def build_hive_step(parsed_step, region=None):
args = parsed_step.get('Args')
emrutils.check_required_field(
structure=constants.HIVE_STEP_CONFIG, name='Args', value=args)
emrutils.check_empty_string_list(name='Args', value=args)
name = _apply_default_value(
arg=parsed_step.get('Name'),
value=constants.DEFAULT_HIVE_STEP_NAME)
action_on_failure = \
_apply_default_value(
arg=parsed_step.get('ActionOnFailure'),
value=constants.DEFAULT_FAILURE_ACTION)
args_list = [
emrutils.build_s3_link(
relative_path=constants.HIVE_SCRIPT_PATH, region=region),
constants.RUN_HIVE_SCRIPT]
args_list.append(constants.HIVE_VERSIONS)
args_list.append(constants.LATEST)
args_list.append(constants.ARGS)
args_list += args
return emrutils.build_step(
jar=emrutils.get_script_runner(region),
args=args_list,
name=name,
action_on_failure=action_on_failure)
def build_pig_step(parsed_step, region=None):
version = parsed_step.get('Version')
args = parsed_step.get('Args')
emrutils.check_required_field(structure=constants.PIG_STEP_CONFIG,
name='Args',
value=args)
name = _apply_default_value(arg=parsed_step.get('Name'),
value=constants.DEFAULT_PIG_STEP_NAME)
action_on_failure = _apply_default_value(
arg=parsed_step.get('ActionOnFailure'),
value=constants.DEFAULT_FAILURE_ACTION)
args_list = [
emrutils.build_s3_link(relative_path=constants.PIG_SCRIPT_PATH,
region=region), constants.RUN_PIG_SCRIPT
]
args_list.append(constants.PIG_VERSIONS)
if version is not None:
args_list.append(version)
else:
args_list.append(constants.LATEST)
args_list.append(constants.ARGS)
args_list += args
return emrutils.build_step(jar=emrutils.get_script_runner(region),
args=args_list,
name=name,
action_on_failure=action_on_failure)
def build_streaming_step(parsed_step, release_label):
name = _apply_default_value(
arg=parsed_step.get('Name'),
value=constants.DEFAULT_STREAMING_STEP_NAME)
action_on_failure = _apply_default_value(
arg=parsed_step.get('ActionOnFailure'),
value=constants.DEFAULT_FAILURE_ACTION)
args = parsed_step.get('Args')
emrutils.check_required_field(
structure=constants.STREAMING_STEP_CONFIG,
name='Args',
value=args)
emrutils.check_empty_string_list(name='Args', value=args)
args_list = []
if release_label:
jar = constants.COMMAND_RUNNER
args_list.append(constants.HADOOP_STREAMING_COMMAND)
else:
jar = constants.HADOOP_STREAMING_PATH
args_list += args
return emrutils.build_step(
jar=jar,
args=args_list,
name=name,
action_on_failure=action_on_failure)
def _build_enable_debugging(self, parsed_globals):
return emrutils.build_step(
name=constants.DEBUGGING_NAME,
action_on_failure=constants.TERMINATE_CLUSTER,
jar=emrutils.get_script_runner(parsed_globals.region),
args=[emrutils.build_s3_link(relative_path=constants.DEBUGGING_PATH, region=parsed_globals.region)],
)
def build_streaming_step(parsed_step, release_label):
name = _apply_default_value(arg=parsed_step.get('Name'),
value=constants.DEFAULT_STREAMING_STEP_NAME)
action_on_failure = _apply_default_value(
arg=parsed_step.get('ActionOnFailure'),
value=constants.DEFAULT_FAILURE_ACTION)
args = parsed_step.get('Args')
emrutils.check_required_field(structure=constants.STREAMING_STEP_CONFIG,
name='Args',
value=args)
emrutils.check_empty_string_list(name='Args', value=args)
args_list = []
if release_label:
jar = constants.COMMAND_RUNNER
args_list.append(constants.HADOOP_STREAMING_COMMAND)
else:
jar = constants.HADOOP_STREAMING_PATH
args_list += args
return emrutils.build_step(jar=jar,
args=args_list,
name=name,
action_on_failure=action_on_failure)
def _build_enable_debugging(self, parsed_globals):
return emrutils.build_step(
name=constants.DEBUGGING_NAME,
action_on_failure=constants.TERMINATE_CLUSTER,
jar=emrutils.get_script_runner(),
args=[emrutils.build_s3_link(
relative_path=constants.DEBUGGING_PATH,
region=parsed_globals.region)])
def build_streaming_step(parsed_step):
name = _apply_default_value(arg=parsed_step.get("Name"), value=constants.DEFAULT_STREAMING_STEP_NAME)
action_on_failure = _apply_default_value(
arg=parsed_step.get("ActionOnFailure"), value=constants.DEFAULT_FAILURE_ACTION
)
args_list = parsed_step.get("Args")
emrutils.check_required_field(structure=constants.STREAMING_STEP_CONFIG, name="Args", value=args_list)
return emrutils.build_step(
jar=constants.STREAMING_JAR_PATH, args=args_list, name=name, action_on_failure=action_on_failure
)
def _build_pig_install_step(region,
action_on_failure=constants.TERMINATE_CLUSTER):
step_args = [
emrutils.build_s3_link(constants.PIG_SCRIPT_PATH, region),
constants.INSTALL_PIG_ARG, constants.BASE_PATH_ARG,
emrutils.build_s3_link(constants.PIG_BASE_PATH, region),
constants.PIG_VERSIONS, constants.LATEST
]
step = emrutils.build_step(name=constants.INSTALL_PIG_NAME,
action_on_failure=action_on_failure,
jar=emrutils.build_s3_link(
constants.SCRIPT_RUNNER_PATH, region),
args=step_args)
return step
def build_custom_jar_step(parsed_step):
name = _apply_default_value(arg=parsed_step.get("Name"), value=constants.DEFAULT_CUSTOM_JAR_STEP_NAME)
action_on_failure = _apply_default_value(
arg=parsed_step.get("ActionOnFailure"), value=constants.DEFAULT_FAILURE_ACTION
)
emrutils.check_required_field(structure=constants.CUSTOM_JAR_STEP_CONFIG, name="Jar", value=parsed_step.get("Jar"))
return emrutils.build_step(
jar=parsed_step.get("Jar"),
args=parsed_step.get("Args"),
name=name,
action_on_failure=action_on_failure,
main_class=parsed_step.get("MainClass"),
properties=emrutils.parse_key_value_string(parsed_step.get("Properties")),
)
def _build_install_hive_step(region, version,
action_on_failure=constants.TERMINATE_CLUSTER):
step_args = [
emrutils.build_s3_link(constants.HIVE_SCRIPT_PATH, region),
constants.INSTALL_HIVE_ARG,
constants.BASE_PATH_ARG,
emrutils.build_s3_link(constants.HIVE_BASE_PATH),
constants.HIVE_VERSIONS,
version]
step = emrutils.build_step(
name=constants.INSTALL_HIVE_NAME,
action_on_failure=action_on_failure,
