hadoop - An unhandled exception in Mapper or Reducer Task would make the task failed? -

I'm thinking that an unused exception function may be unsuccessful in the Operating Operations (or Reader's work) Will the input pair be ignored? I think this is ex-but not so sure

exception is not recurring (i.e. retry Will be successful), then this task will fail, the task will be retried (or speculative execution running in parallel can move forward). If the exception always occurs (like poor input data) then the effort will fail again and in the end the whole job will fail.