Need help on Spark Upgrade

[#1]

The spark versions are different in PROD and Staging environment and I have been asked to match.

In DEV it is of higher version :

[hdfs@admin0.exp spark]$ more RELEASE

Spark 1.5.2.2.3.4.0-3485 built for Hadoop 2.7.1.2.3.4.0-3485

[hdfs@admin0.exp spark]$ pwd

/usr/hdp/2.3.4.0-3485/spark

In PROD it is of lower version :

[hdfs@admin1.prod.iad spark]$ more RELEASE

Spark 1.3.1.2.3.0.0-2557 built for Hadoop 2.7.1.2.3.0.0-2557

[hdfs@admin1.prod.iad spark]$ pwd

/usr/hdp/2.3.0.0-2557/spark

If you see closely HDP version/release is also different : In DEV it is 2.3.4.0-3485, whereas in PROD it is 2.3.0.0-2557

Do you advise me to upgrade the HDP or Can I just upgrade just the spark ?

[#2 · Accepted solution]

@Kumar Veerappan

We recommend to upgrade HDP stack so that you can get appropriate version of spark which comes with stack instead of manually installing spark.

[#3]

Please start from below doc according to your current and target version.

https://docs.hortonworks.com/HDPDocuments/Ambari-2.1.1.0/bk_upgrading_Ambari/content/_upgrading_hdp_...

[#4]

Hi @Kumar Veerappan please let me know if you need any other info on this or kindly accept the answer to close this thread.

[#5]

I am little confused..In Ambari the spark version is as follows :

Spark

1.5.2.2.3

Installed

Apache Spark is a fast and general engine for large-scale data processing.

[#6]

This is after upgrade? once you install it you can do cat RELEASE which will show you full version.

[#7]

Hi @Kumar Veerappan any further issue on this?

[#8]

@Jitendra Yadav This is even before the upgrade. I am little confused because in AMBARI the spark version says 1.5.2 whereas in side the node the version says 1.3.1

[#9]

@Kumar Veerappan did you finalized the upgrade?

[#10]

@Jitendra Yadav

Yes I have to upgrade the complete HDP stack from 2.3.0 to 2.3.4

[#11]

once upgrade completed we also need to clicked on finalize button to complete the upgrade process. if you have done this and still see previous version of spark then you should login to node and manually check spark package version through "rpm -qa|grep spark" command.

[#12]

@Jitendra Yadav

..thanks for your suggestion. Me being a newbie to this system : Can I upgrade through Ambari ?

[#13]

Agreed, you should at least upgrade the lower HDP version (...2.3.0...) to the newer HDP version (2.3.4.0-3485). It is best to get the default Spark version from the HDP install. Please see Table 1.1 at this link which describes the version associations for HDP, Ambari, and Spark:

http://docs.hortonworks.com/HDPDocuments/HDP2/HDP-2.4.2/bk_spark-guide/content/ch_introduction-spark...