def set_up_cluster():
# get a handle on the instance of CM that we have running
api = ApiResource(cm_host, cm_port, cm_username, cm_password, version=7)
# get the CM instance
cm = ClouderaManager(api)
# activate the CM trial license
cm.begin_trial()
# create the management service
service_setup = ApiServiceSetupInfo(name=cm_service_name, type="MGMT")
cm.create_mgmt_service(service_setup)
# install hosts on this CM instance
cmd = cm.host_install(host_username, host_list, password=host_password, cm_repo_url=cm_repo_url)
print "Installing hosts. This might take a while."
while cmd.success == None:
sleep(5)
cmd = cmd.fetch()
if cmd.success != True:
print "cm_host_install failed: " + cmd.resultMessage
exit(0)
print "cm_host_install succeeded"
# first auto-assign roles and auto-configure the CM service
cm.auto_assign_roles()
cm.auto_configure()
# create a cluster on that instance
cluster = create_cluster(api, cluster_name, cdh_version)
# add all our hosts to the cluster
cluster.add_hosts(host_list)
cluster = api.get_cluster("Cluster 1")
parcels_list = []
# get and list all available parcels
print "Available parcels:"
for p in cluster.get_all_parcels():
print '\t' + p.product + ' ' + p.version
if p.version.startswith(cdh_version_number) and p.product == "CDH":
parcels_list.append(p)
if len(parcels_list) == 0:
print "No " + cdh_version + " parcel found!"
exit(0)
cdh_parcel = parcels_list[0]
for p in parcels_list:
if p.version > cdh_parcel.version:
cdh_parcel = p
# download the parcel
print "Starting parcel download. This might take a while."
cmd = cdh_parcel.start_download()
if cmd.success != True:
print "Parcel download failed!"
exit(0)
# make sure the download finishes
while cdh_parcel.stage != 'DOWNLOADED':
sleep(5)
cdh_parcel = get_parcel(api, cdh_parcel.product, cdh_parcel.version, cluster_name)
print cdh_parcel.product + ' ' + cdh_parcel.version + " downloaded"
# distribute the parcel
print "Starting parcel distribution. This might take a while."
cmd = cdh_parcel.start_distribution()
if cmd.success != True:
print "Parcel distribution failed!"
exit(0)
# make sure the distribution finishes
while cdh_parcel.stage != "DISTRIBUTED":
sleep(5)
cdh_parcel = get_parcel(api, cdh_parcel.product, cdh_parcel.version, cluster_name)
print cdh_parcel.product + ' ' + cdh_parcel.version + " distributed"
# activate the parcel
cmd = cdh_parcel.activate()
if cmd.success != True:
print "Parcel activation failed!"
exit(0)
# make sure the activation finishes
while cdh_parcel.stage != "ACTIVATED":
cdh_parcel = get_parcel(api, cdh_parcel.product, cdh_parcel.version, cluster_name)
print cdh_parcel.product + ' ' + cdh_parcel.version + " activated"
# inspect hosts and print the result
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(0)
print "Hosts successfully inspected: \n" + cmd.resultMessage
# create all the services we want to add; we will only create one instance
# of each
for s in service_types_and_names.keys():
service = cluster.create_service(service_types_and_names[s], s)
# we will auto-assign roles; you can manually assign roles using the
# /clusters/{clusterName}/services/{serviceName}/role endpoint or by using
# ApiService.createRole()
cluster.auto_assign_roles()
cluster.auto_configure()
# this will set up the Hive and the reports manager databases because we
# can't auto-configure those two things
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_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)
# start the management service
cm_service = cm.get_service()
cm_service.start().wait()
# this will set the Reports Manager database password
# first we find the correct role
rm_role = None
for r in cm.get_service().get_all_roles():
if r.type == "REPORTSMANAGER":
rm_role = r
if rm_role == None:
print "No REPORTSMANAGER role found!"
exit(0)
# then we get the corresponding role config group -- even though there is
# only once instance of each CM management service, we do this just in case
# it is not placed in the base group
rm_role_group = rm_role.roleConfigGroupRef
rm_rcg = get_role_config_group(api, rm_role.type, \
rm_role_group.roleConfigGroupName, None)
# update the appropriate fields in the config
rm_rcg_config = { "headlamp_database_host" : reports_manager_host, \
"headlamp_database_name" : reports_manager_name, \
"headlamp_database_user" : reports_manager_username, \
"headlamp_database_password" : reports_manager_password, \
"headlamp_database_type" : reports_manager_database_type }
rm_rcg.update_config(rm_rcg_config)
# restart the management service with new configs
cm_service.restart().wait()
# execute the first run command
print "Excuting first run command. This might take a while."
cmd = cluster.first_run()
while cmd.success == None:
cmd = cmd.fetch()
if cmd.success != True:
print "The first run command failed: " + cmd.resultMessage()
exit(0)
print "First run successfully executed. Your cluster has been set up!"
def set_up_cluster():
# get a handle on the instance of CM that we have running
api = ApiResource(cm_host, cm_port, cm_username, cm_password, version=19)
# get the CM instance
cm = ClouderaManager(api)
print "*************************************"
print " Starting Auto Deployment of Cluster "
print "*************************************"
# {'owner': ROAttr(), 'uuid': ROAttr(), 'expiration': ROAttr(),}
TRIAL = False
try:
trial_active = cm.get_license()
print trial_active
if trial_active.owner == "Trial License":
print "Trial License is already set - will NOT continue now."
