Hi Vandana

From the configurations, it looks like none of the NodeManagers are registered 
with RM because of configuration “yarn.resourcemanager.resource- 
tracker.address” issue.  May be you can confirm any NM’s are registered with RM.

In the below, there is space after “resource-“ but “resource-tracker” is single 
without any space. Check after removing space.
<name>yarn.resourcemanager.resource- tracker.address</name>

Similarly I see same issue in “yarn.nodemanager.aux- 
services.mapreduce.shuffle.class” where space after “aux-”!!!

Hope it helps you to resolve issue

Thanks & Regards
Rohith Sharma K S

From: Vandana kumari [mailto:[email protected]]
Sent: 15 April 2015 15:33
To: [email protected]
Subject: Mapreduce job got stuck

i had setup a 3 node hadoop cluster on centos 6.5 but nodemanager is not 
running on master and is running on slave nodes. Alse when i submit a job then 
job get stuck. the same job runs well on sinle node setup. I am unable to 
figure out the problem. Attaching all the configuration files.
Any help will be highly appreciated.

--
Thanks and regards
  Vandana kumari

Reply via email to