langchain-ai/langgraph
1
0
Fork 0
mirror of https://github.com/langchain-ai/langgraph.git synced 2026-07-15 12:35:43 -04:00
Code Issues 314 Packages Projects Releases 17 Wiki Activity

graph.astream checkpoint_id not working properly #519

New Issue
Closed
opened 2026-02-20 17:40:32 -05:00 by yindo · 5 comments
yindo commented 2026-02-20 17:40:32 -05:00
Owner
Copy Link
Copy Source

Originally created by @us on GitHub (Mar 18, 2025).

Checked other resources

  • This is a bug, not a usage question. For questions, please use GitHub Discussions.
  • I added a clear and detailed title that summarizes the issue.
  • I read what a minimal reproducible example is (https://stackoverflow.com/help/minimal-reproducible-example).
  • I included a self-contained, minimal example that demonstrates the issue INCLUDING all the relevant imports. The code run AS IS to reproduce the issue.

Example Code

async def astream_graph(state, config):
    try:
        async with AsyncMongoDBSaver.from_conn_string(os.getenv("MONGODB_URI"),
                                                      db_name=os.getenv("MONGODB_DB_NAME")) as checkpointer:
            graph = build_graph(checkpointer)
            # print("graph.get_state_history(config)", [state for state in graph.get_state_history(config)])
            async for event in graph.astream(state, config, stream_mode=["messages", "debug", "custom"]):
                yield event
    except Exception as e:
        print(f"Error in streaming: {str(e)}")
        raise

def build_graph(checkpointer):
    """Build and return the graph with tools."""
    try:
        workflow = StateGraph(AgentState)

        workflow.add_node("agent", call_model)
        workflow.add_node("tools", call_tools)
        workflow.add_edge(START, "agent")

        workflow.add_conditional_edges(
            "agent",
            should_continue,
            {
                "tools": "tools",
                END: END
            }
        )

        workflow.add_edge("tools", "agent")

        return workflow.compile(checkpointer=checkpointer)
    except Exception as e:
        print(f"Error building graph: {str(e)}")
        raise

async def call_model(state: AgentState) -> Dict:
    """Process messages and get model response."""
    try:
        messages = state["messages"]
        ...
        response = await model.ainvoke(messages)
    if hasattr(response, 'additional_kwargs') and 
       response.additional_kwargs.get('tool_calls'):
            return {"messages": [response]}

        return {"messages": [response]}

    except Exception as e:
        print(f"Error in call_model: {str(e)}")
        raise e

async def call_tools(state: AgentState) -> Dict:
    """Execute tools asynchronously."""
    try:
        messages = state["messages"]
        last_message = cast(AIMessage, messages[-1])
        ...
        result = await tool_node.ainvoke({"messages": [last_message]})
        # The result contains tool messages that we can return directly
        return {"messages": result["messages"]}

    except Exception as e:
        print(f"Error in call_tools: {str(e)}")
        raise e

Error Message and Stack Trace (if applicable)

Description

When I send checkpoint_id inside config['configurable'], its resume from there, i was expecting the continue in from that checkpoint like messages should properly show only the prev messages from that checkpoint, but its gives me full messages!

System Info

osx/linux

Originally created by @us on GitHub (Mar 18, 2025). ### Checked other resources - [x] This is a bug, not a usage question. For questions, please use GitHub Discussions. - [x] I added a clear and detailed title that summarizes the issue. - [x] I read what a minimal reproducible example is (https://stackoverflow.com/help/minimal-reproducible-example). - [x] I included a self-contained, minimal example that demonstrates the issue INCLUDING all the relevant imports. The code run AS IS to reproduce the issue. ### Example Code ```python async def astream_graph(state, config): try: async with AsyncMongoDBSaver.from_conn_string(os.getenv("MONGODB_URI"), db_name=os.getenv("MONGODB_DB_NAME")) as checkpointer: graph = build_graph(checkpointer) # print("graph.get_state_history(config)", [state for state in graph.get_state_history(config)]) async for event in graph.astream(state, config, stream_mode=["messages", "debug", "custom"]): yield event except Exception as e: print(f"Error in streaming: {str(e)}") raise def build_graph(checkpointer): """Build and return the graph with tools.""" try: workflow = StateGraph(AgentState) workflow.add_node("agent", call_model) workflow.add_node("tools", call_tools) workflow.add_edge(START, "agent") workflow.add_conditional_edges( "agent", should_continue, { "tools": "tools", END: END } ) workflow.add_edge("tools", "agent") return workflow.compile(checkpointer=checkpointer) except Exception as e: print(f"Error building graph: {str(e)}") raise async def call_model(state: AgentState) -> Dict: """Process messages and get model response.""" try: messages = state["messages"] ... response = await model.ainvoke(messages) if hasattr(response, 'additional_kwargs') and response.additional_kwargs.get('tool_calls'): return {"messages": [response]} return {"messages": [response]} except Exception as e: print(f"Error in call_model: {str(e)}") raise e async def call_tools(state: AgentState) -> Dict: """Execute tools asynchronously.""" try: messages = state["messages"] last_message = cast(AIMessage, messages[-1]) ... result = await tool_node.ainvoke({"messages": [last_message]}) # The result contains tool messages that we can return directly return {"messages": result["messages"]} except Exception as e: print(f"Error in call_tools: {str(e)}") raise e ``` ### Error Message and Stack Trace (if applicable) ```shell ``` ### Description When I send checkpoint_id inside config['configurable'], its resume from there, i was expecting the continue in from that checkpoint like messages should properly show only the prev messages from that checkpoint, but its gives me full messages! ### System Info osx/linux
yindo added the question label 2026-02-20 17:40:32 -05:00
yindo closed this issue 2026-02-20 17:40:32 -05:00
yindo commented 2026-02-20 17:40:32 -05:00
Author
Owner
Copy Link
Copy Source

