Can I migrate from RPMs to Parcels for Manager Express Version?

[#1]

The manager software did not give me a parcels option.  I moved forward using RPMs.  Can I migrate to parcels now? I have installed MRv1, YARN, Hive, Impala, Hue, HBase.

[#2]

Have a read if this page and let us know if you have any questions

http://www.cloudera.com/content/cloudera/en/documentation/core/latest/topics/cm_ig_migrating_package...

[#3]

Thanks for the link, I have seen the article and gone through it.  A couple of follow on questions:

1. Does the Cloudera manager express version support RPMs not parcels?  Is this expected behavior?

2. The cloudera manager is still express version - will moving to parcels from RPMs be allowed?

3. Will the uninstall packages step undo the parcel based install and activation?

Thanks!

[#4]

> 1. Does the Cloudera manager express version support RPMs not parcels?  Is this expected behavior?

Cloudera Manager supports parcels and RPMs, doesn't matter if it is express or enterprise.

> 2. The cloudera manager is still express version - will moving to parcels from RPMs be allowed?

Yes

> 3. Will the uninstall packages step undo the parcel based install and activation?

No.

[#5]

Thanks, Gautam.  That leaves me confused as to why my installation didnt find any parcels.  Hopefully, I can move forward through the migration process.  The reason for this line of questions is: I need to add Spark Service.  I am now currently RPM based installed.  When I look available spark parcels, it's based on cdh4.  I tried that it failed in the activation phase due to the current RPM based install. The other option seems to be RPM based on a bunch of hosts.

Question: How do I install Spark?  I cannot seem to add a service now. Do I have any options?

Thanks, 

Anant

[#6]

You could have mentioned the Spark requirement earlier 😉 I believe
there are Spark RPMs, see here:
http://archive.cloudera.com/cdh5/redhat/6/x86_64/cdh/5/RPMS/noarch/

So download the CDH5 repo

# cd /etc/yum.repos.d
# wget http://archive.cloudera.com/cdh5/redhat/6/x86_64/cdh/cloudera-cdh5.repo
# yum makecache
# yum list | grep spark

[#7]

I am trying to determine if it is better to move to a parcel and then install spark vs directly installing via RPMs.

You have given me the RPM path.  There are the following RPMs: spark-core, spark-history-server, spark-master, spark-python, spark-worker.

I have 4 master nodes and 3 data nodes. Can you please recommend which RPMs run where?  I cant seem to find proper documentation on that. 

Thanks for your patience,

Anant

[#8]

To make things easy, you can install all the RPMs on all the nodes. Then visit CM and add the Spark service. I have been told to inform you that it is better to use Spark on YARN as of now instead of standalone Spark. This is due to performance considerations

[#9 · Accepted solution]

I was planning to run spark with yarn, glad to be on the same page with you.  Thank you for walking me through the steps! It is so much clearer, thanks for your patience and directions.  

Regards,

Anant