-
Notifications
You must be signed in to change notification settings - Fork 1.8k
Description
What's the issue?
I am using the docker-compose approach with
- 1 service for the deamon
- 1 service for dagit
- 1 service for my code location, running on its own docker container
And the workspace.yml
loads from grpc_server
Without any code change on my side, I updated all my packages (which included an update of dasgter minor version as well as moving to dagster-sahred
1.0) and while docker-compose started correctly
- the service for the deamon failed
- running any job failed with errors like
dagster._core.errors.DagsterCodeLocationNotFoundError: Location example_user_code does not exist in workspace
File "/opt/dagster/dagster_home/.venv/lib/python3.12/site-packages/dagster/_daemon/run_coordinator/queued_run_coordinator_daemon.py", line 411, in _dequeue_run
instance.run_launcher.launch_run(LaunchRunContext(dagster_run=run, workspace=workspace))
File "/opt/dagster/dagster_home/.venv/lib/python3.12/site-packages/dagster/_core/launcher/default_run_launcher.py", line 112, in launch_run
code_location = context.workspace.get_code_location(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/opt/dagster/dagster_home/.venv/lib/python3.12/site-packages/dagster/_core/workspace/context.py", line 204, in get_code_location
raise DagsterCodeLocationNotFoundError(
After updating my dependency on dagster-shared
to <1.0.0
and regenerating my uv lock file, the next deployment worked as expected.
Looking at the docs, I just realize that the docker-compose example shared from last time I checked.
My command is
CMD ["dagster", "api", "grpc", ...
When the doc now show:
CMD ["dagster", "code-server", "start", ...
What did you expect to happen?
I didn't expect an update to the packages to break my deployment
How to reproduce?
No response
Dagster version
1.11.9
Deployment type
Docker Compose + GRPC
Deployment details
No response
Additional information
No response
Message from the maintainers
Impacted by this issue? Give it a 👍! We factor engagement into prioritization.