mahout-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Elaine Gan <elaine-...@gmo.jp>
Subject Re: Number of MR running mahout kmeans
Date Mon, 03 Dec 2012 09:42:14 GMT
Dear Owen,

Thank you.

> This isn't specific to Mahout. In Hadoop, your capacity has no bearing
> on how many jobs Hadoop chooses to run. The ".maximum" properties are
> just expressing how many tasks to run concurrently on one worker.
> 
> Hadoop will always run 1 reducer unless you tell it otherwise with
> mapred.reduce.tasks. I think most of the jobs don't try to set this
> for you and you'll have to do so yourself.
> 

No wonder i see only 1 reducer no matter how many times i tried to
execute it.
Thank you!

> The number of maps depends on the size and number of input files. I
> wouldn't override it, but, if you must you can do so by decreasing the
> maximum split size that one mapper takes.

Noted!

Thanks

Mime
View raw message