Here: https://apacheignite-fs.readme.io/docs/installation-deployment  I found 
“Embedded Deployment” section where is it written: 

Embedded deployment means that Apache Ignite nodes are started inside Apache 
Spark job processes and are stopped when job dies. There is no need for 
additional deployment steps in this case. Apache Ignite code will be 
distributed to the worker machines using Apache Spark deployment mechanism and 
nodes will be started on all workers as a part of IgniteContext initialization.

How it can be? Is any examples? Or I need just to start IgniteContext without 
additional settings in my app and all will works?

From: Andrey Yatsuk
Sent: 13 декабря 2017 г. 17:15
To: [email protected]
Subject: Ignite yarn cluster deployment

Hi!
Our application is launched from remotely pc via spark-submit in yarn-cluster 
mode with Kerberos keytab and principals by this guide: 
https://spark.apache.org/docs/latest/running-on-yarn.htmll (I can describe more 
in detail if it is required.) The advantages of this approach are that I have 
my own version of the spark on any cluster.

Is it possible to automatically deploy Apache Ignite into a Hadoop YARN cluster 
without sftp/ssh only by running a bash-script with 
HADOOP_CONF_DIR/YARN_CONF_DIR configs,

Also on this guide: https://apacheignite.readme.io/docs/yarn-deployment you 
need to set up IGNITE_NODE_COUNT. Why you can’t get number of nodes in the 
cluster from YarnClient?

Best regards,
Andrey Yatsuk.

Reply via email to