Hi, Have you tried submitting applications to other queue? Could you tell us does it worked on other queues? Queue.default has reached 99.7% and this may cause issues. Thanks AK
... View more
@PJ You can also look at the .out files which are present in the same yarn log dir location. Alternatively, you can launch the process in the foreground using "yarn nodemanager" CLI to see why the process is shutting down immediately.
... View more
Hi @Sandeep SIngh, Each mapper and reducer is a java process and we need some reserve memory to store the java code. Hence the Map/Reducer memory should be greater than the JVM heap size.
... View more
Hi @Predrag Minovic and @ssathish, I tried with WebHCat, it works. I already tried with the Distributed Shell application, but I didn't success. Maybe I did wrong. But I think I will try with Oozie soon. Thanks for you help
... View more
@ssathish
hdfs user is not usually allowed to access encryption keys. This ensures that even Hadoop admin cannot access the encrypted data. Check in your Ranger KMS who is authorized to access keys and use that user. You should not be using HDFS user to access encryption keys. https://community.hortonworks.com/content/supportkb/49505/how-to-correctly-setup-the-hdfs-encryption-using-r.html
... View more
@Hoang Le The max application master resource is calculated using the value "yarn.scheduler.capacity.maximum-am-resource-percent" which is set in capacity-scheduler.xml. The default value is 0.1 which means 10% of the cluster resources. More info can be found in http://docs.hortonworks.com/HDPDocuments/HDP2/HDP-2.4.0/bk_yarn_resource_mgt/content/setting_application_limits.html
... View more
