Make dataproc zone configurable (#760)

Author: rtitle

leonardo-example.conf

@@ -6,6 +6,9 @@
 # Google Cloud dataproc configuration
 dataproc {
   dataprocDefaultRegion = "YOUR_REGION"  # the google region for your dataproc
+  #dataprocZone = "YOUR_ZONE"            # Optional. The Google zone for the dataproc cluster.
+                                         # If not specified, Dataproc will automatically choose a zone
+                                         # within the configured region.
   leoGoogleProject = "GOOGLE_PROJECT"    # the name of the google project to use during cluster startup
                                          # *this is not the project the cluster will be created in
   clusterUrlBase = "https://YOUR_DOMAIN/notebooks" # the base url to access your cluster

src/main/scala/org/broadinstitute/dsde/workbench/leonardo/Boot.scala

@@ -75,7 +75,7 @@ object Boot extends App with LazyLogging {
     }
 
     val (leoServiceAccountEmail, leoServiceAccountPemFile) = serviceAccountProvider.getLeoServiceAccountAndKey
-    val gdDAO = new HttpGoogleDataprocDAO(dataprocConfig.applicationName, Pem(leoServiceAccountEmail, leoServiceAccountPemFile), "google", NetworkTag(dataprocConfig.networkTag), dataprocConfig.vpcNetwork.map(VPCNetworkName), dataprocConfig.vpcSubnet.map(VPCSubnetName), dataprocConfig.dataprocDefaultRegion, dataprocConfig.defaultExecutionTimeout)
+    val gdDAO = new HttpGoogleDataprocDAO(dataprocConfig.applicationName, Pem(leoServiceAccountEmail, leoServiceAccountPemFile), "google", NetworkTag(dataprocConfig.networkTag), dataprocConfig.vpcNetwork.map(VPCNetworkName), dataprocConfig.vpcSubnet.map(VPCSubnetName), dataprocConfig.dataprocDefaultRegion, dataprocConfig.dataprocZone, dataprocConfig.defaultExecutionTimeout)
     val googleComputeDAO = new HttpGoogleComputeDAO(dataprocConfig.applicationName, Pem(leoServiceAccountEmail, leoServiceAccountPemFile), "google")
     val googleIamDAO = new HttpGoogleIamDAO(dataprocConfig.applicationName, Pem(leoServiceAccountEmail, leoServiceAccountPemFile), "google")
     val googleStorageDAO = new HttpGoogleStorageDAO(dataprocConfig.applicationName, Pem(leoServiceAccountEmail, leoServiceAccountPemFile), "google")

src/main/scala/org/broadinstitute/dsde/workbench/leonardo/config/DataprocConfig.scala

@@ -7,6 +7,7 @@ import scala.concurrent.duration.FiniteDuration
 case class DataprocConfig(
                            applicationName: String,
                            dataprocDefaultRegion: String,
+                           dataprocZone: Option[String],
                            leoGoogleProject: GoogleProject,
                            dataprocDockerImage: String,
                            clusterUrlBase: String,

src/main/scala/org/broadinstitute/dsde/workbench/leonardo/config/package.scala

@@ -21,6 +21,7 @@ package object config {
     DataprocConfig(
       config.getString("applicationName"),
       config.getString("dataprocDefaultRegion"),
+      config.getAs[String]("dataprocZone"),
       GoogleProject(config.getString("leoGoogleProject")),
       config.getString("dataprocDockerImage"),
       config.getString("clusterUrlBase"),

src/main/scala/org/broadinstitute/dsde/workbench/leonardo/dao/google/HttpGoogleDataprocDAO.scala

@@ -38,6 +38,7 @@ class HttpGoogleDataprocDAO(appName: String,
                             vpcNetwork: Option[VPCNetworkName],
                             vpcSubnet: Option[VPCSubnetName],
                             defaultRegion: String,
+                            zoneOpt: Option[String],
                             defaultExecutionTimeout: FiniteDuration)
                            (implicit override val system: ActorSystem, override val executionContext: ExecutionContext)
   extends AbstractHttpGoogleDAO(appName, googleCredentialMode, workbenchMetricBaseName) with GoogleDataprocDAO {
@@ -244,6 +245,11 @@ class HttpGoogleDataprocDAO(appName: String,
       .setMachineTypeUri(machineConfig.masterMachineType.get)
       .setDiskConfig(new DiskConfig().setBootDiskSizeGb(machineConfig.masterDiskSize.get))
 
+    // Set the zone, if specified. If not specified, Dataproc will pick a zone within the configured region.
+    zoneOpt.foreach { zone =>
+      gceClusterConfig.setZoneUri(zone)
+    }
+
     // Create a Cluster Config and give it the GceClusterConfig, the NodeInitializationAction and the InstanceGroupConfig
     createClusterConfig(machineConfig, credentialsFileName)
       .setGceClusterConfig(gceClusterConfig)

src/test/resources/reference.conf

@@ -32,6 +32,7 @@ dataproc {
   applicationName = "test:leonardo"
   serviceAccountEmail = "test@test.com"
   dataprocDefaultRegion = "testregion"
+  dataprocZone = "test-zone"
   leoGoogleProject = "test-bucket"
   dataprocDockerImage = "testrepo/test"
   clusterUrlBase = "http://leonardo/"