Support Questions
Find answers, ask questions, and share your expertise
Announcements
Alert: Welcome to the Unified Cloudera Community. Former HCC members be sure to read and learn how to activate your account here.

Hortonworks Spark ODBC Driver Connection Keeps Failing

Highlighted

Hortonworks Spark ODBC Driver Connection Keeps Failing

New Contributor

I am running HDP 2.5 on my 4 node cluster and have both Spark1 and Spark2 Installed. I want to connect to the data through Tableau and I installed both 32 bit and 64 bit hortonworks Spark ODBC Drivers from the Hortonworks website.

My problems are:

1. Both the Spark Thrift Servers keep failing suddenly out of the blue. I am not sure if it is some configuration issue (like not having enough heap size so even though it starts up when I start it, eventually it fails).

2. Can I have both the Sparks running simulatneously? Or will that cause any memory overload on the cluster?

3. In the ODBC Driver DSN setup, when I click on "Test" option, sometimes it fails even when the thrift server is up and running. The error is: "[Hortonworks][Hardy] (34) Error from server: connect() failed: errno = 10061."

I am using the 32 bit Driver for Spark2 and 64 bit for Spark1 (I donot know why but that is how it worked- exactly as someone suggested the same in my previous question on this website)

Can someone please elaborate what is happening? I am not able to find any help from the web. Thanks in advance!

1 REPLY 1

Re: Hortonworks Spark ODBC Driver Connection Keeps Failing

Guru

@Sree Kupp,

1. Both the Spark Thrift Servers keep failing suddenly out of the blue. I am not sure if it is some configuration issue (like not having enough heap size so even though it starts up when I start it, eventually it fails).

A cluster can have spark1 and spark2 thrift server running together. Is spark1 and spark2 thrift server deployed on same host ? Can you please check what is the error message for spark thrift server failure?

2. Can I have both the Sparks running simulatneously? Or will that cause any memory overload on the cluster?

Yes , you can have both the spark running simultaneously. Regarding memory overload, If you are using yarn-client or yarn-cluster mode to run the spark applications, It won't memory overload the client machine.

3. In the ODBC Driver DSN setup, when I click on "Test" option, sometimes it fails even when the thrift server is up and running. The error is: "[Hortonworks][Hardy] (34) Error from server: connect() failed: errno = 10061."

I found few good links to handle this issue. Seems like many people hit similar issue. I hope this helps.

http://kb.tableau.com/articles/issue/error-connect-failed-hadoop-hive

https://community.hortonworks.com/questions/33046/hortonworks-hive-odbc-driver-dsn-setup.html

https://community.hortonworks.com/questions/10192/facing-issue-with-odbc-connection.html

Don't have an account?
Coming from Hortonworks? Activate your account here