You are viewing a plain text version of this content. The canonical link for it is here.

Posted to common-user@hadoop.apache.org by Saptarshi Guha <sa...@gmail.com> on 2010/02/12 18:07:48 UTC

A unique number for every task (map/reduce)

Hello,
The name of a map task attempt is

attempt_201002092101_0063_m_N_X (N=1,.... # of map tasks)

and that of a reduce is

attempt_201002092101_0063_r_M_X where X is maybe 0,1,2 ..
(M =1 .... # o reduce tasks)

Thus during the map stage, the unique number is given by N or M.

Is there configuration variable so that I can get the name of this task file?
Then i can arrive at the components: date. job-number, task-#, task atempt #

Thank you
Saptarshi

Re: A unique number for every task (map/reduce)

Posted by Amareshwari Sri Ramadasu <am...@yahoo-inc.com>.

You can get task attempt id, tipid, job id though the configuration properties  mapred.task.id, mapred.tip.id and mapred.job.id respectively.

For e.g. the attempt attempt_201002092101_0063_m_000000_0 : attemptid - attempt_201002092101_0063_m_000000_0; tipid - task_201002092101_0063_m_000000; jobid - job_201002092101_0063
Hope this helps!

Thanks
Amareshwari

On 2/12/10 10:37 PM, "Saptarshi Guha" <sa...@gmail.com> wrote:

Hello,
The name of a map task attempt is

attempt_201002092101_0063_m_N_X (N=1,.... # of map tasks)

and that of a reduce is

attempt_201002092101_0063_r_M_X where X is maybe 0,1,2 ..
(M =1 .... # o reduce tasks)

Thus during the map stage, the unique number is given by N or M.

Is there configuration variable so that I can get the name of this task file?
Then i can arrive at the components: date. job-number, task-#, task atempt #

Thank you
Saptarshi