Internal Error 500

[husky-parul]

Hi

I am trying to use OpenWhisk over OpenShift. The goal is to run embarrassingly parallelise algorithmic tasks and spinning large number of workers to test the scalability of the system. My concern is not the problem I am solving. It is to find out how many workers can be spinned.

I set these properties

name: WHISK_ACTIONS_INVOKES_CONCURRENT
  description: Default number of concurrenct actions per user
  value: "1000000"
  required: true
- name: WHISK_ACTIONS_INVOKES_CONCURRENT_IN_SYSTEM
  description: Number of concurrent actions allowed across the entire system
  value: "1000000"
  required: true
- name: WHISK_ACTIONS_INVOKES_PER_MINUTE
  description: Default number of action invocations per minute per user
  value: "1000000"
   required: true

For values greater than 10000, I get 500 Internal error. Is there a limit to these values. If yes, is there a way to override them?

[bbrowning]

I don't think there's any built-in limit to those values, although I could be wrong. When you say you get a 500 error, what component gave that error? Where did you see the error? You should be able to check the logs of the controller and/or invoker pods to see what's going on.

With that said, adjusting these values won't magically make the system scale to extremely high levels. You have to actually deploy a large enough cluster to handle the amount of load you want to throw at it. If you wanted to be able to run 1 million concurrent actions in the cluster as you show here you'll need a gigantic cluster with terabytes of RAM and hundreds if not thousands of Invoker nodes.