#Getting connection pool is full error

1 messages · Page 1 of 1 (latest)

sturdy dome
#

[2025-07-16 18:45:47,481][WARNING] Connection pool is full, discarding connection: backend.composio.dev. Connection pool size: 10
[2025-07-16 18:45:47,503][WARNING] Connection pool is full, discarding connection: backend.composio.dev. Connection pool size: 10
[2025-07-16 18:45:47,512][WARNING] Connection pool is full, discarding connection: backend.composio.dev. Connection pool size: 10
[2025-07-16 18:45:47,537][WARNING] Connection pool is full, discarding connection: backend.composio.dev. Connection pool size: 10
[2025-07-16 18:45:47,654][WARNING] Connection pool is full, discarding connection: backend.composio.dev. Connection pool size: 10

I'm testing composio for my startup and this is th eissue i'm getting using langchain and the composio toolset

random hatchBOT
#

Hey, Could you share your reproducible code snippet? It would help us understand the issue and provide resolution!

sturdy dome
#

llm = ChatOpenAI(
    model="gpt-4o",
    temperature=0,
    api_key=OPENAI_API_KEY,
)
composio_toolset = ComposioToolSet(llm=llm)


tools = composio_toolset.get_tools(
    entity_id="[email protected]",
    # apps=[App.LINEAR],
    # tags=["EXA_READONLYHINT"],
    actions=[
        Action.LINEAR_GET_LINEAR_ISSUE,
        Action.LINEAR_LIST_LINEAR_ISSUES,
        Action.LINEAR_LIST_LINEAR_STATES,
        Action.LINEAR_LIST_LINEAR_PROJECTS,
        Action.LINEAR_LIST_LINEAR_TEAMS,
    ],
)
print(tools)
agent = create_react_agent(llm, tools)
system_message = SystemMessage(
        content=(
            "You have access to multiple tools that can help answer queries. "
            "Use them dynamically and efficiently based on the user's request. "
            "Todays date is " + datetime.now().strftime("%Y-%m-%d")
        )
    )
task = "Can you tell me all the issues are in progress on linear?"
async for chunk in agent.astream(
    {"messages": [system_message, HumanMessage(content=task)]}
):
    core = ''
    if 'agent' in chunk:
        core = 'agent'
    elif 'tools' in chunk:
        core = 'tools'
    else:
        print(chunk)
    if core:
        for messages in chunk[core]['messages']:
            print(messages.text())

#

@vestal oyster

random hatchBOT
#

Hey, could you please share the imports you’re currently using along with the SDK version?

sturdy dome
#
from langchain_mcp_adapters.client import MultiServerMCPClient
from langgraph.prebuilt import create_react_agent
from langchain_openai import ChatOpenAI
from composio_langchain import ComposioToolSet, App, Action
from langchain_core.messages import SystemMessage, HumanMessage
from datetime import datetime

let me figure out my sdk version

#
composio-0.1.1.dist-info
composio_langchain-0.7.20.dist-info
composio_core-0.7.20.dist-info
#

should I swap to the newer composio?

random hatchBOT
#

We’ll attempt to reproduce the issue on our end and update you. Yes, switching to version 3 is recommended as it’s more reliable for frameworks like Langchain.

random hatchBOT
#

Hi, We Couldn't reproduce the Issue from our Side. Here is the snippet we use for ref:

from langgraph.prebuilt import create_react_agent
from langchain_openai import ChatOpenAI
from composio_langchain import ComposioToolSet, App, Action
from langchain_core.messages import SystemMessage, HumanMessage
from datetime import datetime

llm = ChatOpenAI(
    model="gpt-4o",
    temperature=0,
    api_key="API-KEY",
)
composio_toolset = ComposioToolSet(api_key="<composio-api-key>",llm=llm)


tools = composio_toolset.get_tools(
    # apps=[App.LINEAR],
    # tags=["EXA_READONLYHINT"],
    actions=[
        Action.LINEAR_GET_LINEAR_ISSUE,
        Action.LINEAR_LIST_LINEAR_ISSUES,
        Action.LINEAR_LIST_LINEAR_STATES,
        Action.LINEAR_LIST_LINEAR_PROJECTS,
        Action.LINEAR_LIST_LINEAR_TEAMS,
    ],
)
print(tools)
agent = create_react_agent(llm, tools)
system_message = SystemMessage(
        content=(
            "You have access to multiple tools that can help answer queries. "
            "Use them dynamically and efficiently based on the user's request. "
            "Todays date is " + datetime.now().strftime("%Y-%m-%d")
        )
    )
async def main():
    task = "Can you tell me all the issues are in progress on linear?"
    async for chunk in agent.astream(
        {"messages": [system_message, HumanMessage(content=task)]}
    ):
        core = ''
        if 'agent' in chunk:
            core = 'agent'
        elif 'tools' in chunk:
            core = 'tools'
        else:
            print(chunk)
        if core:
            for messages in chunk[core]['messages']:
                print(messages.text())

if __name__ == "__main__":
    import asyncio
    asyncio.run(main())

—Let me know if you still face any Issues.

sturdy dome
#

thank you I think after upgrading @vestal oyster i stopped getitng the error