Monday, June 19, 2017

OOZIE Job configuration from HUE


Check OOZIE Status:

oozie admin -oozie http://quickstart.cloudera:11000/oozie -status

For getting Name node URL:
cd /etc/hadoop/conf

ls -ltr

view core-site.xml

Property name is,
fs.defaultFS

For JobTracker URL:

view yarn-site.xml

Property name is
yarn.resourcemanager.address

Eg: quickstart.cloudera:8032  OR localhost:8032

Config file path of job.properties to be LOCAL PATH

To create Work flow Job from HUE:

In HUE go to
Workflows -> Editors -> Workflows and create Work flows

Few points:

-> Parameters are the ones for passing job specific parametets like Executor memory etc.

-> Files - are to be used to give input and output files

-> While creation/Edit of workflow a folder icon beside settings button will show the path of workflow.xml and job.properties in HDFS.

-> But job.properties should be in local folder - ?
As this is a single node setup ${namenode} and ${jobtracker} were taken by default.

Note:
Usually in production the jobs will be run via command line where we pass job.prpoerties from local file system using --config flag.

Job Progress and logs can be seen in HUE under the Job.
Also, Job progress can be seen in Job Tracker.

References:

Running Hive script in OOZIE:

Steps:
Follow work flow steps in OOZIE Work flow job creation.
Make sure to mention
oozie.use.system.libpath=True in job.properties.

Running Spark Job in OOZIE 

1. Running a spark Job that executes Hive Queries.

If came across error "The root scratch dir: /tmp/hive on HDFS should be writable. Current permissions are: rwx------"




Watch OOZIE Spark Job submission video under
https://www.youtube.com/watch?v=w5Be9ubK_Po

Watch OOZIE Hive Job submission video under
https://www.youtube.com/watch?v=i1QW7NoAiwM
Youtube, OOZIE Documentation in OOZIE Page

1 comment: