I was wondering if there is a way to let fair scheduler ignore the user and
submit a job to a specific pool.
I would like to have 3/4 pools:
1. Very short (~1 min) routine jobs.
2. Normal processing time (<1 hr) routine jobs.
3. Long (days) experimental jobs.
4. ? ad hoc immediate jobs ?
Users matter less in my use case.
(Running 20 nodes cluster with Hadoop 1.0.4)