How to reconnect to stream after <class 'httpx.TransportError'> Exceeded maximum SSE reconnection attempts

LangSmith Product Help Deployment
,
1

When running a langgraph agent for a long time I eventually get:

<class ‘httpx.TransportError’> Exceeded maximum SSE reconnection attempts

This comes from a loop using this:
async for chunk in client.runs.stream

How do I reconnect to the run without dropping output? (join_stream docs say it doesnt buffer output)

Docs also mention:
A background run can execute for longer than 1 hour, but a client must reconnect to the server (e.g. join stream via POST /threads/{thread_id}/runs/{run_id}/stream) to retrieve output from the run if the run is taking longer than 1 hour.

join_stream uses GET but stream doesn’t take a run_id argument

Edit: I set BG_JOB_TIMEOUT_SECS to 10 hours, this disconnection happens long before that and I can see the run continuing in the langsmith trace

2

@sasharosca To clarify—what is “a long time”, typically? Are you typically seeing the Exceeded maximum SSE reconnection attemptserror before the 1-hour maximum duration mentioned in the docs you cited?

In any case, you should be able to resume a stream after it disconnects by calling join_stream() again with last_event_id(the stream must be resumable for this to work properly).