mirror of
https://github.com/langchain-ai/langgraph.git
synced 2026-10-14 12:15:25 +02:00
gethin-langchainandGitHub
ce14d33687
## Summary Originally this auto-selected a workspace's sole listener for self-hosted/hybrid control planes the same way #9056 already does for Cloud. Per review, that's reverted: in self-hosted environments, leaving placement blank is the deliberate signal to deploy to the bundled operator, not an unresolved default. Auto-selecting a sole listener there would silently take that option away from anyone who has both a bundled operator and one listener configured, with no way to get it back. The actual need behind this was simpler: a way to find a listener's id without hardcoding it, so it can be passed to `--listener-id` explicitly. `HostBackendClient.list_listeners()` already existed and is used internally for placement resolution, but had no CLI surface of its own. This adds `langgraph deploy listeners list`, mirroring the existing `deploy list` / `deploy revisions list` pattern, to look up a workspace's listeners and their namespaces. No change to default placement behavior. ## Test plan - [x] `make format && make lint` clean in `libs/cli` - [x] `format_listeners_table` unit test in `test_util.py` - [x] CLI-level tests for `deploy listeners list` (with and without results) in `test_cli.py` - [x] Full `libs/cli` suite: 508 passed, no regressions