leonardo-example.conf

# Example config file for leonardo.
# Anything commented out here will have have defaults specified or is not necessary to run leo.
# All values can be modified for your specific google environment, database, and security certificates


# 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

  # This tag is used to associate the cluster's master with Leonardo's firewall rule.
  networkTag = "leonardo"

  # Optional. If defined, reads the labels on the Google project in which the cluster is launched, and looks
  # up the value of this key in the labels to determine the VPC network in which to launch it.
  # This config parameter takes precedence over vpcNetwork below.
  #projectVPCNetworkLabel = "vpc-network-name"

  # Optional. If defined, reads the labels on the Google project in which the cluster is launched, and looks
  # up the value of this key in the labels to determine the VPC subnetwork in which to launch it.
  # This config parameter takes precedence over vpcSubnet below.
  #projectVPCSubnetLabel = "vpc-subnetwork-name"

  # Optional. The VPC network of the cluster, and the network the firewall rule is applied to.
  # If unset, the default network of the project is used, if it exists.
  # A full URL, partial URI, or short name are valid. Examples:
  # - https://www.googleapis.com/compute/v1/projects/[project_id]/regions/global/default
  # - projects/[project_id]/regions/global/default
  # - default
  #vpcNetwork = "default"

  # Optional. The subnet on which the cluster is launched. If BOTH vpcNetwork and vpcSubnet are specified,
  # only vpcSubnet is passed to Dataproc. If vpcSubnet is unset, a project-default is used.
  # If the subnet is in a shared VPC, this field can accept a partial URI (ex.
  # "projects/[projectId]/regions/us-east1/sub0") in addition to the short-name form "sub0".
  #vpcSubnet = "default"

  # Optional. The custom dataproc image to use. If not specified Leo will choose a public Dataproc
  # image appropriate for the docker image being installed.
  # See: https://cloud.google.com/dataproc/docs/guides/dataproc-images
  #customDataprocImage = "projects/[project_id]/global/images/[image_name]"
}

# These are the cluster defaults that have already been configured.
# You should only set them in your config if you want a different value
#clusterDefaults {
  #numberOfWorkers = 0            # Standard, non-preemptible workers; Can't be 1
  #masterMachineType = "n1-standard-4"
  #masterDiskSize = 100           # Minimum is 10
  #workerMachineType = "n1-standard-4"
  #workerDiskSize = 100           # Minimum is 10
  #numberOfWorkerLocalSSDs = 0    # Can be 0-8
  #numberOfPreemptibleWorkers = 0 # Must be 0 if number of workers is 0
#}

# Database connection information
mysql {
  db {
    url = "jdbc:mysql://YOUR_DB_HOST/leonardo?allowPublicKeyRetrieval=true&useSSL=false&rewriteBatchedStatements=true&nullNamePatternMatchesAll=true"
    user = "USER_NAME"
    password = "PASSWORD"
  }
}

# To use the swagger page, supply a service account id and realm
#swagger {
  #googleClientId = ""
  #realm = ""
#}

# Secure, 2-way communication between leo and jupyter
# See https://github.com/DataBiosphere/leonardo/CERTIFICATES.md for info on setting up the certificates

# Jupyter proxy server configuration
proxy {
  # Should match the jupyter wildcard cert specified in command above
  jupyterDomain = "JUPYTER_DOMAIN_NAME"
}

# Keys and certificate authorities for cluster
clusterFiles {
  configFolderPath = "/etc/"
  jupyterServerCrt = "jupyter-server.crt"
  jupyterServerKey = "jupyter-server.key"
  jupyterRootCaPem = "rootCA.pem"
  jupyterRootCaKey = "rootCA.key"
}


akka {
  loglevel = INFO

  http {
    server.idle-timeout = 1 hour
    client.idle-timeout = 1 hour
    server.request-timeout = 60 seconds
    parsing.max-content-length = 64m
  }

  #Add your ssl config info below
  ssl-config {
    # Key store with leo client credentials
    keyManager = {
      stores = [
        {
          type = "PKCS12"
          path = "/etc/leo-client.p12"
          password = "PW_KEYSTORE"
        }
      ]
    }
    trustManager = {
      stores = [
        # Certificate store that recognizes the jupyter certificate
        {
          type = "PEM"
          path = "/etc/rootCA.pem"
        },
        # Default trust store - no need to modify
        {
          path: ${java.home}/lib/security/cacerts
          password = "changeit"
        }
      ]
    }
  }
}

# Authorization implementation config
auth {
  # To use your own provider implementation, subclass org.broadinstitute.dsde.workbench.leonardo.model.LeoAuthProvider and provide the class name here
  providerClass = "org.broadinstitute.dsde.workbench.leonardo.auth.AllowlistAuthProvider"  # A default auth provider
  providerConfig { # Do not remove providerConfig, even if empty
    # Any provider specific config should be specified here
    # For the allowlist provider, specify a comma separated list of user emails that should be allowed access
    allowlist = [EMAIL1,EMAIL2]

    # Amount of time Leo will wait for a provider response before timing out.
    # This should be set less than akka.http.server.request-timeout.
    # This option is supported for ANY provider implementation. If not specified, the default is 30 seconds.
    providerTimeout = 30 seconds

    # Permissions used for IamProxyAuthProvider. The `resourcemanager.projects.update` permission is shared by
    # project owners and project editors, while the `resourcemanager.projects.setIamPolicy` is only given to
    # project owners and organization admin roles.
    # requiredProjectIamPermissions = ["resourcemanager.projects.update"]

    # Application name used by the IamProxyAuthProvider (should match application name used by the dataproc config).
    # applicationName = "firecloud:leonardo"
  }

}

# Implement and specify a class that will provide appropriate service accounts
serviceAccounts {
  # To implement your own service account provisioning, subclass org.broadinstitute.dsde.workbench.leonardo.model.ServiceAccountProvider
  # and specify your concrete class here
  providerConfig { # Do not remove config, even if empty
    # Specify any config information your class needs here

    # Amount of time Leo will wait for a provider response before timing out.
    # This should be set less than akka.http.server.request-timeout.
    # This option is supported for ANY provider implementation. If not specified, the default is 30 seconds.
    providerTimeout = 30 seconds
  }
}

#Configs to auto stop cluster after idle period
autoFreeze {
  #This is used to enable auto freeze
  enableAutoFreeze = true
  #This is the duration after which the date accessed is persisted in the ClusterDateAccessedActor
  dateAccessedMonitorScheduler = 1 minute
  #Idle duration after which the cluster is frozen/stopped automatically
  autoFreezeAfter = 30 minutes
  #This is the polling period to poll for clusters that can be auto stopped/frozen
  autoFreezeCheckScheduler = 1 minutes
}

# Configures the "zombie cluster" monitor, which periodically sweeps the database and checks for clusters
# and projects that no longer exist in Google.
zombieClusterMonitor {
  # Whether the zombie cluster monitor is enabled
  enableZombieClusterMonitor = true
  # The period to check for zombie clusters and projects
  pollPeriod = 30 minutes
}