def main():
parser = cm_args_parser()
args = parser.parse_args()
print "conencting to host:",args.cm_host+"..."
api = ApiResource(args.cm_host, username=args.cm_user, password=args.cm_password)
cm = ClouderaManager(api)
cm.update_config({"dssd_enabled":"true"})
def install_cds(cm_host, host_list):
print "Installing CDS for Spark2"
api = ApiResource(cm_host, cm_port, cm_username, cm_password, version=7)
cm = ClouderaManager(api)
config = cm.get_config(view='full')
# Add parcel repository
repolist = config['REMOTE_PARCEL_REPO_URLS']
value = repolist.value or repolist.default
value += ',' + cds_parcel_repo
cm.update_config({'REMOTE_PARCEL_REPO_URLS': value})
sleep(10)
# Install CDS parcel
cluster = api.get_cluster(cluster_name)
parcel = cluster.get_parcel(cds_parcel_name, cds_parcel_version)
print "Downloading CDS parcel. This might take a while."
if parcel.stage == "AVAILABLE_REMOTELY":
parcel = wait_for_parcel(parcel.start_download(), api, parcel,
cluster_name, 'DOWNLOADED')
print "Distributing CDS parcel. This might take a while."
if parcel.stage == "DOWNLOADED":
parcel = wait_for_parcel(parcel.start_distribution(), api, parcel,
cluster_name, 'DISTRIBUTED')
print "Activating CDS parcel. This might take a while."
if parcel.stage == "DISTRIBUTED":
parcel = wait_for_parcel(parcel.activate(), api, parcel, cluster_name,
'ACTIVATED')
service = cluster.create_service(cds_service_name, cds_service_type)
slaves = [host for host in host_list if 'slave' in host]
edges = [host for host in host_list if 'edge' in host]
service.create_role('SPARK2_YARN_HISTORY_SERVER-1',
'SPARK2_YARN_HISTORY_SERVER', cm_host)
service.create_role('SPARK2_ON_YARN-GW_MASTER1', 'GATEWAY', cm_host)
for (i, edge) in enumerate(edges):
service.create_role('SPARK2_ON_YARN-GW_EDGE%s' % i, 'GATEWAY', edge)
for (i, slave) in enumerate(slaves):
service.create_role('SPARK2_ON_YARN-GW_SLAVE%s' % i, 'GATEWAY', slave)
cluster.auto_configure()
# Restart Cloudera Management Service and cluster
cm_service = cm.get_service()
cm_service.restart().wait()
cluster.restart(restart_only_stale_services=True,
redeploy_client_configuration=True).wait()
# Due to (presumably) CM bug, auto_configure() after Kafka installation creates additional
# role config group for HDFS gateway, which breaks further use of auto_configure().
# Below we remove it if it exists.
try:
hdfs_service = cluster.get_service("HDFS-1")
hdfs_service.delete_role_config_group("HDFS-1-GATEWAY-1")
except cm_api.api_client.ApiException:
print(
"Not removing HDFS Gateway role config group as it doesn't exist")
else:
print("Removed additional HDFS Gateway role config group")
print "CDS is now installed."
def main():
module = AnsibleModule(
argument_spec=dict(
cm_host=dict(required=True, type='str'),
cm_port=dict(required=False, type='int', default=7180),
cm_username=dict(required=True, type='str'),
cm_password=dict(required=True, type='str', no_log=True),
cm_tls=dict(required=False, type='bool', default=False),
cm_version=dict(required=False, type='int', default=10),
name=dict(required=False, type='str'),
value=dict(required=False, type='str'),
state=dict(default='present',
choices=['present', 'absent', 'list'])
)
)
cm_host = module.params.get('cm_host')
cm_port = module.params.get('cm_port')
cm_username = module.params.get('cm_username')
cm_password = module.params.get('cm_password')
cm_tls = module.params.get('cm_tls')
cm_version = module.params.get('cm_version')
cm_config_key = module.params.get('name')
cm_config_value = module.params.get('value')
state = module.params.get('state')
changed = False
if not HAS_CM_API:
module.fail_json(changed=changed, msg='cm_api required for this module')
try:
cm_conn = ApiResource(cm_host, server_port=cm_port,
username=cm_username, password=cm_password,
use_tls=cm_tls, version=cm_version)
cms = ClouderaManager(cm_conn)
except ApiException as e:
module.fail_json(changed=changed,
msg="Can't connect to API: {}".format(e))
try:
_settings = cms.get_config('full')
except ApiException as e:
module.fail_json(changed=changed, msg="{}".format(e))
_settings = cms.get_config('full')
settings = dict()
for key in _settings:
settings[key] = dict()
settings[key]['name'] = _settings[key].name
settings[key]['default'] = _settings[key].default
settings[key]['value'] = None
if _settings[key].value is not None:
settings[key]['value'] = _settings[key].value
if state == 'list':
module.exit_json(changed=changed, settings=settings)
if cm_config_key is None:
module.fail_json(changed=changed, msg='Missing `name` option.')
