Each agent should check if enough replicas are available to ensure the configured consistency level.

the custom-sentinel entry shold be optional into the application.conf

An agent executes a snapshot requested by the orchestrator.

Create a REST Endpoint to the orchestrator

Define default configuration values for "predefined" sentinels.

To avoid complexity, a single sentinel with a Jmx ObjectName and ObjectAttribute will perform the validation on the underlying metric (converted in double).

This issue overrides the issues #11 to #16

Currently, all ThreadPool sentinels check if there are some BlockingTasks.

We should also provide a way to notify if there are too many pending tasks.

With VNodes, nodetool describering return more than one TokenRange each one with the list of replicas.
This list of replicas may contain the same Endpoints so if a consistency issue is detected several Notifications are pushed for a single set of nodes.

We have to send only one notification per set of endpoints

Create a sentinel that checks the GCInspector state.
If the GC duration is too long, send a notification.

The cassandra version should be specified into the configuration file to adapt the JMX request and avoid the manual switch off of the Sentinels.

Cassandra 2.1 return an array of 6 elements instead of 7 in Cassandra 2.2 and more.

Notifier should be able to send notification only if the 'Level' is relevant.

With Cassandra 2.1

[ERROR] [10/16/2016 04:02:45.720] [Argos-akka.actor.default-dispatcher-2479] [akka://Argos/user/SentinelOrchestrator/notification-jmx] [I cannot be cast to java.util.HashMap
java.lang.ClassCastException: [I cannot be cast to java.util.HashMap
at io.argos.agent.sentinels.InternalNotificationsSentinel$$anonfun$processProtocolElement$1.applyOrElse(InternalNotificationsSentinel.scala:28)
at scala.runtime.AbstractPartialFunction.apply(AbstractPartialFunction.scala:36)
at io.argos.agent.sentinels.Sentinel$$anonfun$receive$1.applyOrElse(Sentinel.scala:33)
at akka.actor.Actor$class.aroundReceive(Actor.scala:484)
at io.argos.agent.sentinels.Sentinel.aroundReceive(Sentinel.scala:22)
at akka.actor.ActorCell.receiveMessage(ActorCell.scala:526)
at akka.actor.ActorCell.invoke(ActorCell.scala:495)
at akka.dispatch.Mailbox.processMailbox(Mailbox.scala:257)
at akka.dispatch.Mailbox.run(Mailbox.scala:224)
at akka.dispatch.Mailbox.exec(Mailbox.scala:234)
at scala.concurrent.forkjoin.ForkJoinTask.doExec(ForkJoinTask.java:260)
at scala.concurrent.forkjoin.ForkJoinPool$WorkQueue.runTask(ForkJoinPool.java:1339)
at scala.concurrent.forkjoin.ForkJoinPool.runWorker(ForkJoinPool.java:1979)
at scala.concurrent.forkjoin.ForkJoinWorkerThread.run(ForkJoinWorkerThread.java:107)

Argos is not very clean, change it!

Scala : 2.11.7 to 2.12.1
Akka: 2.4.7 to 2.4.16
ScalaTest : 2.2.6 to 3.0.1

Create an agent orchestrator.
The orchestrator will receive some information from each agent (like the LoadAvg of a node...) and it will be able to send information/Request to each agent (ex : request a snapshot)

After a first Dropped Message alert, other notifications are sent because, we test the last minute rate.
This is a float and even after 40 minutes, this value will be greater than 0...