Updates to the readme, and build.sbt

README.md

@@ -6,7 +6,10 @@ I'm sure there are other distributed tools that will do this for me, but I wante
 
 ## Design Notes
 
-Akka actor system
+Using jssh, bouncycastle, and jzlib for the SSH parts.
+
+
+### Akka actor system
 
 TimerActor -> tells when to start -> BackupCoordinatorActor
 
@@ -15,17 +18,33 @@ BackupCoordinatorActor loads the configuration, which details what hosts to back
 Creates a per host actor, that is responsible for that host, and sending the commands.
  * Per host actor will have a list of commands to be run on each host
  * commands are assumed to be in dependency order
+ * Hosts are executed in sequence for now
 
 HostActor creates an actor for each SSH command.
  * HostActor is responsible for the state of command execution, and should keep timer metrics and stuff 
+ * HostActor should maintain the SSH connection and hand it to command actors to use.
+ * It's also responsible for fixing the SSH connection if it dies.
+ * commands will look something like:
+     * Mount network backup location
+     * run pre-backup command (mysql and databases or whatever) 
+     * Run backup command <This command may loop up one for many backups on the same host>
+     * un mount network backup location
 
 SSHCommandActor Only responsible for running the command on the remote host.
- * Returns the output of the command to the HostActor (probably buffering it right now) 
+ * Returns the output of the command to the HostActor
+     * Would be cool to do live streaming updates of stdout v2
+ * will receive an ssh context from the HostActor, since it maintains SSH state
+ 
  
 BackupCoordinatorActor needs to run a command when it's all done. Say an rclone up into the cloud.
-Need to specify a host for that to run on, or it runs locally. That is also it's own actor
+Need to specify a host for that to run on, or it runs locally. That is also it's own actor.
+Should always run the command, even if one of the hosts failed.
 
-PostRunActor Only responsible for the post run command, after all hosts have backed up, do this work
+PostRunActor Only responsible for the post run command, after all hosts have backed up, do this work.
+
+ReportingActor should then email a report of all the things that have been done.
+Data will be in the BackupCoordinatorActor, buffered until it's all finished, and then delivered to the ReportingActor.
+Then to be emailed or archived, or whatever.
 
 ### Example Configuration
 
@@ -41,6 +60,9 @@ PostRunActor Only responsible for the post run command, after all hosts have bac
 
 #### Configuration yaml
 
+TODO: configuration might need something that has a schema. That might be easier to deal with. 
+JSON is terrible at multiline strings, and XML can be way too verbose. HM
+
 ```yaml
 backupTemplate: |
   borg create \

build.sbt

@@ -9,9 +9,13 @@ val akkaTestkit = "com.typesafe.akka" %% "akka-testkit" % akkaV % Test
 val scalaTest = "org.scalactic" %% "scalactic" % scalaTestV
 val scalacticScalaTest = "org.scalatest" %% "scalatest" % scalaTestV
 
+val sshj = "com.hierynomus" % "sshj" % "0.26.0"
+val bouncycastle = "org.bouncycastle" % "bcprov-jdk15on" % "1.60"
+val jzlib = "com.jcraft" % "jzlib" % "1.1.3"
+
 lazy val backupCoordinator = (project in file("."))
   .settings(
     name := "Backup Coordinator",
-    libraryDependencies ++= Seq(akka, akkaTestkit, scalaTest, scalacticScalaTest)
+    libraryDependencies ++= Seq(akka, akkaTestkit, scalaTest, scalacticScalaTest, sshj, bouncycastle, jzlib)
   )