if cm_config_key not in settings:
module.fail_json(changed=changed,
msg='{} is not a valid configuration entry'.format(cm_config_key))
if state == "absent":
cm_config_value = None
elif cm_config_value is None:
module.fail_json(changed=changed, msg='Missing `value` option.')
if cm_config_value != settings[cm_config_key]['value']:
try:
update = dict()
update[cm_config_key] = cm_config_value
rc = cms.update_config(update)
module.exit_json(changed=True, settings=rc)
except Exception as e:
module.fail_json(changed=False, msg="{}".format(e))
module.exit_json(changed=False, settings=cms.get_config('summary'))
def set_up_cluster(cm_host, host_list):
print "Setting up CDH cluster..."
api = ApiResource(cm_host, cm_port, cm_username, cm_password, version=7)
cm = ClouderaManager(api)
print "Creating mgmg service."
try:
service_setup = ApiServiceSetupInfo(name=cm_service_name, type="MGMT")
cm.create_mgmt_service(service_setup)
except ApiException as exc:
if exc.code != 400:
print "create MGMT service failed: " + exc
exit(1)
print "Installing hosts. This might take a while."
cmd = cm.host_install(host_username, host_list,
password=host_password).wait()
if cmd.success != True:
print "cm_host_install failed: " + cmd.resultMessage
exit(2)
print "Auto-assign roles and auto-configure the CM service"
if not is_cluster_installed(api):
cm.auto_assign_roles()
cm.auto_configure()
print "Creating cluster."
if not is_cluster_installed(api):
cluster = create_cluster(api, cluster_name, cdh_version)
cluster.add_hosts(host_list)
cluster = api.get_cluster(cluster_name)
cdh_parcel = get_cdh_parcel(cluster)
print "Downloading CDH parcel. This might take a while."
if cdh_parcel.stage == "AVAILABLE_REMOTELY":
cdh_parcel = wait_for_parcel(cdh_parcel.start_download(), api,
cdh_parcel, cluster_name, 'DOWNLOADED')
print "Distributing CDH parcel. This might take a while."
if cdh_parcel.stage == "DOWNLOADED":
cdh_parcel = wait_for_parcel(cdh_parcel.start_distribution(), api,
cdh_parcel, cluster_name, 'DISTRIBUTED')
print "Activating CDH parcel. This might take a while."
if cdh_parcel.stage == "DISTRIBUTED":
cdh_parcel = wait_for_parcel(cdh_parcel.activate(), api, cdh_parcel,
cluster_name, 'ACTIVATED')
# if cdh_parcel.stage != "ACTIVATED":
# print "CDH parcel activation failed. Parcel in stage: " + cdh_parcel.stage
# exit(14)
print "Inspecting hosts. This might take a few minutes."
cmd = cm.inspect_hosts()
while cmd.success == None:
cmd = cmd.fetch()
if cmd.success != True:
print "Host inpsection failed!"
exit(8)
print "Hosts successfully inspected: \n" + cmd.resultMessage
print "Creating specified services."
for s in service_types_and_names.keys():
try:
cluster.get_service(service_types_and_names[s])
except:
print "Creating service: " + service_types_and_names[s]
service = cluster.create_service(service_types_and_names[s], s)
slaves = [host for host in host_list if 'slave' in host]
edges = [host for host in host_list if 'edge' in host]
#assign master roles to master node
for service in cluster.get_all_services():
if service.name == 'HDFS-1':
service.create_role('NAMENODE-1', 'NAMENODE', cm_host)
service.create_role('SECONDARYNAMENODE', 'SECONDARYNAMENODE',
cm_host)
service.create_role('BALANCER-1', 'BALANCER', cm_host)
service.create_role('HTTPFS-1', 'HTTPFS', cm_host)
service.create_role('HDFS-GW_MASTER1', 'GATEWAY', cm_host)
for (i, edge) in enumerate(edges):
service.create_role('HDFS-GW_EDGE%s' % i, 'GATEWAY', edge)
for (i, slave) in enumerate(slaves):
service.create_role('DATANODE-%s' % i, 'DATANODE', slave)
if service.name == 'ZOOKEEPER-1':
service.create_role('ZOOKEEPERSERVER-1', 'SERVER', cm_host)
if service.name == 'HBASE-1':
service.create_role('MASTER-1', 'MASTER', cm_host)
service.create_role('HBASETHRIFTSERVER-1', 'HBASETHRIFTSERVER',
cm_host)
for (i, slave) in enumerate(slaves):
service.create_role('HBASE-RS-%s' % i, 'REGIONSERVER', slave)
if service.name == 'HUE-1':
service.create_role('HUE-MASTER1', 'HUE_SERVER', cm_host)
service.create_role('HUE-LB_MASTER1', 'HUE_LOAD_BALANCER', cm_host)
