Hey folks! Is there a way to use aupdate_state in LangGraph without potentially blowing away state like human-in-the-loop interrupts? We’re doing a fairly simple update where we mutate messages as part of a migration: aupdate_state(values={"messages": messages}) We’re noticing that when we do this…

hi @tharshan-elvex here is the results: having this agent invocation: import os from dotenv import load_dotenv from langchain_core.messages import SystemMessage, HumanMessage from langgraph.checkpoint.postgres import PostgresSaver, ShallowPostgresSaver from langgraph.graph import START, StateGr…

@tharshan-elvex is this something that could help figure out the migration? Small update - to migrate the input field too: for tool_call in msg.tool_calls: if tool_call['name'] == 'add': # Rewrite to add_v2 with new signature ne…

Hey @pawel-twardziak Is there a finding here specific to preserving the interrupt state? Right now we have a working migration script, that we have validated. The state that gets migrated and written is correct for the the v2 of the tool. However, to bring back the initial focus of the issue at ha…

hi @tharshan-elvex yes, you are right, I lost track I will follow up tomorrow, already too tired today

hi @tharshan-elvex I am back :slight_smile: Question: do any threads have pending tasks (interrupts) in checkpointer tables?

Hello @pawel-twardziak , yes if I run this query SELECT * FROM "public"."checkpoint_writes" WHERE "channel"::TEXT LIKE '%__interrupt__%' LIMIT 100 OFFSET 0; Which returns 62 rows. Which I believe tells us there are 62 threads currently at the interrupt state.

Hi @tharshan-elvex Ok, thanks for the answer, will continue on investigating :+1:

Hi @tharshan-elvex I am checking in here to keep you in the loop - I still can’t find any native way to proceed with the migration. Each approach wipes out all pending tasks so far. I will surely inform you if any findings emerge :slight_smile:

Thanks for the update @pawel-twardziak . What other options do we have if there is no native way?

Hi @tharshan-elvex you mentioned that you already have a script that does the migration in the correct way, am I right?

How to update graph state while preserving interrupts?

OSS Product Help LangGraph

tharshan-elvex October 21, 2025, 6:09pm 36

After:

branch:to:llm is lost
two interrupt channels exist, vs a replacement
messages channel task_path is lost

Topic		Replies	Views
Restoring to checkpoint doesn't resume from node of checkpoint LangGraph python-help	3	370	November 21, 2025
How to time travel to before interrupt and resume with a different value? LangGraph js-help	11	411	December 21, 2025
Persisting HITL payloads LangGraph self-hosted , python-help	7	149	April 20, 2026
MESSAGE_COERCION_FAILURE Using Redis Checkpointer with langraph LangGraph python-help	2	512	September 28, 2025
Request for Guidance: When using the Interrupts mechanism in LangGraph, how should the graph utilize ainvoke? When multiple interrupts exist, how to integrate (the Interrupts mechanism) with other frameworks (e.g., the FastAPI framework)? LangChain Academy python-help	9	285	January 29, 2026

How to update graph state while preserving interrupts?

Related topics