You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Shorten the comments and changelog added by this change
Several comments restated the code or carried the reasoning behind an
earlier revision, which belongs in the commit messages. Each one now states
the API behaviour or decision the code depends on and stops there.
The changelog entry explained the mechanism at length; it now names the fix
and keeps the regression for unsupported endpoint types.
direct: Fixed `telemetry_config` on model serving endpoints never converging, and edits to it being planned and then silently dropped. The backend creates a telemetry profile from `table_names` and returns only the resulting `telemetry_profile_id`, so every `bundle plan` after a deploy reported an update, and the deploy that followed made no telemetry call at all. Wiring the telemetry API also means a `telemetry_config` on an endpoint that serves nothing or only external models now fails on a deploy that applies the field, with the error that API returns (create still drops the field) ([#6106](https://github.com/databricks/cli/pull/6106)).
1
+
direct: Fixed model serving `telemetry_config` drift and applied planned telemetry updates. Unsupported endpoint types now fail when telemetry is applied; create may still succeed because it drops the field ([#6106](https://github.com/databricks/cli/pull/6106)).
0 commit comments