - An unexpected error occurred: Task <Task pending name='Task-918' coro=<SQLAlchemyDataLayer.create_step() running at D:\Projects\researcher\.venv\Lib\site-packages\chainlit\data\utils.py:25>> got Future <Future pending cb=[BaseProtocol._on_waiter_completed()]> attached to a different loop
2025-02-17 14:46:12 - HTTP Request: POST https://lite.duckduckgo.com/lite/ "HTTP/2 200 OK"
2025-02-17 14:46:12 - SQLAlchemy: update_step, step_id=260c53bc-bcdd-45a0-bea8-db23ce7b32a2
2025-02-17 14:46:12 - SQLAlchemy: create_step, step_id=260c53bc-bcdd-45a0-bea8-db23ce7b32a2
2025-02-17 14:46:12 - An unexpected error occurred: Task <Task pending name='Task-937' coro=<SQLAlchemyDataLayer.update_step() running at D:\Projects\researcher\.venv\Lib\site-packages\chainlit\data\utils.py:25>> got Future <Future pending cb=[BaseProtocol._on_waiter_completed()]> attached to a different loop
2025-02-17 14:46:12 - SQLAlchemy: update_step, step_id=bc940f63-791b-44b9-8166-c2982952db47

further down...

2025-02-17 14:46:19 - SQLAlchemy: create_step, step_id=0953c1ba-0c8a-441a-b6cb-ef976ba55ebf
2025-02-17 14:46:20 - An error occurred: (sqlalchemy.dialects.postgresql.asyncpg.Error) <class 'asyncpg.exceptions.DeadlockDetectedError'>: deadlock detected
DETAIL:  Process 1364 waits for ShareLock on transaction 1131; blocked by process 1361.
Process 1361 waits for ShareLock on transaction 1136; blocked by process 1364.
HINT:  See server log for query details.
[SQL:
            INSERT INTO steps ("name", "type", "id", "threadId", "parentId", "streaming", "metadata", "input", "isError", "output", "createdAt", "start", "end", "defaultOpen", "showInput", "generation")
            VALUES ($1, $2, $3, $4, $5, $6, $7, $8, $9, $10, $11, $12, $13, $14, $15, $16)
            ON CONFLICT (id) DO UPDATE
            SET "name" = $1, "type" = $2, "threadId" = $4, "parentId" = $5, "streaming" = $6, "metadata" = $7, "input" = $8, "isError" = $9, "output" = $10, "createdAt" = $11, "start" = $12, "end" = $13, "defaultOpen" = $14, "showInput" = $15, "generation" = $16;
        ]
[parameters: ('LLMChain', 'undefined', '5fcd21be-b8fc-4b04-a67a-3e3d2b4a3d1f', '7c8b7eb2-018f-4493-a07f-6e4fd975148e', 'ea756a5d-5c88-443e-ae14-5b92a1c4997e', False, '{"language": "json"}', '{\'input\': StringPromptValue(text="Prepare plan for task execution. (e.g. retrieve current date to find weather forecast)\n\n    Tools to use: wikip ... (7443 characters truncated) ...  With rising demand, customizable tools offer personalized solutions, merging function with unique design.\nThought:\', \'stop\': [\'Observation:\']}', False, '```json\n{\n  "action": "Web Search",\n  "action_input": "innovative 3D printed home solutions"\n}\n```', '2025-02-17T12:46:18.436626Z', '2025-02-17T12:46:18.436626Z', '2025-02-17T12:46:19.096666Z', False, 'json', 'null')]
(Background on this error at: https://sqlalche.me/e/20/dbapi)