Adding new gateway to Hadoop cluster

[#1]

We have a requirement to add a new gateway to hadoop cluster in additional to the one we already have. I have a couple of queries if anyone can help.

1. Are there any concerns in adding multiple gateways to a CDH cluster?

2. I came across the below article describing how to set up a gateway outside the cluster.

https://cloudera-portal.force.com/articles/KB_Article/Install-Cloudera-Client-on-non-Cloudera-server...

We have a kerberized cluster. Is it recommended to set up a gateway outside and will it be possible to submit jobs from this external gateway? (We have Oozie and Httpfs services running in a different host within the cluster). Are there any disadvantages to this approach?

Thank you in advance for your valuable inputs.

[#2 · Accepted solution]

Hello,

You can add as many gateways as you want. If you add a gateway outside of Cloudera Manager  then the only disadvantage is that you need to do every step manually. For example these:

• installation of client configurations
• installation of client packages
• configuration of kerberos (/etc/krb5.conf)
• etc.

Gabor

[#3]

@roczei Thank you for your response.

Do you mean to say that functionally, there will be no difference in a gateway set up outside? I can submit applications and access HDFS as I would do from a gateway within the cluster, correct?

[#4]

@AnandMSYes.

[#5]

Thank you.