jar=emrutils.build_s3_link(constants.SCRIPT_RUNNER_PATH, region),
args=step_args)
return step
def build_streaming_step(parsed_step):
name = _apply_default_value(arg=parsed_step.get('Name'),
value=constants.DEFAULT_STREAMING_STEP_NAME)
action_on_failure = _apply_default_value(
arg=parsed_step.get('ActionOnFailure'),
value=constants.DEFAULT_FAILURE_ACTION)
args_list = parsed_step.get('Args')
emrutils.check_required_field(structure=constants.STREAMING_STEP_CONFIG,
name='Args',
value=args_list)
return emrutils.build_step(jar=constants.STREAMING_JAR_PATH,
args=args_list,
name=name,
action_on_failure=action_on_failure)
def _run_main_command(self, parsed_args, parsed_globals):
steps = []
args = self._build_hbase_disable_backups_args(parsed_args)
step_config = emrutils.build_step(
constants.HBASE_JAR_PATH,
constants.HBASE_SCHEDULE_BACKUP_STEP_NAME,
constants.CANCEL_AND_WAIT, args)
steps.append(step_config)
parameters = {'JobFlowId': parsed_args.cluster_id, 'Steps': steps}
emrutils.call_and_display_response(self._session, 'AddJobFlowSteps',
parameters, parsed_globals)
return 0
def _build_enable_debugging(self, parsed_args, parsed_globals):
if parsed_args.release_label:
jar = constants.COMMAND_RUNNER
args = [constants.DEBUGGING_COMMAND]
else:
jar = emrutils.get_script_runner(self.region)
args = [emrutils.build_s3_link(
relative_path=constants.DEBUGGING_PATH,
region=self.region)]
return emrutils.build_step(
name=constants.DEBUGGING_NAME,
action_on_failure=constants.TERMINATE_CLUSTER,
jar=jar,
args=args)
def _build_enable_debugging(self, parsed_args, parsed_globals):
if parsed_args.release_label:
jar = constants.COMMAND_RUNNER
args = [constants.DEBUGGING_COMMAND]
else:
jar = emrutils.get_script_runner(self.region)
args = [emrutils.build_s3_link(
relative_path=constants.DEBUGGING_PATH,
region=self.region)]
return emrutils.build_step(
name=constants.DEBUGGING_NAME,
action_on_failure=constants.TERMINATE_CLUSTER,
jar=jar,
args=args)
def _build_pig_install_step(region,
action_on_failure=constants.TERMINATE_CLUSTER):
step_args = [
emrutils.build_s3_link(constants.PIG_SCRIPT_PATH, region),
constants.INSTALL_PIG_ARG,
constants.BASE_PATH_ARG,
emrutils.build_s3_link(constants.PIG_BASE_PATH, region),
constants.PIG_VERSIONS,
constants.LATEST]
step = emrutils.build_step(
name=constants.INSTALL_PIG_NAME,
action_on_failure=action_on_failure,
jar=emrutils.build_s3_link(constants.SCRIPT_RUNNER_PATH, region),
args=step_args)
return step
def _build_install_hive_step(region,
version,
action_on_failure=constants.TERMINATE_CLUSTER):
step_args = [
emrutils.build_s3_link(constants.HIVE_SCRIPT_PATH, region),
constants.INSTALL_HIVE_ARG, constants.BASE_PATH_ARG,
emrutils.build_s3_link(constants.HIVE_BASE_PATH),
constants.HIVE_VERSIONS, version
]
step = emrutils.build_step(name=constants.INSTALL_HIVE_NAME,
action_on_failure=action_on_failure,
jar=emrutils.build_s3_link(
constants.SCRIPT_RUNNER_PATH, region),
args=step_args)
return step
def _run_main_command(self, parsed_args, parsed_globals):
steps = []
args = hbaseutils.build_hbase_restore_from_backup_args(
parsed_args.dir, parsed_args.backup_version)
step_config = emrutils.build_step(
jar=constants.HBASE_JAR_PATH,
name=constants.HBASE_RESTORE_STEP_NAME,
action_on_failure=constants.CANCEL_AND_WAIT,
args=args)
steps.append(step_config)
parameters = {'JobFlowId': parsed_args.cluster_id, 'Steps': steps}
emrutils.call_and_display_response(self._session, 'AddJobFlowSteps',
parameters, parsed_globals)
return 0
def build_spark_step(parsed_step, region):
name = _apply_default_value(arg=parsed_step.get('Name'),
value=constants.DEFAULT_SPARK_STEP_NAME)
action_on_failure = _apply_default_value(
arg=parsed_step.get('ActionOnFailure'),
value=constants.DEFAULT_FAILURE_ACTION)
args_list = [constants.SPARK_SUBMIT_COMMAND]
args = parsed_step.get('Args')
emrutils.check_required_field(structure=constants.SPARK_STEP_CONFIG,
name='Args',
value=args)
args_list += args
return emrutils.build_step(jar=emrutils.get_script_runner(region),
args=args_list,
name=name,
action_on_failure=action_on_failure)
def _run_main_command(self, parsed_args, parsed_globals):
steps = []
args = self._build_hbase_backup_args(parsed_args)
step_config = emrutils.build_step(
jar=constants.HBASE_JAR_PATH,
name=constants.HBASE_BACKUP_STEP_NAME,
action_on_failure=constants.CANCEL_AND_WAIT,
args=args)
steps.append(step_config)
parameters = {'JobFlowId': parsed_args.cluster_id,
'Steps': steps}
emrutils.call_and_display_response(self._session, 'AddJobFlowSteps',
parameters, parsed_globals)
return 0
def build_custom_jar_step(parsed_step):
name = _apply_default_value(arg=parsed_step.get('Name'),
value=constants.DEFAULT_CUSTOM_JAR_STEP_NAME)
action_on_failure = _apply_default_value(
arg=parsed_step.get('ActionOnFailure'),
value=constants.DEFAULT_FAILURE_ACTION)
emrutils.check_required_field(structure=constants.CUSTOM_JAR_STEP_CONFIG,
name='Jar',
value=parsed_step.get('Jar'))
return emrutils.build_step(jar=parsed_step.get('Jar'),
args=parsed_step.get('Args'),
name=name,
action_on_failure=action_on_failure,
main_class=parsed_step.get('MainClass'),
properties=emrutils.parse_key_value_string(
parsed_step.get('Properties')))
def _build_install_hive_site_step(region, hive_site_path,
action_on_failure=constants.CANCEL_AND_WAIT):
step_args = [
emrutils.build_s3_link(constants.HIVE_SCRIPT_PATH, region),
constants.BASE_PATH_ARG,
emrutils.build_s3_link(constants.HIVE_BASE_PATH),
constants.INSTALL_HIVE_SITE_ARG,
hive_site_path,
constants.HIVE_VERSIONS,
constants.LATEST]
step = emrutils.build_step(
name=constants.INSTALL_HIVE_SITE_NAME,
action_on_failure=action_on_failure,
jar=emrutils.build_s3_link(constants.SCRIPT_RUNNER_PATH, region),
args=step_args)
return step
def build_spark_step(parsed_step, release_label, region=None):
name = _apply_default_value(arg=parsed_step.get('Name'),
value=constants.DEFAULT_SPARK_STEP_NAME)
action_on_failure = _apply_default_value(