print "Assuming Cluster is already setup"
TRIAL = True
else:
print "Setting up `Trial License`."
cm.begin_trial()
except:
cm.begin_trial()
if TRIAL:
exit(0)
# create the management service
service_setup = ApiServiceSetupInfo(name=cm_service_name, type="MGMT")
try:
if not cm.get_service().name:
cm.create_mgmt_service(service_setup)
else:
print "Service already exist."
except:
cm.create_mgmt_service(service_setup)
# install hosts on this CM instance
cmd = cm.host_install(host_username,
host_list,
password=host_password,
cm_repo_url=cm_repo_url,
unlimited_jce=True)
print "Installing hosts. This might take a while."
while cmd.success == None:
sleep(5)
cmd = cmd.fetch()
print cmd
if cmd.success != True:
print "cm_host_install failed: " + cmd.resultMessage
exit(0)
print "cm_host_install succeeded"
# first auto-assign roles and auto-configure the CM service
cm.auto_assign_roles()
cm.auto_configure()
# create a cluster on that instance
cluster = create_cluster(api, cluster_name, cdh_version)
# add all our hosts to the cluster
cluster.add_hosts(host_list)
cluster = api.get_cluster(cluster_name)
parcels_list = []
# get and list all available parcels
print "Available parcels:"
for p in cluster.get_all_parcels():
print '\t' + p.product + ' ' + p.version
if p.version.startswith(cdh_version_number) and p.product == "CDH":
parcels_list.append(p)
if len(parcels_list) == 0:
print "No " + cdh_version + " parcel found!"
exit(0)
cdh_parcel = parcels_list[0]
for p in parcels_list:
if p.version > cdh_parcel.version:
cdh_parcel = p
# download the parcel
print "Starting parcel download. This might take a while."
cmd = cdh_parcel.start_download()
if cmd.success != True:
print "Parcel download failed!"
exit(0)
# make sure the download finishes
while cdh_parcel.stage != 'DOWNLOADED':
sleep(5)
cdh_parcel = get_parcel(api, cdh_parcel.product, cdh_parcel.version,
cluster_name)
print cdh_parcel.product + ' ' + cdh_parcel.version + " downloaded"
# distribute the parcel
print "Starting parcel distribution. This might take a while."
cmd = cdh_parcel.start_distribution()
if cmd.success != True:
print "Parcel distribution failed!"
exit(0)
# make sure the distribution finishes
while cdh_parcel.stage != "DISTRIBUTED":
sleep(5)
cdh_parcel = get_parcel(api, cdh_parcel.product, cdh_parcel.version,
cluster_name)
print cdh_parcel.product + ' ' + cdh_parcel.version + " distributed"
# activate the parcel
cmd = cdh_parcel.activate()
if cmd.success != True:
print "Parcel activation failed!"
exit(0)
# make sure the activation finishes
while cdh_parcel.stage != "ACTIVATED":
cdh_parcel = get_parcel(api, cdh_parcel.product, cdh_parcel.version,
cluster_name)
print cdh_parcel.product + ' ' + cdh_parcel.version + " activated"
# inspect hosts and print the result
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(0)
print "Hosts successfully inspected: \n" + cmd.resultMessage
# create all the services we want to add; we will only create one instance
# of each
for s in service_types_and_names.keys():
service = cluster.create_service(service_types_and_names[s], s)
# we will auto-assign roles; you can manually assign roles using the
# /clusters/{clusterName}/services/{serviceName}/role endpoint or by using
# ApiService.createRole()
cluster.auto_assign_roles()
cluster.auto_configure()
# # this will set up the Hive and the reports manager databases because we
# # can't auto-configure those two things
# 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_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)
# start the management service
cm_service = cm.get_service()
cm_service.start().wait()
# this will set the Reports Manager database password
# first we find the correct role
rm_role = None
for r in cm.get_service().get_all_roles():
if r.type == "REPORTSMANAGER":
rm_role = r
if rm_role == None:
print "No REPORTSMANAGER role found!"
exit(0)
# then we get the corresponding role config group -- even though there is
# only once instance of each CM management service, we do this just in case
# it is not placed in the base group
rm_role_group = rm_role.roleConfigGroupRef
rm_rcg = get_role_config_group(api, rm_role.type, \
rm_role_group.roleConfigGroupName, None)
# update the appropriate fields in the config
rm_rcg_config = {"headlamp_database_host": reports_manager_host, \
"headlamp_database_name": reports_manager_name, \
"headlamp_database_user": reports_manager_username, \
"headlamp_database_password": reports_manager_password, \
"headlamp_database_type": reports_manager_database_type}
rm_rcg.update_config(rm_rcg_config)
# restart the management service with new configs
cm_service.restart().wait()
# execute the first run command
print "Excuting first run command. This might take a while."
cmd = cluster.first_run()
while cmd.success == None:
cmd = cmd.fetch()
if cmd.success != True:
print "The first run command failed: " + cmd.resultMessage()
exit(0)
print "First run successfully executed. Your cluster has been set up!"
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."