How to add R studio service in Cloudera manager ?

[#1]

Hi friendz,

I am using Cloudera Quickstart 5.8 VMware.

In the same machine we have installed R & R studio .

All working fine.

Now my question is :

How to add R studio service in cloudera manager screen.

Please guide me.

Thanks,

Syam.

[#2]

I do not think you can add R through Cloudera Manager. You need to install from backend.

[#3]

Hi Vina,

Thanks for the reply,

R and R studio installed successfully.
I want to add R studio service in Cloudera manager to Stop, start, restart, monitor etc.
How much load it is taking etc activities ?

Thanks,
Syam.

[#4]

Hi,

As per I know, you cannot add extra service and it's operations along with
stats in Cloudera Manager. The CDH would be bundled with it's own setups.

[#5]

Oh!!
Will see, What update will Cloudera will give.
Thanks for the reply.

[#6]

You can add additional services but it will take work. You would need to build a CSD or Parcel for it. This is sometimes down by the vendor but I can't find anything for RStudio server.

The CSD provide the configuration information for the service (i.e. start, stop, restart, configuration, etc.). The parcel includes the CSD but also how to install the software.

https://www.cloudera.com/documentation/enterprise/5-5-x/topics/cm_mc_addon_services.html

[#7]

Hi syamsri!

Cloudera Manager needs a service descriptor for each service to monitor. Although the extension interface is open (see the documentation and extension tools) I’m not aware of such a connector for R-studio.

Please note though, that R-studio beeing a desktop application, it’s resource usage is probably not relevant in a cluster environment. If you’re using R-studio with sparklyr to submit jobs to a cluster, then your jobs will be executed by spark. You can consult our documentation on how to monitor spark applications.

[#8]

There is a way to deploy R Server 9.1 on a Cloudera cluster using parcels and run it as a service from Cloudera Manager. I have tested recently and the following guide from Microsoft explains the required steps:

https://docs.microsoft.com/en-us/machine-learning-server/install/r-server-install-cloudera

I have recently performend an installation or R Server and R Studio on Cloudera Manager & CDH 5.13, following the guides. The instructions are quite good, there are some gotchas though:

1) You can download the parcel from the sources listed under 

https://docs.microsoft.com/en-us/machine-learning-server/install/r-server-install-cloudera-generate-...

If you just want to test the setup the easiest way is to download R Studio from Visual Studio Dev Essentials. Just make sure you download the correct file for Hadoop (not for Linux):

Microsoft R Server 9.1.0 for Hadoop (x64) - (English)
No additional details are available for this file.
Released: 4/19/2017
SHA1: F6A03F858BB6733C8B074458CF18C102B64DD3A8
File name: en_microsoft_r_server_910_for_hadoop_x64_10323951.tar.gz

2) You can unpack the files and run the generate_mrs_parcel.sh script to generate the parcel, parchel SHA and the CSD file:

MRS-9.1.0-[..].parcel (file name depends on select OS version)

MRS-9.1.0-[..].parcel.sha

MRS-9.1.0-CONFIG.JAR

3) You may also want to download Microsoft R Open for Microsoft R Server. Fortunately, R Open is also available as a Cloudera parcel: https://mran.revolutionanalytics.com/download/mro-for-mrs/

Select the following:

R Open for Microsoft R Server – Hadoop Cloudera Manager Parcel
sha256: 317f6b455411303363d6694e7b9c055bffef010d5681e52054be31e32b57b6f9

The archive contains both a parcel and parcel sha file. Both should go under /opt/cloudera/parcel-repo/ as well

I was installing on CentOS 6.8 and I ran into an issue where the file names for the generated parcel were not correct (MRO-3.3.2-_dist_.parcel). If you have something like this, make sure to rename the parcel and parcel sha files according to your RHEL/CentOS version:

MRO-3.3.2-el6.parcel

MRO-3.3.2-el6.parcel.sha

Preferably do this in the folder where the files were generated and only move to /opt/cloudera/parcel-repo/ AFTER you have renamed them. (Otherwise Cloudera Manager may not find the new parcels and an additional restart of the cloudera-scm-server service on the CM host may be needed.)

Conclusion

If you have done everything correctly, clicking Check for New Parcels in Cloudera Manager (top right corner -> Parcels page) will refresh the list of available parcels for MRO and MRS should show up. If they do not show, check the exact filename (should represent the OS version), their location (/opt/cloudera/parcel-repo/), owner (cloudera-scm:cloudera-scm) and permissions (644). You may also need to restart the cloudera-scm-server service.

Installing R Studio Server

As mentioned earlier, there is no parcel available for R Studio Server, this needs to be managed manually.

There is a short guide on how to get it up and running on Cloudera

https://support.rstudio.com/hc/en-us/articles/235841328-Using-RStudio-Server-with-Microsoft-R-Server...

I also needed some additional steps, which are very well documented in a Microsoft blog post:

https://blogs.msdn.microsoft.com/microsoftrservertigerteam/2017/04/30/using-rstudio-server-with-micr...