Problems with oracle BDA & Cloudera

[#1]

I am a beginner with oracle BDA and Cloudera ,after installing BDA to my cluster all other services like zookeper , hive  went to halt , if I disable the BDA all of the services started working. Please help

[#2]

@BSambit

Have you tried to configure Oracle BDA on an existing cluster ? (or) created new cluster using mammoth (bdacli) commands ?

If your answer is yes for my first question, what are the steps you have followed to enable/disable Oracle BDA on an existing cluster?

[#3]

Hi ,

That's a long process .


1. Download V860496-01.zip (which is for Cloudera) from oracle

2. There is a PDF available from oracle named E76071-02.pdf which having the installation steps

3. Configure the /usr/java/default to point jdk of cloudera

4. Configure a parallel python as this BDA don't accept 2.6 version so you have to configure 2.7 onwards . basically a yum install

5. Download others mentioned in the pdf .

6. Then config the bda-config.json file to point to oracle Db.

7. Start the setup with the json file

8. After that it will available as a parcel .. then in the cloudera manager go to host >parcel to see the parcel available . activate it

9. It will come below to the zookeeper of the main page, from there you can start it like other services .

Regards
Sambit

[#4]

@BSambit

From your details, my understanding is that you have Cloudera manager configured already and trying to setup Oracle BDA on top of that? is my understanding true? if so, if CM is already working then what is the point of setting up BDA on top of it?

[#5]

Hi,

Yes its an existing cloudera and its working , but here have a problem we need to move some 20 tb data from oracle RDBMS system to Hadoop and we dont want to use sqoop or any kind of such technology , so we want to read directly from oracle data pump file and there is only one serde available and which comes as part of BDA . so we want the BDA.

[#6]

@BSambit

If your requirement is to access 20 tb data from Oracle to hive/impala. And for only this purpose if you are trying to configure Oracle BDA then I don't think this is a right choice. Because BDA is not meant for that purpose and it will control your entire cluster. So just for one db/table you cannot switch your cluster to BDA, if so, you may get into different issues in future  (ex: you may get error when upgrade your environment). So either configure your cluster via BDA (or) without BDA but not both (this is my understanding).

I understand your requirement that you are looking for a serde which is available with BDA. Did you check the serde availability anywhere in github? 

Ex: use the below link only for understanding... it shows how to download, install & enable a library to read and write in JSON format from Hive. Similar to that you have to download your library from github or somewhere else to meet your requirement instead of configure BDA

https://github.com/rcongiu/Hive-JSON-Serde