The manual installation of UCX Assessment in Databricks works with the default <ALL> values, but automation scripts that set WORKSPACE_GROUPS="<ALL>" DATABASES="<ALL>" often encounter a SCHEMA_NOT_FOUND error related to 'ALL' not being recognized as a valid schema name.
Why Manual Works, Automation Fails
-
During manual installs, the CLI/UI prompts for inputs and internally interprets <ALL> as a wildcard, triggering logic that migrates all schemas/databases. The CLI does not literally use 'ALL' as the schema name but understands it as "all available options".
-
When automating, explicitly passing <ALL> as an environment variable likely leads the script to treat it as a literal name, causing attempts to crawl a schema named 'ALL', which does not exist—hence, the failure.
Needed Parameters for Automation
-
Instead of setting DATABASES="<ALL>", leave the variable empty or unset; most automation tools will default to including all schemas if not explicitly specified.
-
Check if the automation command supports wildcards or a specific keyword (like an empty string or "*"), as opposed to "<ALL>". The official UCX docs and troubleshooting guide recommend relying on default behavior rather than overriding it with a literal parameter, especially for automation workflows.
-
You may need to consult the UCX README notebook generated on install, which details accepted parameters for jobs and automation.
Resolving Schema Lookup Issues
-
Try omitting the explicit DATABASES and WORKSPACE_GROUPS parameters, letting the script use defaults. If parameters must be set, confirm with UCX documentation what value (empty, wildcard, etc.) triggers the “all” behavior.
-
Verify your automation script is running in the same context/catalog as your manual install. Workspace default catalog settings, such as not using hive_metastore, can also impact schema discovery.
-
Review the UCX troubleshooting guide and error conditions for advanced configuration or known issues regarding schema crawl operations. Sometimes, a reinstall of UCX may be needed if assessment results are already stored and an incremental run isn't supported.
Recommendations
-
For automation, remove the explicit <ALL> values or carefully check how UCX expects to receive "all" options in automation. This will resolve the SCHEMA_NOT_FOUND error by avoiding an invalid literal schema name.
-
Always check the Databricks Catalog Explorer and diagnosis dashboards post-install to confirm schema detection is successful.