arg=parsed_step.get('ActionOnFailure'),
value=constants.DEFAULT_FAILURE_ACTION)
args = parsed_step.get('Args')
emrutils.check_required_field(structure=constants.SPARK_STEP_CONFIG,
name='Args',
value=args)
return emrutils.build_step(jar=_get_runner_jar(release_label, region),
args=_build_spark_args(args, release_label,
region),
name=name,
action_on_failure=action_on_failure)
def build_impala_step(parsed_step, region):
name = _apply_default_value(arg=parsed_step.get("Name"), value=constants.DEFAULT_IMPALA_STEP_NAME)
action_on_failure = _apply_default_value(
arg=parsed_step.get("ActionOnFailure"), value=constants.DEFAULT_FAILURE_ACTION
)
args_list = [
emrutils.build_s3_link(relative_path=constants.IMPALA_INSTALL_PATH, region=region),
constants.RUN_IMPALA_SCRIPT,
]
args = parsed_step.get("Args")
emrutils.check_required_field(structure=constants.IMPALA_STEP_CONFIG, name="Args", value=args)
args_list += args
return emrutils.build_step(
jar=emrutils.get_script_runner(region), args=args_list, name=name, action_on_failure=action_on_failure
)
def build_spark_step(parsed_step, release_label, region=None):
name = _apply_default_value(
arg=parsed_step.get('Name'),
value=constants.DEFAULT_SPARK_STEP_NAME)
action_on_failure = _apply_default_value(
arg=parsed_step.get('ActionOnFailure'),
value=constants.DEFAULT_FAILURE_ACTION)
args = parsed_step.get('Args')
emrutils.check_required_field(
structure=constants.SPARK_STEP_CONFIG, name='Args', value=args)
return emrutils.build_step(
jar=_get_runner_jar(release_label, region),
args=_build_spark_args(args, release_label, region),
name=name,
action_on_failure=action_on_failure)
def _run_main_command(self, parsed_args, parsed_globals):
steps = []
self._check_type(parsed_args.type)
self._check_unit(parsed_args.unit)
args = self._build_hbase_schedule_backup_args(parsed_args)
step_config = emrutils.build_step(
jar=constants.HBASE_JAR_PATH,
name=constants.HBASE_SCHEDULE_BACKUP_STEP_NAME,
action_on_failure=constants.CANCEL_AND_WAIT,
args=args)
steps.append(step_config)
parameters = {'JobFlowId': parsed_args.cluster_id, 'Steps': steps}
emrutils.call_and_display_response(self._session, 'AddJobFlowSteps',
parameters, parsed_globals)
return 0
def build_spark_step(parsed_step, region):
name = _apply_default_value(
arg=parsed_step.get('Name'),
value=constants.DEFAULT_SPARK_STEP_NAME)
action_on_failure = _apply_default_value(
arg=parsed_step.get('ActionOnFailure'),
value=constants.DEFAULT_FAILURE_ACTION)
args_list=[constants.SPARK_SUBMIT_COMMAND]
args = parsed_step.get('Args')
emrutils.check_required_field(
structure=constants.SPARK_STEP_CONFIG, name='Args', value=args)
args_list += args
return emrutils.build_step(
jar=emrutils.get_script_runner(region),
args=args_list,
name=name,
action_on_failure=action_on_failure)
def build_streaming_step(parsed_step):
name = _apply_default_value(
arg=parsed_step.get('Name'),
value=constants.DEFAULT_STREAMING_STEP_NAME)
action_on_failure = _apply_default_value(
arg=parsed_step.get('ActionOnFailure'),
value=constants.DEFAULT_FAILURE_ACTION)
args = parsed_step.get('Args')
emrutils.check_required_field(
structure=constants.STREAMING_STEP_CONFIG,
name='Args',
value=args)
emrutils.check_empty_string_list(name='Args', value=args)
return emrutils.build_step(
jar=constants.STREAMING_JAR_PATH,
args=args,
name=name,
action_on_failure=action_on_failure)
def build_impala_step(parsed_step, region):
name = _apply_default_value(arg=parsed_step.get('Name'),
value=constants.DEFAULT_IMPALA_STEP_NAME)
action_on_failure = _apply_default_value(
arg=parsed_step.get('ActionOnFailure'),
value=constants.DEFAULT_FAILURE_ACTION)
args_list = [
emrutils.build_s3_link(relative_path=constants.IMPALA_INSTALL_PATH,
region=region), constants.RUN_IMPALA_SCRIPT
]
args = parsed_step.get('Args')
emrutils.check_required_field(structure=constants.IMPALA_STEP_CONFIG,
name='Args',
value=args)
args_list += args
return emrutils.build_step(jar=emrutils.get_script_runner(region),
args=args_list,
name=name,
action_on_failure=action_on_failure)
def build_pig_step(parsed_step, region=None):
args = parsed_step.get("Args")
emrutils.check_required_field(structure=constants.PIG_STEP_CONFIG, name="Args", value=args)
name = _apply_default_value(arg=parsed_step.get("Name"), value=constants.DEFAULT_PIG_STEP_NAME)
action_on_failure = _apply_default_value(
arg=parsed_step.get("ActionOnFailure"), value=constants.DEFAULT_FAILURE_ACTION
)
args_list = [
emrutils.build_s3_link(relative_path=constants.PIG_SCRIPT_PATH, region=region),
constants.RUN_PIG_SCRIPT,
]
args_list.append(constants.PIG_VERSIONS)
args_list.append(constants.LATEST)
args_list.append(constants.ARGS)
args_list += args
return emrutils.build_step(
jar=emrutils.get_script_runner(region), args=args_list, name=name, action_on_failure=action_on_failure
)
def build_impala_step(parsed_step, release_label, region=None):
if release_label:
raise exceptions.UnknownStepTypeError(step_type=constants.IMPALA)
name = _apply_default_value(
arg=parsed_step.get('Name'),
value=constants.DEFAULT_IMPALA_STEP_NAME)
action_on_failure = _apply_default_value(
arg=parsed_step.get('ActionOnFailure'),
value=constants.DEFAULT_FAILURE_ACTION)
args_list = [
emrutils.build_s3_link(
relative_path=constants.IMPALA_INSTALL_PATH, region=region),
constants.RUN_IMPALA_SCRIPT]
args = parsed_step.get('Args')
emrutils.check_required_field(
structure=constants.IMPALA_STEP_CONFIG, name='Args', value=args)
args_list += args
return emrutils.build_step(
jar=emrutils.get_script_runner(region),
args=args_list,
name=name,
action_on_failure=action_on_failure)
def _run_main_command(self, parsed_args, parsed_globals):
params = {}
bootstrap_actions = []
params['Name'] = parsed_args.name
service_role_validation_message = (
" Either choose --use-default-roles or use both --service-role "
"<roleName> and --ec2-attributes InstanceProfile=<profileName>.")