for (i, edge) in enumerate(edges):
service.create_role('HUE-EDGE%s' % i, 'HUE_SERVER', edge)
if service.name == 'HIVE-1':
service.create_role('HIVEMETASTORE-1', 'HIVEMETASTORE', cm_host)
service.create_role('HIVESERVER-1', 'HIVESERVER2', cm_host)
service.create_role('HIVE-GW_MASTER1', 'GATEWAY', cm_host)
for (i, edge) in enumerate(edges):
service.create_role('HIVE-GW_EDGE%s' % i, 'GATEWAY', edge)
for (i, slave) in enumerate(slaves):
service.create_role('HIVE-GW_SLAVE%s' % i, 'GATEWAY', slave)
if service.name == 'IMPALA-1':
service.create_role('STATESTORE-1', 'STATESTORE', cm_host)
service.create_role('CATALOGSERVER-1', 'CATALOGSERVER', cm_host)
for (i, slave) in enumerate(slaves):
service.create_role('IMPALAD-%s' % i, 'IMPALAD', slave)
if service.name == 'OOZIE-1':
service.create_role('OOZIE_SERVER-1', 'OOZIE_SERVER', cm_host)
if service.name == 'SPARK_ON_YARN-1':
service.create_role('SPARK_YARN_HISTORY_SERVER-1',
'SPARK_YARN_HISTORY_SERVER', cm_host)
service.create_role('SPARK_ON_YARN-GW_MASTER%s' % i, 'GATEWAY',
cm_host)
for (i, edge) in enumerate(edges):
service.create_role('SPARK_ON_YARN-GW_EDGE%s' % i, 'GATEWAY',
edge)
for (i, slave) in enumerate(slaves):
service.create_role('SPARK_ON_YARN-GW_SLAVE%s' % i, 'GATEWAY',
slave)
if service.name == 'SQOOP-1':
service.create_role('SQOOP_SERVER-1', 'SQOOP_SERVER', cm_host)
if service.name == 'YARN-1':
service.create_role('RESOURCEMANAGER-1', 'RESOURCEMANAGER',
cm_host)
service.create_role('JOBHISTORY-1', 'JOBHISTORY', cm_host)
for (i, slave) in enumerate(slaves):
service.create_role('NODEMANAGER-%s' % i, 'NODEMANAGER', slave)
#print "Auto assigning roles."
#cluster.auto_assign_roles()
cluster.auto_configure()
print "Updating Hive config."
hive_metastore_host = cm_host # let's assume that
hive = cluster.get_service(service_types_and_names["HIVE"])
hive_config = { "hive_metastore_database_host" : hive_metastore_host, \
"hive_metastore_database_name" : hive_metastore_name, \
"hive_metastore_database_user" : hive_metastore_user, \
"hive_metastore_database_password" : hive_metastore_password, \
"hive_metastore_database_port" : hive_metastore_database_port, \
"hive_metastore_database_type" : hive_metastore_database_type }
hive.update_config(hive_config)
print "Updating Hue config."
hue_db_host = cm_host # let's assume that
hue = cluster.get_service(service_types_and_names["HUE"])
hue_config = { "database_host" : hue_db_host, \
"database_name" : hue_db_name, \
"database_user" : hue_db_user, \
"database_password" : hue_db_password, \
"database_port" : hue_db_port, \
"database_type" : hue_db_type }
hue.update_config(hue_config)
# Set Java version to OpenJDK
cm.update_all_hosts_config({'java_home': '/usr/lib/jvm/java-openjdk'})
print "Starting management service."
cm_service = cm.get_service()
cm_service.start().wait()
print "Excuting first run command. This might take a while."
cmd = cluster.first_run().wait()
if cmd.success != True:
print "The first run command failed: " + cmd.resultMessage
exit(11)
print "First run successfully executed. Your cluster has been set up!"
config = cm.get_config(view='full')
repolist = config['REMOTE_PARCEL_REPO_URLS']
value = repolist.value or repolist.default
value += ',' + anaconda_repo
cm.update_config({'REMOTE_PARCEL_REPO_URLS': value})
sleep(10)
cluster = api.get_cluster(cluster_name)
parcel = cluster.get_parcel('Anaconda', anaconda_parcel_version)
print "Downloading Anaconda parcel. This might take a while."
if parcel.stage == "AVAILABLE_REMOTELY":
parcel = wait_for_parcel(parcel.start_download(), api, parcel,
cluster_name, 'DOWNLOADED')
print "Distributing Anaconda parcel. This might take a while."
if parcel.stage == "DOWNLOADED":
parcel = wait_for_parcel(parcel.start_distribution(), api, parcel,
cluster_name, 'DISTRIBUTED')
print "Activating Anaconda parcel. This might take a while."
if parcel.stage == "DISTRIBUTED":
parcel = wait_for_parcel(parcel.activate(), api, parcel, cluster_name,
'ACTIVATED')
print "Anaconda is now installed."
