Why are my Tasks Succeeded above Tasks Total in Spark UI?

Question

I had thought that succeeded would equal total plus or minus failed.

Where do these numbers come from?

Do these numbers indicate an issue, or do they matter at all?

Answer 1

The first number (3660) is the number of succeeded tasks. The second number (1830) is the number of tasks in the stage. Your stage ran twice, so the number of succeeded tasks is twice the number of tasks in the stage.

A stage can partially or in whole re-executed when multiple later stages depend on them and they are not cached or not have been partially evicted from the cache.

Another way you can have a higher number of succeeded tasks than the total is using speculative execution.