if parsed_args.use_default_roles is True and \
parsed_args.service_role is not None:
raise exceptions.MutualExclusiveOptionError(
option1="--use-default-roles",
option2="--service-role",
message=service_role_validation_message)
if parsed_args.use_default_roles is True and \
parsed_args.ec2_attributes is not None and \
'InstanceProfile' in parsed_args.ec2_attributes:
raise exceptions.MutualExclusiveOptionError(
option1="--use-default-roles",
option2="--ec2-attributes InstanceProfile",
message=service_role_validation_message)
instances_config = {}
instances_config['InstanceGroups'] = \
instancegroupsutils.validate_and_build_instance_groups(
instance_groups=parsed_args.instance_groups,
instance_type=parsed_args.instance_type,
instance_count=parsed_args.instance_count)
is_valid_ami_version = re.match('\d?\..*', parsed_args.ami_version)
if is_valid_ami_version is None:
raise exceptions.InvalidAmiVersionError(
ami_version=parsed_args.ami_version)
params['AmiVersion'] = parsed_args.ami_version
emrutils.apply_dict(
params, 'AdditionalInfo', parsed_args.additional_info)
emrutils.apply_dict(params, 'LogUri', parsed_args.log_uri)
if parsed_args.use_default_roles is True:
parsed_args.service_role = EMR_ROLE_NAME
if parsed_args.ec2_attributes is None:
parsed_args.ec2_attributes = {}
parsed_args.ec2_attributes['InstanceProfile'] = EC2_ROLE_NAME
emrutils.apply_dict(params, 'ServiceRole', parsed_args.service_role)
if (
parsed_args.no_auto_terminate is False and
parsed_args.auto_terminate is False):
parsed_args.no_auto_terminate = True
instances_config['KeepJobFlowAliveWhenNoSteps'] = \
emrutils.apply_boolean_options(
parsed_args.no_auto_terminate,
'--no-auto-terminate',
parsed_args.auto_terminate,
'--auto-terminate')
instances_config['TerminationProtected'] = \
emrutils.apply_boolean_options(
parsed_args.termination_protected,
'--termination-protected',
parsed_args.no_termination_protected,
'--no-termination-protected')
if (
parsed_args.visible_to_all_users is False and
parsed_args.no_visible_to_all_users is False):
parsed_args.visible_to_all_users = True
params['VisibleToAllUsers'] = \
emrutils.apply_boolean_options(
parsed_args.visible_to_all_users,
'--visible-to-all-users',
parsed_args.no_visible_to_all_users,
'--no-visible-to-all-users')
params['Tags'] = emrutils.parse_tags(parsed_args.tags)
params['Instances'] = instances_config
if parsed_args.ec2_attributes is not None:
self._build_ec2_attributes(
cluster=params, parsed_attrs=parsed_args.ec2_attributes)
debugging_enabled = emrutils.apply_boolean_options(
parsed_args.enable_debugging,
'--enable-debugging',
parsed_args.no_enable_debugging,
'--no-enable-debugging')
if parsed_args.log_uri is None and debugging_enabled is True:
raise exceptions.LogUriError
if debugging_enabled is True:
self._update_cluster_dict(
cluster=params,
key='Steps',
value=[self._build_enable_debugging(parsed_globals)])
if parsed_args.applications is not None:
app_list, ba_list, step_list = applicationutils.build_applications(
region=self.region,
parsed_applications=parsed_args.applications,
ami_version=params['AmiVersion'])
self._update_cluster_dict(
params, 'NewSupportedProducts', app_list)
self._update_cluster_dict(
params, 'BootstrapActions', ba_list)
self._update_cluster_dict(
params, 'Steps', step_list)
hbase_restore_config = parsed_args.restore_from_hbase_backup
if hbase_restore_config is not None:
args = hbaseutils.build_hbase_restore_from_backup_args(
dir=hbase_restore_config.get('Dir'),
backup_version=hbase_restore_config.get('BackupVersion'))
step_config = emrutils.build_step(
jar=constants.HBASE_JAR_PATH,
name=constants.HBASE_RESTORE_STEP_NAME,
action_on_failure=constants.CANCEL_AND_WAIT,
args=args)
self._update_cluster_dict(
params, 'Steps', [step_config])
if parsed_args.bootstrap_actions is not None:
self._build_bootstrap_actions(
cluster=params,
parsed_boostrap_actions=parsed_args.bootstrap_actions)
if parsed_args.emrfs is not None:
emr_fs_ba_config_list = emrfsutils.build_bootstrap_action_configs(
self.region, parsed_args.emrfs)
self._update_cluster_dict(
cluster=params, key='BootstrapActions',
value=emr_fs_ba_config_list)
if parsed_args.steps is not None:
steps_list = steputils.build_step_config_list(
parsed_step_list=parsed_args.steps,
region=self.region)
self._update_cluster_dict(
cluster=params, key='Steps', value=steps_list)
self._validate_required_applications(parsed_args)
run_job_flow_response = emrutils.call(
self._session, 'run_job_flow', params, self.region,
parsed_globals.endpoint_url, parsed_globals.verify_ssl)
constructed_result = self._construct_result(run_job_flow_response)
emrutils.display_response(self._session, 'run_job_flow',
constructed_result, parsed_globals)
return 0
def _run_main_command(self, parsed_args, parsed_globals):
params = {}
params['Name'] = parsed_args.name
self._validate_release_label_ami_version(parsed_args)
service_role_validation_message = (
" Either choose --use-default-roles or use both --service-role "
"<roleName> and --ec2-attributes InstanceProfile=<profileName>.")
if parsed_args.use_default_roles is True and \
parsed_args.service_role is not None:
raise exceptions.MutualExclusiveOptionError(
option1="--use-default-roles",
option2="--service-role",
message=service_role_validation_message)
if parsed_args.use_default_roles is True and \
parsed_args.ec2_attributes is not None and \
'InstanceProfile' in parsed_args.ec2_attributes:
raise exceptions.MutualExclusiveOptionError(
option1="--use-default-roles",
option2="--ec2-attributes InstanceProfile",
message=service_role_validation_message)
if parsed_args.instance_groups is not None and \
parsed_args.instance_fleets is not None:
raise exceptions.MutualExclusiveOptionError(
option1="--instance-groups",
option2="--instance-fleets")
instances_config = {}
if parsed_args.instance_fleets is not None:
instances_config['InstanceFleets'] = \
instancefleetsutils.validate_and_build_instance_fleets(
parsed_args.instance_fleets)
else:
instances_config['InstanceGroups'] = \
instancegroupsutils.validate_and_build_instance_groups(
instance_groups=parsed_args.instance_groups,
instance_type=parsed_args.instance_type,
instance_count=parsed_args.instance_count)
if parsed_args.release_label is not None:
params["ReleaseLabel"] = parsed_args.release_label
if parsed_args.configurations is not None:
try:
params["Configurations"] = json.loads(
parsed_args.configurations)
except ValueError:
raise ValueError('aws: error: invalid json argument for '
'option --configurations')
if (parsed_args.release_label is None and
parsed_args.ami_version is not None):
is_valid_ami_version = re.match('\d?\..*', parsed_args.ami_version)
if is_valid_ami_version is None:
