def test_allow_string_cluster(self):
spot = "2"
withSpotPrice = "WithSpotPrice"
cluster = emr.Cluster(
'Cluster',
# AdditionalInfo="Additional Info",
Applications=[
emr.Application(Name="Hadoop"),
emr.Application(Name="Hive"),
emr.Application(Name="Mahout"),
emr.Application(Name="Pig"),
emr.Application(Name="Spark")
],
BootstrapActions=[
emr.BootstrapActionConfig(
Name='Dummy bootstrap action',
ScriptBootstrapAction=emr.ScriptBootstrapActionConfig(
Path='file:/usr/share/aws/emr/scripts/install-hue',
Args=["dummy", "parameter"]))
],
Configurations=[
emr.Configuration(Classification="core-site",
ConfigurationProperties={
'hadoop.security.groups.cache.secs':
'250'
})
],
Instances=emr.JobFlowInstancesConfig(
Ec2KeyName="KeyName",
Ec2SubnetId="SubnetId",
MasterInstanceGroup=emr.InstanceGroupConfigProperty(
InstanceCount="1",
InstanceType=M4_LARGE,
AutoScalingPolicy=emr.AutoScalingPolicy(
Constraints=emr.ScalingConstraints(MinCapacity="1",
MaxCapacity="3"),
Rules=self.generate_rules("MasterAutoScalingPolicy")),
),
CoreInstanceGroup=emr.InstanceGroupConfigProperty(
Name="Core Instance",
BidPrice=If(withSpotPrice, Ref(spot), Ref("AWS::NoValue")),
Market=If(withSpotPrice, "SPOT", "ON_DEMAND"),
InstanceCount="1",
InstanceType=M4_LARGE,
AutoScalingPolicy=emr.AutoScalingPolicy(
Constraints=emr.ScalingConstraints(MinCapacity="1",
MaxCapacity="3"),
Rules=self.generate_rules("CoreAutoScalingPolicy"),
)),
),
JobFlowRole="EMRJobFlowRole",
LogUri="s3://cluster-logs",
Name="EMR Cluster",
ReleaseLabel="emr-5.5.0",
ServiceRole="EMRServiceRole",
AutoScalingRole="EMR_AutoScaling_DefaultRole",
VisibleToAllUsers="true",
Tags=Tags(Name="EMR Sample Cluster"))
cluster.to_dict()
})
])
],
JobFlowRole=Ref(emr_instance_profile),
ServiceRole=Ref(emr_service_role),
AutoScalingRole=emr_autoscaling_role,
Instances=emr.JobFlowInstancesConfig(
Ec2KeyName=Ref(keyname),
Ec2SubnetId=Ref(subnet),
MasterInstanceGroup=emr.InstanceGroupConfigProperty(
Name="Master Instance",
InstanceCount="1",
InstanceType=M4_LARGE,
Market="ON_DEMAND",
AutoScalingPolicy=emr.AutoScalingPolicy(
Constraints=emr.ScalingConstraints(MinCapacity="1",
MaxCapacity="3"),
Rules=generate_rules("MasterAutoScalingPolicy"))),
CoreInstanceGroup=emr.InstanceGroupConfigProperty(
Name="Core Instance",
BidPrice=If(withSpotPrice, Ref(spot), Ref("AWS::NoValue")),
Market=If(withSpotPrice, "SPOT", "ON_DEMAND"),
AutoScalingPolicy=emr.AutoScalingPolicy(
Constraints=emr.ScalingConstraints(MinCapacity="1",
MaxCapacity="3"),
Rules=generate_rules("CoreAutoScalingPolicy"),
),
EbsConfiguration=emr.EbsConfiguration(EbsBlockDeviceConfigs=[
emr.EbsBlockDeviceConfigs(
VolumeSpecification=emr.VolumeSpecification(
SizeInGB="10", VolumeType="gp2"),
VolumesPerInstance="1")
def test_allow_string_cluster(self):
cluster_security_configuration = emr.SecurityConfiguration(
'emrsecurityconfiguration',
Name="EMRSecurityConfiguration",
SecurityConfiguration=security_configuration)
spot = "2"
withSpotPrice = "WithSpotPrice"
cluster = emr.Cluster(
'Cluster',
# AdditionalInfo="Additional Info",
Applications=[
emr.Application(Name="Hadoop"),
emr.Application(Name="Hive"),
emr.Application(Name="Mahout"),
emr.Application(Name="Pig"),
emr.Application(Name="Spark")
],
BootstrapActions=[
emr.BootstrapActionConfig(
Name='Dummy bootstrap action',
