a visual debugger for Akka actor systems (experimental)
Current version is built agains akka 2.4.10.
To get started you need to add dependencies for "monitoring" jar:
resolvers += Resolver.bintrayRepo("lustefaniak", "maven")
libraryDependencies += "com.blstream.akkaviz" %% "monitoring" % "0.1.7"
After that you must configure your application to run with AspectJ weaver.
Simpler way is to use sbt plugin, which will preconfigure required bits, in your project directory create file project/akkaviz.sbt
with:
resolvers += Resolver.url("lustefaniak/sbt-plugins", url("https://dl.bintray.com/lustefaniak/sbt-plugins/"))(Resolver.ivyStylePatterns)
addSbtPlugin("com.blstream.akkaviz" % "sbt-akka-viz" % "0.1.7")
Please check lustefaniak/akka-viz-demo for simple, preconfigured SBT project.
- see which actors exchange messages on a graph
- monitor internal state via reflection
- track FSM transitions
- monitor actor creation
- display messages in realtime (with contents) for selected actors
- filter messages by class
- delay processing of messages
- display information about exceptions in actors and used supervisor strategies
To change one of the akka-viz default settings create application.conf
file as defined in Typesafe Config documentation
Default settings are:
akkaviz {
internalSystemName = akka-viz
interface = 127.0.0.1
port = 8888
bufferSize = 10000
maxSerializationDepth = 3
inspectObjects = false
autoStartReporting = true
enableArchive = true
cassandra {
keyspace = akkaviz
preparedStatementCacheSize = 100
session {
contactPoint = 127.0.0.1
}
}
}
Akka-viz can use Cassandra database to store all messages so they could be replayed in the frontend for easier debugging.
To use it make sure akkaviz.enableArchive = true
and akkaviz.cassandra
points to correct Cassandra configuration.
You need to manually create keyspace and tables. Execute included DDL via cqlsh
We supply a few demos with the source code, so you can explore without hooking up to an existing ActorSystem.
Clone project and run with demo/reStart
in SBT. Go to http://localhost:8888
in your favourite browser and play around!
Include jar in classpath of your app and run your main()
. Server should be listening on http://localhost:8888
.
To hook into Akka's internals, we use AspectJ (capturing actor creation, intercepting of messages). WebSocket handling is provided by Akka HTTP and Akka Streams. Filtering is done per client; slowing down of message processing, however, affects the whole system. Frontend is powered by Scala.js, Scala.Rx and ScalaTags.
We're using custom-made serialization with help of upickle for transporting the messages to frontend clients. To serialize some specific type of message - ie. your custom class, that you want to have displayed in web frontend - you have to implement a AkkaVizSerializer.