Human in the loop interrupt with Redis Saver is not working when interrupt is called inside a tool in subgraph

[adityarana420]

I have a main graph connected to various subgraphs. I invoke main graph like:

if message.get('resume'):
      _, events = MasterAssistant().get_graph().invoke(Command(resume=query), conversation_config, subgraphs=True)
else:
      _, events = MasterAssistant().get_graph().invoke(query_state, conversation_config, subgraphs=True)

The issue is, my interrupt is happening inside a subgraph connected to Master Assistant. On running the second interrupt command. I get langgraph redis issue:

 File ".../redis_helper.py", line 56, in parse_redis_checkpoint_key
    namespace, thread_id, checkpoint_ns, checkpoint_id = redis_key.split(
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ValueError: too many values to unpack (expected 4)

The redis key value generally looks like: checkpoint local-chat8 1f023664-e009-6c26-8002-c48ddc7d9a09, but after langgraph interruption inside the tool, it looks like: checkpoint:local-chat8:troubleshoot_subgraph:0b11a927-2f15-9d21-7645-8b014a688fb8:1f023664-f2cc-6b42-bfff-9112b3fd9406 (5 values to unpack)`

This is causing to above error. How to make interrupt inside tools inside subgraphs and handle them using langgraph redis saver implementation.

I am using RedisSaver implementation mentioned on: https://github.com/langchain-ai/langgraph/blob/main/docs/docs/how-tos/persistence_redis.ipynb