raise exceptions.InvalidAmiVersionError(
ami_version=parsed_args.ami_version)
params['AmiVersion'] = parsed_args.ami_version
emrutils.apply_dict(
params, 'AdditionalInfo', parsed_args.additional_info)
emrutils.apply_dict(params, 'LogUri', parsed_args.log_uri)
if parsed_args.log_encryption_kms_key_id is not None:
emrutils.apply_dict(params, 'LogEncryptionKmsKeyId',
parsed_args.log_encryption_kms_key_id)
if parsed_args.use_default_roles is True:
parsed_args.service_role = EMR_ROLE_NAME
if parsed_args.ec2_attributes is None:
parsed_args.ec2_attributes = {}
parsed_args.ec2_attributes['InstanceProfile'] = EC2_ROLE_NAME
emrutils.apply_dict(params, 'ServiceRole', parsed_args.service_role)
if parsed_args.instance_groups is not None:
for instance_group in instances_config['InstanceGroups']:
if 'AutoScalingPolicy' in instance_group.keys():
if parsed_args.auto_scaling_role is None:
raise exceptions.MissingAutoScalingRoleError()
emrutils.apply_dict(params, 'AutoScalingRole', parsed_args.auto_scaling_role)
if parsed_args.scale_down_behavior is not None:
emrutils.apply_dict(params, 'ScaleDownBehavior', parsed_args.scale_down_behavior)
if (
parsed_args.no_auto_terminate is False and
parsed_args.auto_terminate is False):
parsed_args.no_auto_terminate = True
instances_config['KeepJobFlowAliveWhenNoSteps'] = \
emrutils.apply_boolean_options(
parsed_args.no_auto_terminate,
'--no-auto-terminate',
parsed_args.auto_terminate,
'--auto-terminate')
instances_config['TerminationProtected'] = \
emrutils.apply_boolean_options(
parsed_args.termination_protected,
'--termination-protected',
parsed_args.no_termination_protected,
'--no-termination-protected')
if (parsed_args.visible_to_all_users is False and
parsed_args.no_visible_to_all_users is False):
parsed_args.visible_to_all_users = True
params['VisibleToAllUsers'] = \
emrutils.apply_boolean_options(
parsed_args.visible_to_all_users,
'--visible-to-all-users',
parsed_args.no_visible_to_all_users,
'--no-visible-to-all-users')
params['Tags'] = emrutils.parse_tags(parsed_args.tags)
params['Instances'] = instances_config
if parsed_args.ec2_attributes is not None:
self._build_ec2_attributes(
cluster=params, parsed_attrs=parsed_args.ec2_attributes)
debugging_enabled = emrutils.apply_boolean_options(
parsed_args.enable_debugging,
'--enable-debugging',
parsed_args.no_enable_debugging,
'--no-enable-debugging')
if parsed_args.log_uri is None and debugging_enabled is True:
raise exceptions.LogUriError
if debugging_enabled is True:
self._update_cluster_dict(
cluster=params,
key='Steps',
value=[
self._build_enable_debugging(parsed_args, parsed_globals)])
if parsed_args.applications is not None:
if parsed_args.release_label is None:
app_list, ba_list, step_list = \
applicationutils.build_applications(
region=self.region,
parsed_applications=parsed_args.applications,
ami_version=params['AmiVersion'])
self._update_cluster_dict(
params, 'NewSupportedProducts', app_list)
self._update_cluster_dict(
params, 'BootstrapActions', ba_list)
self._update_cluster_dict(
params, 'Steps', step_list)
else:
params["Applications"] = []
for application in parsed_args.applications:
params["Applications"].append(application)
hbase_restore_config = parsed_args.restore_from_hbase_backup
if hbase_restore_config is not None:
args = hbaseutils.build_hbase_restore_from_backup_args(
dir=hbase_restore_config.get('Dir'),
backup_version=hbase_restore_config.get('BackupVersion'))
step_config = emrutils.build_step(
jar=constants.HBASE_JAR_PATH,
name=constants.HBASE_RESTORE_STEP_NAME,
action_on_failure=constants.CANCEL_AND_WAIT,
args=args)
self._update_cluster_dict(
params, 'Steps', [step_config])
if parsed_args.bootstrap_actions is not None:
self._build_bootstrap_actions(
cluster=params,
parsed_boostrap_actions=parsed_args.bootstrap_actions)
if parsed_args.emrfs is not None:
self._handle_emrfs_parameters(
cluster=params,
emrfs_args=parsed_args.emrfs,
release_label=parsed_args.release_label)
if parsed_args.steps is not None:
steps_list = steputils.build_step_config_list(
parsed_step_list=parsed_args.steps,
region=self.region,
release_label=parsed_args.release_label)
self._update_cluster_dict(
cluster=params, key='Steps', value=steps_list)
if parsed_args.security_configuration is not None:
emrutils.apply_dict(
params, 'SecurityConfiguration', parsed_args.security_configuration)
if parsed_args.custom_ami_id is not None:
emrutils.apply_dict(
params, 'CustomAmiId', parsed_args.custom_ami_id
)
if parsed_args.ebs_root_volume_size is not None:
emrutils.apply_dict(
params, 'EbsRootVolumeSize', int(parsed_args.ebs_root_volume_size)
)
if parsed_args.repo_upgrade_on_boot is not None:
emrutils.apply_dict(
params, 'RepoUpgradeOnBoot', parsed_args.repo_upgrade_on_boot
)
if parsed_args.kerberos_attributes is not None:
emrutils.apply_dict(
params, 'KerberosAttributes', parsed_args.kerberos_attributes)
if parsed_args.step_concurrency_level is not None:
params['StepConcurrencyLevel'] = parsed_args.step_concurrency_level
if parsed_args.managed_scaling_policy is not None:
emrutils.apply_dict(
params, 'ManagedScalingPolicy', parsed_args.managed_scaling_policy)
if parsed_args.placement_group_configs is not None:
emrutils.apply_dict(
params, 'PlacementGroupConfigs',
parsed_args.placement_group_configs)
self._validate_required_applications(parsed_args)
run_job_flow_response = emrutils.call(
self._session, 'run_job_flow', params, self.region,
parsed_globals.endpoint_url, parsed_globals.verify_ssl)
constructed_result = self._construct_result(run_job_flow_response)
emrutils.display_response(self._session, 'run_job_flow',
constructed_result, parsed_globals)
return 0
def _run_main(self, parsed_args, parsed_globals):
emr = self._session.get_service('emr')
params = {}
bootstrap_actions = []
params['Name'] = parsed_args.name
instances_config = {}
instances_config['InstanceGroups'] = \
instancegroupsutils.validate_and_build_instance_groups(
instance_groups=parsed_args.instance_groups,
instance_type=parsed_args.instance_type,
instance_count=parsed_args.instance_count)
is_valid_ami_version = re.match('\d?\..*', parsed_args.ami_version)
if is_valid_ami_version is None:
raise exceptions.InvalidAmiVersionError(
ami_version=parsed_args.ami_version)
params['AmiVersion'] = parsed_args.ami_version
emrutils.apply_dict(params, 'AdditionalInfo',
parsed_args.additional_info)
emrutils.apply_dict(params, 'LogUri', parsed_args.log_uri)
if parsed_args.use_default_roles is True:
parsed_args.service_role = EMR_ROLE_NAME
if parsed_args.ec2_attributes is None:
parsed_args.ec2_attributes = {}
parsed_args.ec2_attributes['InstanceProfile'] = EC2_ROLE_NAME
