Will my spark accumulator see all the records written by the executors?

Question

In the following code, is the BLOCK 2 loop guaranteed to be executed only after all the executor tasks spawned by BLOCK 1 have finished, or is it possible for it to run while some of the executor tasks are still running?

If it is possible for the 2 blocks to run concurrently, what is the best way to prevent this? I need to process the contents of the accumulator, but only once all the executors have finished.

When running with a master url of local[4] as shown, it looks like BLOCK 2 waits for BLOCK 1 to finish, however I am seeing errors when running with a url of yarn which suggest that BLOCK 2 is running concurrently with the executor tasks in BLOCK 1

object Main {
    def main(args: Array[String]) {
        val sparkContext = SparkSession.builder.master("local[4]").getOrCreate().sparkContext
        val myRecordAccumulator = sparkContext.collectionAccumulator[MyRecord]

        // BLOCK 1
        sparkContext.binaryFiles("/my/files/here").foreach(_ => {
            for(i <- 1 to 100000) {
                val record = buildRecord()
                myRecordAccumulator.add(record)
            }
        })

        // BLOCK 2
        myRecordAccumulator.value.asScala.foreach((record: MyRecord) => {
            processIt(record)
        })
    }
}