Details
-
Sub-task
-
Status: Resolved
-
Minor
-
Resolution: Duplicate
-
3.1.0
-
None
-
None
Description
When aborting a job, the staging committers list staged files in the cluster FS to abort...all exceptions are caught & downgraded to log events.
We shouldn't even log FNFEs except at debug level, as all it means is "the job is aborting before things got that far. Printing the full stack simply creates confusion about what the problem is
Attachments
Issue Links
- Is contained by
-
HADOOP-15107 Stabilize/tune S3A committers; review correctness & docs
- Resolved