Notes

I was given 7 tasks which consisted of Hadoop shell commands, analyzing data using PIG, creating tables and analyzing data in HIVE, importing/exporting from/to MYSQL using SQOOP and configuring a FLUME agent and starting it. I was able to complete all the tasks within 1 hour and 30 minutes.

I was given 7 questions, of which I had/have (still awaiting results) to get 5 correct to get certified. I was given a question each from Sqoop and Flume, 3 Pig questions and 2 Hive questions. For instance, in the Flume question, they asked me to create a memory channel with capacity of 5000. While 5000 was highlighted, memory was not, and I had to read the question again to figure out what type of channel I needed to configure.

A few tips for Pig questions - One, ALWAYS name your GENERATEd columns. If there is a schema with 20 columns, they’ll tell you which ones are relevant, but when you do a FOREACH.. GENERATE.., make sure you give the selected columns names. I didn’t do it for one question (at first) and lost about 10 minutes wondering why the output was not showing up. I then added column names and the script ran through in a breeze. It doesn’t matter what you name them, but to be safe, just do.
Two, when they give you the relevant column numbers, and information, they do not account for Pig’s zero-indexing.
Three, if you come across a question where they ask you to populate a Hive table as a result of a Pig transformation, remember to use HCatalog/HCatStorer/HCatLoader, and more importantly, to invoke the grunt shell using the flag -useHCatalog. I was on the right track with the question and had even written the long org.apache.hive.hcatalog.pig.HCatStorer(), but without the -useHCatalog flag, it was useless. Since I was rushed for time and had two more questions to finish all 7 in only 15 minutes, I skipped it and finished the Sqoop and Flume questions.
Four, in the grunt shell, Backspace does not work. It’s extremely frustrating. It works everywhere else - in Hive, in the terminal, in the vi editor. Just not in grunt. When almost 50% of your questions are in Pig, it is doubly frustrating. Even though I’d written the script and was mostly copy-pasting, checks in between were needed, and it was annoying when I made a typo.
Five, use “pig -x tez” before you start testing your scripts. It’ll be much faster to test and use than without tez.


https://cookeddata.wordpress.com/2015/12/10/hortonworks-data-platform-certified-developer-hdpcd-certification/
http://www.gocit.vn/files/Oreilly.Programming.Pig-www.gocit.vn.pdf?bcsi_scan_3350bd09186420d9=w/G9ibzJx8BjEskUd3mAPk+Az1EFAAAAYKJVFg==&bcsi_scan_filename=Oreilly.Programming.Pig-www.gocit.vn.pdf

https://github.com/rashidaligee/HDPCD-Certification

During the exam, be prepared to click on a small “Can’t type?” red button as you’ll keep losing focus of the keyboard within the virtual machine. I needed to use this about once every minute or even more at times. Preparing for this mentally is better, as it then won’t be as much of an annoyance as it normally is. These are small tips that, if you’re prepared for, and know what to expect, will help save a minute here or there.



There are few issues in the practice exam.

You will not have access to put your scripts in the mentioned directory. It is just a permission error and is taken care of in the actual exam.
For executing SQOOP commands you need to grand permission to root for the tables in MYSQL. This is also taken care of in the actual exam. During the Practice exam to execute SQOOP commands do the following in the mentioned order:
SSH into the namenode -> ssh root@namenode(password is “hadoop”)
Login to MYSQL -> mysql –username root –p(this should prompt you for a password, which is “hadoop”)
Grand permission for all the tables -> GRANT ALL ON *.* TO ‘root’@’localhost’ identified by ‘hadoop’.