Akka pre-mature termination - akka

Question

I'm experiencing early termination of the AkkaSystem after first record being read ,entering into the Dead letter without executing the task for all the records
I have 10records in my file , 2 of the records has the matching filename to be pushed to s3.
what could be going wrong here. Please suggest
Sample file record:
xxxxx,ABC,2019-05-10 00:11:00
yyyyyy,XYZ,2019-05-10 00:41:00
import akka.actor.{Actor, ActorSystem, Props}
import scala.io.Source
import scala.sys.process._
class HelloActor extends Actor {
def receive: Receive = {
case line: String => {
// print("Ahshan"+line)
val row = line.split ( "," )
val stdout = new StringBuilder
val stderr = new StringBuilder
val status = Seq ( "/bin/sh", "-c", "ls /Users/ahshan.md/Downloads/".concat ( row ( 2 ).substring ( 0, 10 ) ).concat ( "/*" ).concat ( row ( 0 ) ).concat ( "*" ) ) ! ProcessLogger ( stdout append _, stderr append _ )
// println(status)
// println("stdout: " + stdout)
// println("stderr: " + stderr)
if (status == 0) {
println ( "/bin/sh", "-c", "aws s3 cp ".concat ( stdout.mkString ).concat ( " " ).concat ( "s3://ahshan/".concat ( row ( 1 ) ).concat ( "/" ).concat ( row ( 0 ) ).concat ( ".email" ) ) )
}
else {
// println ( "File Not Found: " + row ( 0 ), row ( 1 ), row ( 2 ).substring ( 0, 10 ) )
println ( "stderr: " + stderr )
}
}
case "finished" => println ( "Hello Ahshan" )
case _ => println ( "Exiting" )
}
}
object AkkaHelloWorld extends App {
// an actor needs an ActorSystem
val system = ActorSystem ( "HelloSystem" )
// create and start the actor
val helloActor = system.actorOf ( Props [HelloActor], name = "helloActor" )
try {
val filename = "/Users/ahshan.md/Downloads/test.txt"
for (line <- Source.fromFile ( filename ).getLines) {
helloActor ! line
}
}
finally {
system.terminate ()
}
}

Answer 1

The ActorSystem and the actors run concurrently with your main thread, sending a message to an actor is async and the sending thread immediately continues, it does not wait for the actor to process the message.
This means that in your code the main thread fires off each line to the actor as fast as it can and then terminates the actor system. In this example it could make sense to move the termination logic into the actor and let the actor terminate the system when it has completed.
That will only work as long as your application is a single actor though, as soon as you add another actor you will have to revisit how you terminate the system.
As an additional note using the process API in Scala like that will block until the process has completed, calling blocking code can have bad consequences, read this section of the docs for more details: https://doc.akka.io/docs/akka/current/typed/dispatchers.html#blocking-needs-careful-management