ScriptBootstrapAction=emr.ScriptBootstrapActionConfig(
Path='file:/usr/share/aws/emr/scripts/install-hue',
Args=["dummy", "parameter"]))
],
Configurations=[
emr.Configuration(Classification="core-site",
ConfigurationProperties={
'hadoop.security.groups.cache.secs':
'250'
})
],
Instances=emr.JobFlowInstancesConfig(
Ec2KeyName="KeyName",
Ec2SubnetId="SubnetId",
MasterInstanceGroup=emr.InstanceGroupConfigProperty(
InstanceCount="1",
InstanceType=M4_LARGE,
AutoScalingPolicy=emr.AutoScalingPolicy(
Constraints=emr.ScalingConstraints(MinCapacity="1",
MaxCapacity="3"),
Rules=self.generate_rules("MasterAutoScalingPolicy")),
),
CoreInstanceGroup=emr.InstanceGroupConfigProperty(
Name="Core Instance",
BidPrice=If(withSpotPrice, Ref(spot), Ref("AWS::NoValue")),
Market=If(withSpotPrice, "SPOT", "ON_DEMAND"),
InstanceCount="1",
InstanceType=M4_LARGE,
AutoScalingPolicy=emr.AutoScalingPolicy(
Constraints=emr.ScalingConstraints(MinCapacity="1",
MaxCapacity="3"),
Rules=self.generate_rules("CoreAutoScalingPolicy"),
)),
),
JobFlowRole="EMRJobFlowRole",
LogUri="s3://cluster-logs",
Name="EMR Cluster",
ReleaseLabel="emr-5.5.0",
SecurityConfiguration=Ref(cluster_security_configuration),
ServiceRole="EMRServiceRole",
AutoScalingRole="EMR_AutoScaling_DefaultRole",
VisibleToAllUsers="true",
Tags=Tags(Name="EMR Sample Cluster"))
cluster.to_dict()
autoscale_policy = emr.AutoScalingPolicy(
Constraints=emr.ScalingConstraints(MinCapacity=0, MaxCapacity=5),
Rules=[
emr.ScalingRule(
Name='ScaleUpContainerPending',
Description='Scale up on over-provisioned '
'containers',
Action=emr.ScalingAction(
SimpleScalingPolicyConfiguration=emr.
SimpleScalingPolicyConfiguration(
AdjustmentType=emr.CHANGE_IN_CAPACITY,
CoolDown=300,
ScalingAdjustment=1)),
Trigger=emr.ScalingTrigger(
CloudWatchAlarmDefinition=emr.
CloudWatchAlarmDefinition(
ComparisonOperator='GREATER_THAN',
MetricName='ContainerPendingRatio',
Period=300,
Threshold=0.75,
Dimensions=[
emr.MetricDimension(Key='JobFlowId',
Value='${emr.clusterId}')
]))),
emr.ScalingRule(
Name='ScaleUpMemory',
Description='Scale up on low memory',
Action=emr.ScalingAction(
SimpleScalingPolicyConfiguration=emr.
SimpleScalingPolicyConfiguration(
AdjustmentType='CHANGE_IN_CAPACITY',
CoolDown=300,
ScalingAdjustment=1)),
Trigger=emr.ScalingTrigger(
CloudWatchAlarmDefinition=emr.
CloudWatchAlarmDefinition(
ComparisonOperator='LESS_THAN',
MetricName='YARNMemoryAvailablePercentage',
Period=300,
Threshold=15,
Dimensions=[
emr.MetricDimension(Key='JobFlowId',
Value='${emr.clusterId}')
]))),
emr.ScalingRule(
Name='ScaleDownMemory',
Description='Scale down on high memory',
Action=emr.ScalingAction(
SimpleScalingPolicyConfiguration=emr.
SimpleScalingPolicyConfiguration(
AdjustmentType=emr.CHANGE_IN_CAPACITY,
CoolDown=300,
ScalingAdjustment=-1)),
Trigger=emr.ScalingTrigger(
CloudWatchAlarmDefinition=emr.
CloudWatchAlarmDefinition(
ComparisonOperator='GREATER_THAN',
MetricName='YARNMemoryAvailablePercentage',
Period=300,
Threshold=75,
Dimensions=[
emr.MetricDimension(Key='JobFlowId',
Value='${emr.clusterId}')
])))
])
emr.InstanceGroupConfig('TaskInstanceGroup',
AutoScalingPolicy=autoscale_policy,
InstanceCount=0,
InstanceType=M4_LARGE,
InstanceRole='TASK',
Market='ON_DEMAND',
Name='Task Instance',
JobFlowId=Ref(cluster))