Hello, how do you use Spark Streaming with Delta Lake, the pattern “forEachBatch”, when the table does not exists yet?

Depends on how you're writing to the table, some of the time it's created for you. But if you're using

MERGE

, for example, the table is not created automatically so you'd have to do it yourself. For things like this, I typically a DeltaTable.createIfNotExists() which will no-op when the table already exists. If you go this way, I suggest using addColumns; with it you can easily take the schema of an existing DataFrame and inject it into the builder.

using delta lake connector will right idea, u can pull data directly in delta format