emrutils.apply_dict(params, 'ServiceRole', parsed_args.service_role)
if (parsed_args.no_auto_terminate is False
and parsed_args.auto_terminate is False):
parsed_args.no_auto_terminate = True
instances_config['KeepJobFlowAliveWhenNoSteps'] = \
emrutils.apply_boolean_options(
parsed_args.no_auto_terminate,
'--no-auto-terminate',
parsed_args.auto_terminate,
'--auto-terminate')
instances_config['TerminationProtected'] = \
emrutils.apply_boolean_options(
parsed_args.termination_protected,
'--termination-protected',
parsed_args.no_termination_protected,
'--no-termination-protected')
if (parsed_args.visible_to_all_users is False
and parsed_args.no_visible_to_all_users is False):
parsed_args.visible_to_all_users = True
params['VisibleToAllUsers'] = \
emrutils.apply_boolean_options(
parsed_args.visible_to_all_users,
'--visible-to-all-users',
parsed_args.no_visible_to_all_users,
'--no-visible-to-all-users')
params['Tags'] = emrutils.parse_tags(parsed_args.tags)
params['Instances'] = instances_config
if parsed_args.ec2_attributes is not None:
self._build_ec2_attributes(cluster=params,
parsed_attrs=parsed_args.ec2_attributes)
debugging_enabled = emrutils.apply_boolean_options(
parsed_args.enable_debugging, '--enable-debugging',
parsed_args.no_enable_debugging, '--no-enable-debugging')
if parsed_args.log_uri is None and debugging_enabled is True:
raise exceptions.LogUriError
if debugging_enabled is True:
self._update_cluster_dict(
cluster=params,
key='Steps',
value=[self._build_enable_debugging(parsed_globals)])
if parsed_args.applications is not None:
app_list, ba_list, step_list = applicationutils.build_applications(
session=self._session,
parsed_applications=parsed_args.applications,
parsed_globals=parsed_globals,
ami_version=params['AmiVersion'])
self._update_cluster_dict(params, 'NewSupportedProducts', app_list)
self._update_cluster_dict(params, 'BootstrapActions', ba_list)
self._update_cluster_dict(params, 'Steps', step_list)
hbase_restore_config = parsed_args.restore_from_hbase_backup
if hbase_restore_config is not None:
args = hbaseutils.build_hbase_restore_from_backup_args(
dir=hbase_restore_config.get('Dir'),
backup_version=hbase_restore_config.get('BackupVersion'))
step_config = emrutils.build_step(
jar=constants.HBASE_JAR_PATH,
name=constants.HBASE_RESTORE_STEP_NAME,
action_on_failure=constants.CANCEL_AND_WAIT,
args=args)
self._update_cluster_dict(params, 'Steps', [step_config])
if parsed_args.bootstrap_actions is not None:
self._build_bootstrap_actions(
cluster=params,
parsed_boostrap_actions=parsed_args.bootstrap_actions)
if parsed_args.emrfs is not None:
emr_fs_ba_args = self._build_emr_fs_args(parsed_args.emrfs)
emr_fs_ba_config = \
emrutils.build_bootstrap_action(
path=emrutils.build_s3_link(
relative_path=constants.CONFIG_HADOOP_PATH,
region=parsed_globals.region),
name=constants.EMR_FS_BA_NAME,
args=emr_fs_ba_args)
self._update_cluster_dict(cluster=params,
key='BootstrapActions',
value=[emr_fs_ba_config])
if parsed_args.steps is not None:
steps_list = steputils.build_step_config_list(
parsed_step_list=parsed_args.steps,
region=parsed_globals.region)
self._update_cluster_dict(cluster=params,
key='Steps',
value=steps_list)
self._validate_required_applications(parsed_args)
run_job_flow = emr.get_operation('RunJobFlow')
run_job_flow_response = emrutils.call(self._session, run_job_flow,
params, parsed_globals.region,
parsed_globals.endpoint_url,
parsed_globals.verify_ssl)
constructed_result = self._construct_result(run_job_flow_response[1])
emrutils.display_response(self._session, run_job_flow,
constructed_result, parsed_globals)
return 0
def _run_main_command(self, parsed_args, parsed_globals):
params = {}
params['Name'] = parsed_args.name
self._validate_release_label_ami_version(parsed_args)
service_role_validation_message = (
" Either choose --use-default-roles or use both --service-role "
"<roleName> and --ec2-attributes InstanceProfile=<profileName>.")
if parsed_args.use_default_roles is True and \
parsed_args.service_role is not None:
raise exceptions.MutualExclusiveOptionError(
option1="--use-default-roles",
option2="--service-role",
message=service_role_validation_message)
if parsed_args.use_default_roles is True and \
parsed_args.ec2_attributes is not None and \
'InstanceProfile' in parsed_args.ec2_attributes:
raise exceptions.MutualExclusiveOptionError(
option1="--use-default-roles",
option2="--ec2-attributes InstanceProfile",
message=service_role_validation_message)
if parsed_args.instance_groups is not None and \
parsed_args.instance_fleets is not None:
raise exceptions.MutualExclusiveOptionError(
option1="--instance-groups",
option2="--instance-fleets")
instances_config = {}
if parsed_args.instance_fleets is not None:
instances_config['InstanceFleets'] = \
instancefleetsutils.validate_and_build_instance_fleets(
parsed_args.instance_fleets)
else:
instances_config['InstanceGroups'] = \
instancegroupsutils.validate_and_build_instance_groups(
instance_groups=parsed_args.instance_groups,
instance_type=parsed_args.instance_type,
instance_count=parsed_args.instance_count)
if parsed_args.release_label is not None:
params["ReleaseLabel"] = parsed_args.release_label
if parsed_args.configurations is not None:
try:
params["Configurations"] = json.loads(
parsed_args.configurations)
except ValueError:
raise ValueError('aws: error: invalid json argument for '
'option --configurations')
if (parsed_args.release_label is None and
parsed_args.ami_version is not None):
is_valid_ami_version = re.match('\d?\..*', parsed_args.ami_version)
if is_valid_ami_version is None:
raise exceptions.InvalidAmiVersionError(
ami_version=parsed_args.ami_version)
params['AmiVersion'] = parsed_args.ami_version
emrutils.apply_dict(
params, 'AdditionalInfo', parsed_args.additional_info)
emrutils.apply_dict(params, 'LogUri', parsed_args.log_uri)
if parsed_args.use_default_roles is True:
parsed_args.service_role = EMR_ROLE_NAME
if parsed_args.ec2_attributes is None:
parsed_args.ec2_attributes = {}
parsed_args.ec2_attributes['InstanceProfile'] = EC2_ROLE_NAME
emrutils.apply_dict(params, 'ServiceRole', parsed_args.service_role)
if parsed_args.instance_groups is not None:
for instance_group in instances_config['InstanceGroups']:
if 'AutoScalingPolicy' in instance_group.keys():
if parsed_args.auto_scaling_role is None:
raise exceptions.MissingAutoScalingRoleError()
emrutils.apply_dict(params, 'AutoScalingRole', parsed_args.auto_scaling_role)
if parsed_args.scale_down_behavior is not None:
emrutils.apply_dict(params, 'ScaleDownBehavior', parsed_args.scale_down_behavior)
if (
parsed_args.no_auto_terminate is False and
parsed_args.auto_terminate is False):
parsed_args.no_auto_terminate = True
