[Bug]: TextIO.read() runs in only one executor when using SparkStructuredStreamingRunner but runs in distributed manner with SparkRunner

[DebajitKumarPhukan]

What happened?

Issue
If you try to read any file line by line using TextIO.read() in Spark cluster, SparkRunner runs the tasks in multiple executors. But when you just switch to SparkStructuredStreamingRunner, it runs only in one executor. The distribution behavior should should remain the same even if the underlying API is changed from RDD to Dataset.

Note: I am just reading one text file.

Issue Priority

Priority: 2 (default / most bugs should be filed as P2)

Issue Components

• Component: Python SDK
• Component: Java SDK
• Component: Go SDK
• Component: Typescript SDK
• Component: IO connector
• Component: Beam YAML
• Component: Beam examples
• Component: Beam playground
• Component: Beam katas
• Component: Website
• Component: Infrastructure
• Component: Spark Runner
• Component: Flink Runner
• Component: Samza Runner
• Component: Twister2 Runner
• Component: Hazelcast Jet Runner
• Component: Google Cloud Dataflow Runner