Hi,
I was working on a scenario where in I am generating a file in close()
function of my Map implementation.
Since Map execution are worked concurrently, this file is overwritten.
I was wondering how to name this file uniquely per map execution basic and
then read in configure() function of re
Hi Aaron,
I am on Hadoop 0.20.0 on Ubuntu, pseudo distributed mode. If I remove the
sleep time from my start-all.sh script, my jobtracker comes up momentarily
and then dies.
Here is a capture of my commands:
sgo...@desktop:~/software/hadoop-0.20.0$ bin/hadoop namenode -format
10/02/13 21:54:19 I
Hi Sonal,
Why are you using Hadoop 0.20.0? It's fairly old and there are lots of
fixes in 0.20.1, and more in 0.20.2 which should be released any
minute now.
In particular, you're missing this change:
https://issues.apache.org/jira/browse/HADOOP-5921
which makes the JobTracker stubbornly wait fo
Yes, thanks Todd. I am looking to upgrade to 0.20.2.
Thanks and Regards,
Sonal
On Sat, Feb 13, 2010 at 11:07 PM, Todd Lipcon wrote:
> Hi Sonal,
>
> Why are you using Hadoop 0.20.0? It's fairly old and there are lots of
> fixes in 0.20.1, and more in 0.20.2 which should be released any
> minute
The number of map tasks is normally different from the number of reduce tasks.
Coding as you planned would limit the flexibility of hadoop.
On Saturday, February 13, 2010, ANKITBHATNAGAR wrote:
>
> Hi,
>
> I was working on a scenario where in I am generating a file in close()
> function of my Map