Created 07-01-2019 09:42 PM
hi all
I want to hear from hortonworks guys about the following idea
we want to build the following architecture
1. ambari cluster that include ( HDP 3.1 )
3-5 kafka machines
3 zookeper servers that are dedicated only for the kafka machines
2. ambari cluster that include all other application ( HDP 3.1 )
HDFS - service
YARN - service
MAPREDUCE - service
HIVE - service
SPARK
3 Zookeepers that dedicated to hdfs , yarn , etc
data nodes machines
so both cluster should talk with each other like one big cluster
is this possible ?
Created 07-01-2019 10:55 PM
Regarding your query: "2 ambari clusters in one big cluster"
Ambari can currently manage only one cluster deployed to it. So from ambari perspective you manage two cluster from same Ambari.
However, You can run two clusters running separately and still be able to establish communication between Kafka cluster to the other cluster which has all the services running.
Like Spark running in Cluster two can interact with External Kafka Broker running inside Cluster one.
Created 07-01-2019 10:55 PM
Regarding your query: "2 ambari clusters in one big cluster"
Ambari can currently manage only one cluster deployed to it. So from ambari perspective you manage two cluster from same Ambari.
However, You can run two clusters running separately and still be able to establish communication between Kafka cluster to the other cluster which has all the services running.
Like Spark running in Cluster two can interact with External Kafka Broker running inside Cluster one.
Created 07-02-2019 03:26 AM
@DEAR Jay "clusters running separately and still be able to establish communication between Kafka cluster to the other cluster which has all the services "
can you please explain in basic details what are the configuration on both cluster to enable contamination between one ambari cluster to the ambari cluster that include the kafka?
Created 07-02-2019 03:32 AM
What i mean to say is this ::: At AmbariServer level you can not manage two clusters as a Single cluster.
But if you want to Produce Messages to Kafka (For example via Spark Application then that can be done even in Non Ambari Managed cluster as well right)
For Example: In your Spark application define the correct Kafka Broker host/port ...etc like following:
https://spark.apache.org/docs/latest/streaming-kafka-0-10-integration.html
Created 07-02-2019 01:19 AM
The above question and the entire reply thread below was originally posted in the Community Help track. On Tue Jul 2 01:01 UTC 2019, a member of the HCC moderation staff moved it to the Design & Architecture track. The Community Help Track is intended for questions about using the HCC site itself, not technical questions.