[TEZ-4147] Reduce NN calls in RecoveryService::handleRecoveryEvent

• Type: Improvement
• Status: Resolved
• Priority: Trivial
• Resolution: Fixed
• Fix Version/s: 0.10.0, 0.9.3
• Assignee: László Bodor
• Reporter: Rajesh Balamohan
• Created: 15/Apr/20 23:20
• Updated: 17/Apr/20 19:51
• Resolved: 17/Apr/20 19:51

Description

https://github.com/apache/tez/blob/master/tez-dag/src/main/java/org/apache/tez/dag/history/recovery/RecoveryService.java#L459

https://github.com/apache/tez/blob/master/tez-dag/src/main/java/org/apache/tez/dag/history/recovery/RecoveryService.java#L467

create(overwrite=false) already checks for file exists and throws relevant exception. It should be possible to reduce recoveryDirFS.exists(dagFilePath) check to help in concurrent job runs (e.g 50-100 AM trying to run very short jobs).