Hi team, I am trying to deploy our langgraph agents to our EU LangSmith Cloud using langgraph deploy CLI instead of using the UI (which we already used to deploy successfully a few deployments).
I set up our Github Actions like this:
Run uvx --from langgraph-cli langgraph deploy --name drova-agents-dev --deployment-type dev --verbose
uvx --from langgraph-cli langgraph deploy --name drova-agents-dev --deployment-type dev --verbose
shell: /usr/bin/bash -e {0}
env:
LANGSMITH_API_KEY: ***
LANGSMITH_TENANT_ID: ***
LANGGRAPH_HOST_URL: https://eu.api.host.langchain.com
The CLI was able to deploy the container to the cloud. However, in the server log there is an error:
[ERROR] Queue entrypoint task failed
Traceback (most recent call last):
File "/api/langgraph_api/queue_entrypoint.py", line 293, in main
File "/api/langgraph_api/queue_entrypoint.py", line 223, in entrypoint
File "/usr/local/lib/python3.12/contextlib.py", line 210, in __aenter__
return await anext(self.gen)
^^^^^^^^^^^^^^^^^^^^^
File "/api/langgraph_api/timing/timer.py", line 227, in combined_lifespan
File "/usr/local/lib/python3.12/contextlib.py", line 659, in enter_async_context
result = await _enter(cm)
^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/contextlib.py", line 210, in __aenter__
return await anext(self.gen)
^^^^^^^^^^^^^^^^^^^^^
File "/storage/langgraph_runtime_postgres/lifespan.py", line 74, in lifespan
File "/api/langgraph_license/validation.py", line 467, in get_license_status
ValueError: Custom authentication is currently available in the cloud version of LangSmith Deployment or with an self-hosting enterprise license. Please visit https://langchain-ai.github.io/langgraph/concepts/deployment_options/ to learn more about deployment options, or contact sales@langchain.com for information on upgrading from the self-hosted Lite plan to an enterprise license.
[INFO] Shutting down queue...
[INFO] Entrypoint task finished
[WARNING] No enterprise license key found, running in lite mode with LangSmith API key. For production use, set LANGGRAPH_CLOUD_LICENSE_KEY in environment.
We are a paid customer and already have a few successfully-deployed deployment so I’m not sure why the errors mention the lack of license key.
Hello @arthurtran-drova welcome to Langchain Community.
Your deployment fails because the deployed container validates its license by making an HTTP request to the LangSmith auth endpoint, but it defaults to the US endpoint (api.smith.langchain.com), not your EU one. Since your API key belongs to an EU workspace, the US endpoint doesn’t recognize it as having LangGraph Cloud entitlement. The runtime falls back to “lite mode,” and lite mode explicitly blocks custom authentication.
The Fix
Set LANGSMITH_ENDPOINT in your GitHub Actions workflow so the deployed container points its license validation at the correct EU endpoint:
env:
LANGSMITH_API_KEY: ${{ secrets.LANGSMITH_API_KEY }}
LANGSMITH_TENANT_ID: ${{ secrets.LANGSMITH_TENANT_ID }}
LANGSMITH_ENDPOINT: https://eu.api.smith.langchain.com
LANGGRAPH_HOST_URL: https://eu.api.host.langchain.com
Or set it in the env block of your langgraph.json so it’s baked into the deployment itself:
{
"env": {
"LANGSMITH_ENDPOINT": "https://eu.api.smith.langchain.com"
}
}
Doc reference: docs.langchain.com/langsmith/env-var#langsmith_endpoint
Thanks for the help. However, I could not make it work by adding LANGSMITH_ENDPOINT. In the CICD deployment log there is this log line from langgraph cli:
Skipping reserved env var: LANGSMITH_ENDPOINT
I took a look at the langgraph_cli/deploy.py source code and found that the env var LANGSMITH_ENDPOINT is part of RESERVED_ENV_VARS and were skipped.
If that also didn’t do the trick, then it might be a server-side bug.
However, can you also please double-check your LANGSMITH_TENANT_ID and make sure it is pointing to the right workspace ID (I have previously made this mistake)
Go to https://eu.smith.langchain.com
Click Settings (bottom left)
Click Workspaces
Copy the UUID — that's your LANGSMITH_TENANT_ID
Once you have double checked and it still does not work, then I would recommend opening a support ticket with LangSmith https://support.langchain.com/
You can through the following guide to understand how to open the ticket LangChain Support Portal
Finally, I am tagging an official maintainer so they can also look over the issue:
@niilooy
ty @keenborder786, will keep an eye out on this one.
Hi @arthurtran-drova, the issue you’re facing is as a result of current limitation with local docker builds with the cli. Try this instead to trigger a remote build.
uvx --from langgraph-cli langgraph deploy --name drova-agents-dev --deployment-type dev –remote --verbose
Thank you. This was the fix that make it work!
