Moving to mapreduce-dev@, bcc general@. Yes, as described in the bug, the CS has high-ram jobs which is a better model for shared multi-tenant clusters. The hadoop-0.20.203 release from Apache has the most current and tested version of the CapacityScheduler.
Arun Sent from my iPhone On Jul 22, 2011, at 9:36 AM, Liang Chenmin <[email protected]> wrote: > Hi all, > I am using hadoop 0.20.2 cdh3 version. The old method to set max > concurrent mapper/reducer in code no longer works. I saw a patch about this, > but the current status is "won't fixed". Is there any update about this? I > am using Fair Scheduler, should I use Capacity Scheduler instead? > https://issues.apache.org/jira/browse/HADOOP-5170 > > Thanks, > chenmin liang
