hadoop-mapreduce-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Himawan Mahardianto <mahardia...@ugm.ac.id>
Subject Re: how namenode decided which datanode are ready for a client
Date Wed, 20 May 2015 01:21:17 GMT
but on what category datanode are choosen by namenode, if some client want
to write some data(block) on it via HDFS
how namenode distinguish on one datanode and the other if only I just have
one rack, so rack awareness feature are unused

On Tue, May 19, 2015 at 3:56 PM, rajesh damera <d.rajesh15@gmail.com> wrote:

> Namenode usually checks for the jobs running on the datanodes, of which
> tasktrackers keep a track of and decides if it still busy running a task.
> On May 19, 2015 14:23, "rajesh damera" <d.rajesh15@gmail.com> wrote:
>
>> Datanode availability is monitored and decided by Namenode. You usually
>> get permissions, based on the type of user you are, to use a certain number
>> of nodes to run your jobs. These limitations are set by administrators at
>> user level
>>
>> Warm regards
>> Rajesh Damera
>> On May 19, 2015 14:08, "Himawan Mahardianto" <mahardianto@ugm.ac.id>
>> wrote:
>>
>>> Hi guys, I know that every datanode where sending heartbeat to prove
>>> that their alive but what method are used by namenode to decide which
>>> datanode are available to client if all datanode where available?
>>>
>>> Best Regards
>>>
>>>
>>

Mime
View raw message