@vbarda commented on GitHub (Mar 18, 2025):

your example is incomplete - please provide how you're actually invoking the graph

from your description it's unclear whether you're trying to fork or replay https://langchain-ai.github.io/langgraph/concepts/time-travel/, but you shouldn't be seeing full messages in either of the two cases.

also, i would also recommend running your code with langgraph.checkpoint.memory.MemorySaver and verifying that it's not an issue with MongoDBSaver

@vbarda commented on GitHub (Mar 18, 2025): your example is incomplete - please provide how you're actually invoking the graph from your description it's unclear whether you're trying to fork or replay https://langchain-ai.github.io/langgraph/concepts/time-travel/, but you shouldn't be seeing full messages in either of the two cases. also, i would also recommend running your code with `langgraph.checkpoint.memory.MemorySaver` and verifying that it's not an issue with MongoDBSaver
yindo commented 2026-02-20 17:40:32 -05:00
Author
Owner
Copy Link
Copy Source

@us commented on GitHub (Mar 19, 2025):

ah sorry, i missed! here how i am invoking that:
(yes i am giving 2 times the config! just to make sure for all cases i need to clean, i know..)

config = {
    "user_id": user_id,
    "thread_id": thread_id,
    "checkpoint_id": checkpoint_id
}

state = {
    "messages": messages,
    "config": config
}

astream_graph(
    state,
    {'configurable': config},
)


class AgentState(TypedDict):
    """State for the agent."""
    messages: Annotated[Sequence[BaseMessage], add]
    config: Dict
@us commented on GitHub (Mar 19, 2025): ah sorry, i missed! here how i am invoking that: *(yes i am giving 2 times the config! just to make sure for all cases i need to clean, i know..)* ```python config = { "user_id": user_id, "thread_id": thread_id, "checkpoint_id": checkpoint_id } state = { "messages": messages, "config": config } astream_graph( state, {'configurable': config}, ) class AgentState(TypedDict): """State for the agent.""" messages: Annotated[Sequence[BaseMessage], add] config: Dict ```
yindo commented 2026-02-20 17:40:32 -05:00
Author
Owner
Copy Link
Copy Source

@vbarda commented on GitHub (Apr 11, 2025):

what's in add? i would recommend using the built-in messages reducer from langgraph.graph.message import add_messages

@vbarda commented on GitHub (Apr 11, 2025): what's in `add`? i would recommend using the built-in messages reducer `from langgraph.graph.message import add_messages`
yindo commented 2026-02-20 17:40:32 -05:00
Author
Owner
Copy Link
Copy Source

@nfcampos commented on GitHub (May 23, 2025):

Have you tried with other checkpointers, eg. the sqlite checkpointer?

@nfcampos commented on GitHub (May 23, 2025): Have you tried with other checkpointers, eg. the sqlite checkpointer?
yindo commented 2026-02-20 17:40:32 -05:00
Author
Owner
Copy Link
Copy Source

@us commented on GitHub (May 25, 2025):

working now!

@us commented on GitHub (May 25, 2025): working now!
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
agents
awaiting author revision
bug
change
chat models
checkpointer
ci
code quality
code quality
dependencies
documentation
drawing
duplicate
enhancement
good first issue
help wanted
invalid
maintainer
meta
missing information
not maintained by langchain
open-swe-auto
open-swe-max
open-swe-max-auto
pending
performance
pull-request
Mirrored from GitHub Pull Request
 question
requires design
sdk
server
stale
studio
tests
type checking
usage question
v0.5
v1.1
No Label question
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: langchain-ai/langgraph#519
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?