Hi,

I found in hadoop 0.19 document, it says "These excess allocated resources can 
be reclaimed and made available to another queue in order to meet its capacity 
guarantee.", and there is a config called 
"mapred.capacity-scheduler.queue.<queue-name>.reclaim-time-limit".

But when I used yarn's capacity scheduler, I found that over allocated capacity 
cannot be reclaimed when needed. So I want to know how capacity is guaranteed 
when elasticity is enabled. Is this feature cut or still under developing?

Thanks

Best Regards,
Peng





Reply via email to