seniorHadoop
What is Hadoop MapReduce task failure recovery mechanism?
Updated May 16, 2026
Short answer
Failed tasks are retried automatically by YARN up to a configured limit.
Deep explanation
If a task fails due to node crash or exception, YARN reschedules it on another node. Intermediate outputs are recomputed. Job success depends on successful completion of all tasks within retry limits.
Real-world example
Recovering ETL jobs when compute nodes fail mid-processing.
Common mistakes
- Assuming partial task output is reused.
Follow-up questions
- What triggers task retry?
- Is state preserved?