Self Hosted LangGraph server Errors (Google Cloud)

[BinariusConsulting]

I have a self hosted version of a very simple graph deployed on Google Cloud run. The docker was cerated using the langgraph cli tools and I have my own postgres and redis instances wroking well and connected. When i call the endpoint of this self hosted version more than 50% of the times i get the below errors. Inadditona there is a warning [WARNING]: LangSmith failed to fetch info on supported operations. Falling back to batch operations and default limits. How can i resolve these errors?

On the other hand, if I use the langgraphjs dev command to run the graph locally in LangGraph Studio, It works without any errors whatsoever.

2025-04-18 23:44:51.927 PDT
^^^^^^^^^
2025-04-18 23:44:51.927 PDT
File "/api/langgraph_api/stream.py", line 269, in consume
2025-04-18 23:44:51.927 PDT
File "/api/langgraph_api/stream.py", line 258, in consume
2025-04-18 23:44:51.927 PDT
File "/api/langgraph_api/stream.py", line 147, in astream_state
2025-04-18 23:44:51.927 PDT
File "/api/langgraph_api/asyncio.py", line 82, in wait_if_not_done
2025-04-18 23:44:51.927 PDT
File "/api/langgraph_api/js/remote.py", line 142, in astream_events
2025-04-18 23:44:51.927 PDT
File "/api/langgraph_api/js/remote.py", line 92, in _client_stream
2025-04-18 23:44:51.927 PDT
File "/api/langgraph_api/js/sse.py", line 134, in aiter_lines_raw
2025-04-18 23:44:51.927 PDT
File "/usr/local/lib/python3.12/site-packages/httpx/_models.py", line 997, in aiter_bytes
2025-04-18 23:44:51.927 PDT
async for raw_bytes in self.aiter_raw():
2025-04-18 23:44:51.927 PDT
File "/usr/local/lib/python3.12/site-packages/httpx/_models.py", line 1055, in aiter_raw
2025-04-18 23:44:51.927 PDT
async for raw_stream_bytes in self.stream:
2025-04-18 23:44:51.927 PDT
File "/usr/local/lib/python3.12/site-packages/httpx/_client.py", line 176, in __aiter__
2025-04-18 23:44:51.927 PDT
async for chunk in self._stream:
2025-04-18 23:44:51.927 PDT
File "/usr/local/lib/python3.12/site-packages/httpx/_transports/default.py", line 270, in __aiter__
2025-04-18 23:44:51.927 PDT
with map_httpcore_exceptions():
2025-04-18 23:44:51.927 PDT
^^^^^^^^^^^^^^^^^^^^^^^^^
2025-04-18 23:44:51.927 PDT
File "/usr/local/lib/python3.12/contextlib.py", line 158, in __exit__
2025-04-18 23:44:51.927 PDT
self.gen.throw(value)
2025-04-18 23:44:51.927 PDT
File "/usr/local/lib/python3.12/site-packages/httpx/_transports/default.py", line 118, in map_httpcore_exceptions
2025-04-18 23:44:51.927 PDT
raise mapped_exc(message) from exc
2025-04-18 23:44:51.927 PDT
httpx.ReadTimeout

[benjamincburns]

cc @dqbd

[benjamincburns]

@BinariusConsulting just to clarify, does the same image that fails in GCP run fine when tested locally?

[BinariusConsulting]

@benjamincburns Hi! I did not thoroughly test the image locally, but I think it worked locally. However, I have not seen the errors from the past three days. My hypothesis is that the errors were being caused by cold starts on Google Cloud Run, so I currently have at least one instance running for my LangGraph server, and that seems to have done the trick. Additionally, I rebuilt the image using the latest LangGraph CLI, so that might also have played a part. I'll let you know if this resurfaces again.

[benjamincburns]

Transferred to LangGraphJS API project as it's related to CLI & OSS platform tooling