Databricks bundle job: UC 'key-manager' UNAVAILABLE on saveAsTable
On a new Databricks Free Edition workspace, a bundle-deployed serverless notebook job failed at the first DataFrame.saveAsTable with AnalysisException ... ErrorClass=BAD_REQUEST UNAVAILABLE: RCS returned no response for service=key-manager, extractor=pctx.metastoreid (UnityCatalogServiceException in createStagingTable). CREATE SCHEMA succeeded just before, and a rerun reproduced it, so it is not a one-off blip. Searches found no documented fix. Diagnostic split: run a one-line CREATE TABLE ... AS SELECT 1 in the SQL editor. If that also fails, it is a Databricks-side metastore/key service problem (wait or open support with the trace ID); if it works, look at the job's run identity. Still unresolved at time of writing.
Update: the same 'RCS returned no response for service=key-manager, extractor=pctx.metastoreid' error also hits CREATE TABLE AS SELECT via a SQL warehouse (Statement Execution API), in every catalog of the metastore (workspace, main), AND SELECT on previously working managed tables. SHOW SCHEMAS/TABLES and reads of the samples catalog still succeed. So it is a metastore-wide key-service outage, not job identity, bundle config or code. Quick triage via CLI: databricks api post /api/2.0/sql/statements with a warehouse_id and 'CREATE OR REPLACE TABLE <cat>.default.probe AS SELECT 1'.