Re: Ignite with Spark Intergration
Hi, In Spark you can use next options: * spark.driver.extraJavaOptions * spark.executor.extraJavaOptions You can path your IGNITE JVM options there like -DIGNITE_QUIET=false. Generally clients will be started on executors during data loading or data reading but you also can start them on driver side using Ignition.start(). BR, Andrei 12/12/2019 12:17 PM, datta пишет: Stopped client node in the machine where spark worker node was running and started spark shell. It started a ignite client node from within. I have only 1 problem that how to specify ignite jvm options from spark. it is taking default -Xms and -Xmx arguments which are very less -- Sent from: http://apache-ignite-users.70518.x6.nabble.com/
Re: Ignite with Spark Intergration
Stopped client node in the machine where spark worker node was running and started spark shell. It started a ignite client node from within. I have only 1 problem that how to specify ignite jvm options from spark. it is taking default -Xms and -Xmx arguments which are very less -- Sent from: http://apache-ignite-users.70518.x6.nabble.com/
Re: Ignite with Spark Intergration
I integrated spark and ignite using thrift client will that do ? https://github.com/kali786516/ApacheIgnitePoc/blob/master/src/main/scala/com/ignite/examples/spark/SparkClientConnectionTest.scala#L73 On Monday, December 9, 2019, Denis Magda wrote: > Hi, just ensure that "clientMode" is set in IgniteConfiguration that you > pass to Spark's IgniteContext object. Spark worker will spin up a client > node automatically for you and that one will reach out to the server > (assuming you properly configured Ignite discovery SPI in the same > IgniteConfiguration). > > - > Denis > > > On Sat, Dec 7, 2019 at 10:46 AM datta wrote: > >> Hi, >> >> Then what I have currently implemented is hopefully not embedded mode >> is it? >> >> Also I wanted to know if should install client nodes on spark worker nodes >> if spark is going to start a client node itself ? >> >> >> >> -- >> Sent from: http://apache-ignite-users.70518.x6.nabble.com/ >> > -- Thanks & Regards Sri Tummala
Re: Ignite with Spark Intergration
Hi, just ensure that "clientMode" is set in IgniteConfiguration that you pass to Spark's IgniteContext object. Spark worker will spin up a client node automatically for you and that one will reach out to the server (assuming you properly configured Ignite discovery SPI in the same IgniteConfiguration). - Denis On Sat, Dec 7, 2019 at 10:46 AM datta wrote: > Hi, > > Then what I have currently implemented is hopefully not embedded mode > is it? > > Also I wanted to know if should install client nodes on spark worker nodes > if spark is going to start a client node itself ? > > > > -- > Sent from: http://apache-ignite-users.70518.x6.nabble.com/ >
Re: Ignite with Spark Intergration
Hi, Then what I have currently implemented is hopefully not embedded mode is it? Also I wanted to know if should install client nodes on spark worker nodes if spark is going to start a client node itself ? -- Sent from: http://apache-ignite-users.70518.x6.nabble.com/
Re: Ignite with Spark Intergration
That’s just how the Spark integration works! I suppose you could use the Spark’s JDBC connection to access Ignite, but you’d lose some of the flexibility. Regards, Stephen > On 6 Dec 2019, at 17:04, datta wrote: > > Hi, > > I have installed ignite in 2 machines . > > 1 - server and 1 as client > > in the client machine i have installed spark and copied the the required > ignite jars inside the spark_home's jars folder. > > the problem is when i am starting spark and try to read a table in ignite > using spark dataframe api it is starting another client node from spark. > > How to prevent this from happening? > > > ignite version 2.5.0 > spark version 2.2.0 > > FYI I am using HortonWorks Ambari Platform. as i could not upgrade spark to > 2.3.0 as it would break other things i had to use ignite 2.5.0 which > supports 2.2.0 spark > > > > -- > Sent from: http://apache-ignite-users.70518.x6.nabble.com/
Ignite with Spark Intergration
Hi, I have installed ignite in 2 machines . 1 - server and 1 as client in the client machine i have installed spark and copied the the required ignite jars inside the spark_home's jars folder. the problem is when i am starting spark and try to read a table in ignite using spark dataframe api it is starting another client node from spark. How to prevent this from happening? ignite version 2.5.0 spark version 2.2.0 FYI I am using HortonWorks Ambari Platform. as i could not upgrade spark to 2.3.0 as it would break other things i had to use ignite 2.5.0 which supports 2.2.0 spark -- Sent from: http://apache-ignite-users.70518.x6.nabble.com/