Adapt the DB config to match with pureconfig switch

[chetanmeh]

For changes done via apache/openwhisk#3369

[jthomas]

Looks like the build has got stuck on this: https://travis-ci.org/apache/incubator-openwhisk-devtools/builds/352461291

Can someone restart it?

[csantanapr]

I restarted here: https://travis-ci.org/apache/incubator-openwhisk-devtools/builds/352461291
@jthomas you have write access to the repo you can restart any travis build.

[csantanapr]

I think is a problem with controller now being https by default instead of http

apache/openwhisk@a8476ab

cc @vvraskin

[csantanapr]

hum, it might not be related to https

until $(curl --output /dev/null --silent --head --fail http://10.20.0.239:8888/ping); do printf '.'; sleep 5; done
..........................................................................................................................................................................................................................................

Since the default in purefonfig is http
CONFIG_whisk_controller_protocol
apache/openwhisk@a8476ab#diff-de8780464afb147e69223943ad23af93R11

[chetanmeh]

The build is getting stuck in controller startup which fails with following error

Exception in thread "main" akka.ConfigurationException: ActorSystem [akka://controller-actor-system] needs to have a 'ClusterActorRefProvider' enabled in the configuration, currently uses [akka.actor.LocalActorRefProvider]
	at akka.cluster.Cluster.<init>(Cluster.scala:71)
	at akka.cluster.Cluster$.createExtension(Cluster.scala:34)
	at akka.cluster.Cluster$.createExtension(Cluster.scala:29)
	at akka.actor.ActorSystemImpl.registerExtension(ActorSystem.scala:880)
	at akka.actor.ExtensionId$class.apply(Extension.scala:78)
	at akka.cluster.Cluster$.apply(Cluster.scala:29)
	at whisk.core.loadBalancer.ShardingContainerPoolBalancer.<init>(ShardingContainerPoolBalancer.scala:60)
	at whisk.core.loadBalancer.ShardingContainerPoolBalancer$.loadBalancer(ShardingContainerPoolBalancer.scala:295)
	at whisk.core.controller.Controller.<init>(Controller.scala:121)
	at whisk.core.controller.Controller$.main(Controller.scala:232)
	at whisk.core.controller.Controller.main(Controller.scala)

This is possibly due to switch of default LoadBalancerProvider to ShardingContainerPoolBalancer which instantiates the cluster. Currently the docker compose setup has CONFIG_akka_actor_provider=local set. Not sure if just setting it to cluster would make it work

@ddragosd As we switched to local earlier and going forward ContainerLoadBalanacer would be removed we would need to enable cluster for docker-compose. Thoughts?

@csantanapr Would it be possible to configure the log upload script for devtools as its configured in main repo. That would help in accessing the logs

[csantanapr]

Yes go ahead and do the work to collect anything you need into a tgz then later I can hook it up with an upload with @rabbah

[chetanmeh]

The build now passed with switch to cluster mode. Locally also it works fine. Would look into the log file upload in a separate PR