What happened?
We have seen a behavior for Python SDK using interactive runner starting 2.42.0 that when a flag which is same as another pipeline option is passed to dataflow runner with single dash, it escapes the valid argument check before job submission and creates problem at runtime. Moreover, the error is logged as info message and the job fails after an hour because of worker stuck error.
For example, passing --f=abc which is a valid shorthand for --flexrs_goal does the argument validation before job submission. But when -f=abc is used, it escapes this check and logs an info message

on dataflow.
Issue Priority
Priority: 2 (default / most bugs should be filed as P2)
Issue Components