instances_config['KeepJobFlowAliveWhenNoSteps'] = \
emrutils.apply_boolean_options(
parsed_args.no_auto_terminate,
'--no-auto-terminate',
parsed_args.auto_terminate,
'--auto-terminate')
instances_config['TerminationProtected'] = \
emrutils.apply_boolean_options(
parsed_args.termination_protected,
'--termination-protected',
parsed_args.no_termination_protected,
'--no-termination-protected')
if (parsed_args.visible_to_all_users is False and
parsed_args.no_visible_to_all_users is False):
parsed_args.visible_to_all_users = True
params['VisibleToAllUsers'] = \
emrutils.apply_boolean_options(
parsed_args.visible_to_all_users,
'--visible-to-all-users',
parsed_args.no_visible_to_all_users,
'--no-visible-to-all-users')
params['Tags'] = emrutils.parse_tags(parsed_args.tags)
params['Instances'] = instances_config
if parsed_args.ec2_attributes is not None:
self._build_ec2_attributes(
cluster=params, parsed_attrs=parsed_args.ec2_attributes)
debugging_enabled = emrutils.apply_boolean_options(
parsed_args.enable_debugging,
'--enable-debugging',
parsed_args.no_enable_debugging,
'--no-enable-debugging')
if parsed_args.log_uri is None and debugging_enabled is True:
raise exceptions.LogUriError
if debugging_enabled is True:
self._update_cluster_dict(
cluster=params,
key='Steps',
value=[
self._build_enable_debugging(parsed_args, parsed_globals)])
if parsed_args.applications is not None:
if parsed_args.release_label is None:
app_list, ba_list, step_list = \
applicationutils.build_applications(
region=self.region,
parsed_applications=parsed_args.applications,
ami_version=params['AmiVersion'])
self._update_cluster_dict(
params, 'NewSupportedProducts', app_list)
self._update_cluster_dict(
params, 'BootstrapActions', ba_list)
self._update_cluster_dict(
params, 'Steps', step_list)
else:
params["Applications"] = []
for application in parsed_args.applications:
params["Applications"].append(application)
hbase_restore_config = parsed_args.restore_from_hbase_backup
if hbase_restore_config is not None:
args = hbaseutils.build_hbase_restore_from_backup_args(
dir=hbase_restore_config.get('Dir'),
backup_version=hbase_restore_config.get('BackupVersion'))
step_config = emrutils.build_step(
jar=constants.HBASE_JAR_PATH,
name=constants.HBASE_RESTORE_STEP_NAME,
action_on_failure=constants.CANCEL_AND_WAIT,
args=args)
self._update_cluster_dict(
params, 'Steps', [step_config])
if parsed_args.bootstrap_actions is not None:
self._build_bootstrap_actions(
cluster=params,
parsed_boostrap_actions=parsed_args.bootstrap_actions)
if parsed_args.emrfs is not None:
self._handle_emrfs_parameters(
cluster=params,
emrfs_args=parsed_args.emrfs,
release_label=parsed_args.release_label)
if parsed_args.steps is not None:
steps_list = steputils.build_step_config_list(
parsed_step_list=parsed_args.steps,
region=self.region,
release_label=parsed_args.release_label)
self._update_cluster_dict(
cluster=params, key='Steps', value=steps_list)
if parsed_args.security_configuration is not None:
emrutils.apply_dict(
params, 'SecurityConfiguration', parsed_args.security_configuration)
if parsed_args.custom_ami_id is not None:
emrutils.apply_dict(
params, 'CustomAmiId', parsed_args.custom_ami_id
)
if parsed_args.ebs_root_volume_size is not None:
emrutils.apply_dict(
params, 'EbsRootVolumeSize', int(parsed_args.ebs_root_volume_size)
)
if parsed_args.repo_upgrade_on_boot is not None:
emrutils.apply_dict(
params, 'RepoUpgradeOnBoot', parsed_args.repo_upgrade_on_boot
)
self._validate_required_applications(parsed_args)
run_job_flow_response = emrutils.call(
self._session, 'run_job_flow', params, self.region,
parsed_globals.endpoint_url, parsed_globals.verify_ssl)
constructed_result = self._construct_result(run_job_flow_response)
emrutils.display_response(self._session, 'run_job_flow',
constructed_result, parsed_globals)
return 0
def _build_hbase_install_step(jar):
return emrutils.build_step(
jar=jar,
name=constants.START_HBASE_NAME,
action_on_failure=constants.TERMINATE_CLUSTER,
args=constants.HBASE_INSTALL_ARG)
def _run_main(self, parsed_args, parsed_globals):
emr = self._session.get_service("emr")
params = {}
bootstrap_actions = []
params["Name"] = parsed_args.name
service_role_validation_message = (
" Either choose --use-default-roles or use both --service-role "
"<roleName> and --ec2-attributes InstanceProfile=<profileName>."
)
if parsed_args.use_default_roles is True and parsed_args.service_role is not None:
raise exceptions.MutualExclusiveOptionError(
option1="--use-default-roles", option2="--service-role", message=service_role_validation_message
)
if (
parsed_args.use_default_roles is True
and parsed_args.ec2_attributes is not None
and "InstanceProfile" in parsed_args.ec2_attributes
):
raise exceptions.MutualExclusiveOptionError(
option1="--use-default-roles",
option2="--ec2-attributes InstanceProfile",
message=service_role_validation_message,
)
instances_config = {}
instances_config["InstanceGroups"] = instancegroupsutils.validate_and_build_instance_groups(
instance_groups=parsed_args.instance_groups,
instance_type=parsed_args.instance_type,
instance_count=parsed_args.instance_count,
)
is_valid_ami_version = re.match("\d?\..*", parsed_args.ami_version)
if is_valid_ami_version is None:
raise exceptions.InvalidAmiVersionError(ami_version=parsed_args.ami_version)
params["AmiVersion"] = parsed_args.ami_version
emrutils.apply_dict(params, "AdditionalInfo", parsed_args.additional_info)
emrutils.apply_dict(params, "LogUri", parsed_args.log_uri)
if parsed_args.use_default_roles is True:
parsed_args.service_role = EMR_ROLE_NAME
if parsed_args.ec2_attributes is None:
parsed_args.ec2_attributes = {}
parsed_args.ec2_attributes["InstanceProfile"] = EC2_ROLE_NAME
emrutils.apply_dict(params, "ServiceRole", parsed_args.service_role)
if parsed_args.no_auto_terminate is False and parsed_args.auto_terminate is False:
parsed_args.no_auto_terminate = True
instances_config["KeepJobFlowAliveWhenNoSteps"] = emrutils.apply_boolean_options(
parsed_args.no_auto_terminate, "--no-auto-terminate", parsed_args.auto_terminate, "--auto-terminate"
)
instances_config["TerminationProtected"] = emrutils.apply_boolean_options(
parsed_args.termination_protected,
"--termination-protected",
parsed_args.no_termination_protected,
"--no-termination-protected",
)
if parsed_args.visible_to_all_users is False and parsed_args.no_visible_to_all_users is False:
parsed_args.visible_to_all_users = True
params["VisibleToAllUsers"] = emrutils.apply_boolean_options(
parsed_args.visible_to_all_users,
"--visible-to-all-users",
parsed_args.no_visible_to_all_users,
"--no-visible-to-all-users",
)
params["Tags"] = emrutils.parse_tags(parsed_args.tags)
