CODEWITHSUNDEEP
elixir
Nathaniel Johnson
Nathaniel Johnson

Reputation: 623

Elixir lang processing lists in parallel

I'm currently having an issue processing lists in elixir in parallel. The reason for the parallelism is that I'm saving the results to an API and if I blast them all at once it gets DDOS'd and shuts down.

The code below is supposed to split the result of a SQL query and process each row in a separate task and when all the tasks have been completed, it should terminate.

What happens is that the first task, after firing a message, causes the script to terminate. I've seen answers where they put the receive in a function and that function calls itself over and over but I feel like there has to be another better way to handle this.

results = Enum.chunk(results, 500)

# Give this process a name
Process.register(self(), :core)

# Loop over the chunks making a process for each
Enum.each results, fn(result) ->
  task = Task.async(fn -> Person.App.process(result, "Test", "1") end)
end

# And listen for messages
receive do
  {:hello, msg} -> IO.inspect msg
  {:world, _} -> "won't match"
end

Upvotes: 9

Views: 2832

Answers (1)

Paweł Obrok
Paweł Obrok

Reputation: 23164

When using Task.async it is most convenient to obtain the result with Task.await:

results
|> Enum.map(fn result -> Task.async(fn -> Person.App.process(result, "Test", "1") end) end)
|> Enum.map(&Task.await/1)
|> Enum.each(&IO.inspect/1)

In fact if you don't await for the result of async it will still be sent to the process that called async and stored in its mailbox, potentially causing a memory leak! If your intention is to create a Task where you don't care about the result use Task.start or Task.start_link instead.

Upvotes: 13

Related Questions