def __init__( self, application ):
super( CreateSparkCluster, self ).__init__( application )
self.option( '--num-slaves', '-s', metavar='NUM',
type=int, default=1,
help='The number of slaves to start.' )
# We want --instance-type for the slaves and --master-instance-type for the master and we
# want --master-instance-type to default to the value of --instance-type.
super( CreateSparkCluster, self ).option(
'--instance-type', '-t', metavar='TYPE', dest='slave_instance_type',
default=SparkBox.recommended_instance_type( ),
help='The type of EC2 instance to launch for the slaves, e.g. t2.micro, '
'm3.small, m3.medium, or m3.large etc. ' )
self.option( '--master-instance-type', metavar='TYPE', dest='instance_type',
help='The type of EC2 instance to launch for the master, e.g. t2.micro, '
'm3.small, m3.medium, or m3.large etc. The default is the instance type '
'used for the slaves.' )
self.option( '--ebs-volume-size', metavar='GB', default=0,
help='The size in GB of an EBS volume to be attached to each node for '
'persistent data such as that backing HDFS. By default HDFS will be '
'backed instance store ( ephemeral) only, or the root volume for '
'instance types that do not offer instance store.' )