params["Instances"] = instances_config
if parsed_args.ec2_attributes is not None:
self._build_ec2_attributes(cluster=params, parsed_attrs=parsed_args.ec2_attributes)
debugging_enabled = emrutils.apply_boolean_options(
parsed_args.enable_debugging, "--enable-debugging", parsed_args.no_enable_debugging, "--no-enable-debugging"
)
if parsed_args.log_uri is None and debugging_enabled is True:
raise exceptions.LogUriError
if debugging_enabled is True:
self._update_cluster_dict(cluster=params, key="Steps", value=[self._build_enable_debugging(parsed_globals)])
if parsed_args.applications is not None:
app_list, ba_list, step_list = applicationutils.build_applications(
session=self._session,
parsed_applications=parsed_args.applications,
parsed_globals=parsed_globals,
ami_version=params["AmiVersion"],
)
self._update_cluster_dict(params, "NewSupportedProducts", app_list)
self._update_cluster_dict(params, "BootstrapActions", ba_list)
self._update_cluster_dict(params, "Steps", step_list)
hbase_restore_config = parsed_args.restore_from_hbase_backup
if hbase_restore_config is not None:
args = hbaseutils.build_hbase_restore_from_backup_args(
dir=hbase_restore_config.get("Dir"), backup_version=hbase_restore_config.get("BackupVersion")
)
step_config = emrutils.build_step(
jar=constants.HBASE_JAR_PATH,
name=constants.HBASE_RESTORE_STEP_NAME,
action_on_failure=constants.CANCEL_AND_WAIT,
args=args,
)
self._update_cluster_dict(params, "Steps", [step_config])
if parsed_args.bootstrap_actions is not None:
self._build_bootstrap_actions(cluster=params, parsed_boostrap_actions=parsed_args.bootstrap_actions)
if parsed_args.emrfs is not None:
emr_fs_ba_args = self._build_emr_fs_args(parsed_args.emrfs)
emr_fs_ba_config = emrutils.build_bootstrap_action(
path=emrutils.build_s3_link(relative_path=constants.CONFIG_HADOOP_PATH, region=parsed_globals.region),
name=constants.EMR_FS_BA_NAME,
args=emr_fs_ba_args,
)
self._update_cluster_dict(cluster=params, key="BootstrapActions", value=[emr_fs_ba_config])
if parsed_args.steps is not None:
steps_list = steputils.build_step_config_list(
parsed_step_list=parsed_args.steps, region=parsed_globals.region
)
self._update_cluster_dict(cluster=params, key="Steps", value=steps_list)
self._validate_required_applications(parsed_args)
run_job_flow = emr.get_operation("RunJobFlow")
run_job_flow_response = emrutils.call(
self._session,
run_job_flow,
params,
parsed_globals.region,
parsed_globals.endpoint_url,
parsed_globals.verify_ssl,
)
constructed_result = self._construct_result(run_job_flow_response[1])
emrutils.display_response(self._session, run_job_flow, constructed_result, parsed_globals)
return 0
def _run_main(self, parsed_args, parsed_globals):
emr = self._session.get_service('emr')
params = {}
bootstrap_actions = []
params['Name'] = parsed_args.name
is_valid_ami = re.match('\d?\..*', parsed_args.ami_version)
if is_valid_ami is None:
raise exceptions.\
InvalidAmiVersionError(ami_version=parsed_args.ami_version)
params['AmiVersion'] = parsed_args.ami_version
emrutils.apply_dict(
params, 'AdditionalInfo', parsed_args.additional_info)
emrutils.apply_dict(params, 'LogUri', parsed_args.log_uri)
if parsed_args.use_default_roles is True:
parsed_args.service_role = EMR_ROLE_NAME
if parsed_args.ec2_attributes is None:
parsed_args.ec2_attributes = {}
parsed_args.ec2_attributes['InstanceProfile'] = EC2_ROLE_NAME
emrutils.apply_dict(params, 'ServiceRole', parsed_args.service_role)
instances_config = {}
instances_config['InstanceGroups'] = \
instancegroupsutils.build_instance_groups(
parsed_args.instance_groups)
if (
parsed_args.no_auto_terminate is False and
parsed_args.auto_terminate is False):
raise exceptions.\
MissingBooleanOptionsError(
true_option='--auto-terminate',
false_option='--no-auto-terminate')
instances_config['KeepJobFlowAliveWhenNoSteps'] = \
emrutils.apply_boolean_options(
parsed_args.no_auto_terminate,
'--no-auto-terminate',
parsed_args.auto_terminate,
'--auto-terminate')
instances_config['TerminationProtected'] = \
emrutils.apply_boolean_options(
parsed_args.termination_protected,
'--termination-protected',
parsed_args.no_termination_protected,
'--no-termination-protected')
params['VisibleToAllUsers'] = \
emrutils.apply_boolean_options(
parsed_args.visible_to_all_users,
'--visible-to-all-users',
parsed_args.no_visible_to_all_users,
'--no-visible-to-all-users')
params['Tags'] = emrutils.parse_tags(parsed_args.tags)
params['Instances'] = instances_config
if parsed_args.ec2_attributes is not None:
self._build_ec2_attributes(
cluster=params, parsed_attrs=parsed_args.ec2_attributes)
debugging_enabled = emrutils.apply_boolean_options(
parsed_args.enable_debugging,
'--enable-debugging',
parsed_args.no_enable_debugging,
'--no-enable-debugging')
if parsed_args.log_uri is None and debugging_enabled is True:
raise exceptions.LogUriError
if debugging_enabled is True:
self._update_cluster_dict(
cluster=params,
key='Steps',
value=[self._build_enable_debugging(parsed_globals)])
if parsed_args.applications is not None:
app_list, ba_list, step_list = applicationutils.build_applications(
parsed_applications=parsed_args.applications,
parsed_globals=parsed_globals,
ami_version=params['AmiVersion'])
self._update_cluster_dict(
params, 'NewSupportedProducts', app_list)
self._update_cluster_dict(
params, 'BootstrapActions', ba_list)
self._update_cluster_dict(
params, 'Steps', step_list)
hbase_restore_config = parsed_args.restore_from_hbase_backup
if hbase_restore_config is not None:
args = hbaseutils.build_hbase_restore_from_backup_args(
dir=hbase_restore_config.get('Dir'),
backup_version=hbase_restore_config.get('BackupVersion'))
step_config = emrutils.build_step(
jar=constants.HBASE_JAR_PATH,
name=constants.HBASE_RESTORE_STEP_NAME,
action_on_failure=constants.CANCEL_AND_WAIT,
args=args)
self._update_cluster_dict(
params, 'Steps', [step_config])
if parsed_args.bootstrap_actions is not None:
self._build_bootstrap_actions(
cluster=params,
parsed_boostrap_actions=parsed_args.bootstrap_actions)
if parsed_args.steps is not None:
steps_list = steputils.build_step_config_list(
parsed_step_list=parsed_args.steps,
region=parsed_globals.region)
self._update_cluster_dict(
cluster=params, key='Steps', value=steps_list)
self._validate_required_applications(parsed_args)
run_job_flow = emr.get_operation('RunJobFlow')
run_job_flow_response = emrutils.call(
self._session, run_job_flow, params,
parsed_globals.region, parsed_globals.endpoint_url,
parsed_globals.verify_ssl)
constructed_result = self._construct_result(run_job_flow_response[1])
emrutils.display_response(self._session, run_job_flow,
constructed_result, parsed_globals)
return 0