def scale_cluster(cluster, instances):
if not instances:
return
if not cmd.is_pre_installed_cdh(instances[0].remote()):
_configure_os(instances)
_install_packages(instances, PACKAGES)
_start_cloudera_agents(instances)
_await_agents(instances)
for instance in instances:
_configure_instance(instance)
cu.update_configs(instance)
if 'DATANODE' in instance.node_group.node_processes:
cu.refresh_nodes(cluster, 'DATANODE', cu.HDFS_SERVICE_NAME)
_configure_swift_to_inst(instance)
if 'DATANODE' in instance.node_group.node_processes:
hdfs = cu.get_service('DATANODE', instance=instance)
cu.start_roles(hdfs, cu.get_role_name(instance, 'DATANODE'))
if 'NODEMANAGER' in instance.node_group.node_processes:
yarn = cu.get_service('NODEMANAGER', instance=instance)
cu.start_roles(yarn, cu.get_role_name(instance, 'NODEMANAGER'))
def scale_cluster(cluster, instances):
if not instances:
return
if not cmd.is_pre_installed_cdh(instances[0].remote()):
_configure_os(instances)
_install_packages(instances, PACKAGES)
_start_cloudera_agents(instances)
_await_agents(instances)
for instance in instances:
_configure_instance(instance)
cu.update_configs(instance)
if 'DATANODE' in instance.node_group.node_processes:
cu.refresh_nodes(cluster, 'DATANODE', cu.HDFS_SERVICE_NAME)
_configure_swift_to_inst(instance)
if 'DATANODE' in instance.node_group.node_processes:
hdfs = cu.get_service('DATANODE', instance=instance)
cu.start_roles(hdfs, cu.get_role_name(instance, 'DATANODE'))
if 'NODEMANAGER' in instance.node_group.node_processes:
yarn = cu.get_service('NODEMANAGER', instance=instance)
cu.start_roles(yarn, cu.get_role_name(instance, 'NODEMANAGER'))
def _add_role(instance, process):
if process in ["MANAGER"]:
return
service = cu.get_service(process, instance=instance)
role = service.create_role(cu.get_role_name(instance, process), process, instance.fqdn())
role.update_config(_get_configs(process, node_group=instance.node_group))
def _add_role(instance, process):
if process in ['MANAGER']:
return
service = cu.get_service(process, instance=instance)
role = service.create_role(cu.get_role_name(instance, process), process,
instance.fqdn())
role.update_config(_get_configs(process, node_group=instance.node_group))
def _add_role(instance, process):
if process in ['CLOUDERA_MANAGER']:
return
process = pu.convert_role_showname(process)
service = cu.get_service(process, instance=instance)
role = service.create_role(cu.get_role_name(instance, process),
process, instance.fqdn())
role.update_config(_get_configs(process, node_group=instance.node_group))
def decommission_cluster(cluster, instances):
dns = []
nms = []
for i in instances:
if 'DATANODE' in i.node_group.node_processes:
dns.append(cu.get_role_name(i, 'DATANODE'))
if 'NODEMANAGER' in i.node_group.node_processes:
nms.append(cu.get_role_name(i, 'NODEMANAGER'))
if dns:
cu.decommission_nodes(cluster, 'DATANODE', dns)
if nms:
cu.decommission_nodes(cluster, 'NODEMANAGER', nms)
cu.delete_instances(cluster, instances)
cu.refresh_nodes(cluster, 'DATANODE', cu.HDFS_SERVICE_NAME)
cu.refresh_nodes(cluster, 'NODEMANAGER', cu.YARN_SERVICE_NAME)
def decommission_cluster(cluster, instances):
dns = []
nms = []
for i in instances:
if 'DATANODE' in i.node_group.node_processes:
dns.append(cu.get_role_name(i, 'DATANODE'))
if 'NODEMANAGER' in i.node_group.node_processes:
nms.append(cu.get_role_name(i, 'NODEMANAGER'))
if dns:
cu.decommission_nodes(cluster, 'DATANODE', dns)
if nms:
cu.decommission_nodes(cluster, 'NODEMANAGER', nms)
cu.delete_instances(cluster, instances)
cu.refresh_nodes(cluster, 'DATANODE', cu.HDFS_SERVICE_NAME)
cu.refresh_nodes(cluster, 'NODEMANAGER', cu.YARN_SERVICE_NAME)
def _get_configs(service, cluster=None, node_group=None):
def get_hadoop_dirs(mount_points, suffix):
return ','.join([x + suffix for x in mount_points])
all_confs = {}
if cluster:
zk_count = v._get_inst_count(cluster, 'ZOOKEEPER_SERVER')
hbm_count = v._get_inst_count(cluster, 'HBASE_MASTER')
snt_count = v._get_inst_count(cluster, 'SENTRY_SERVER')
ks_count = v._get_inst_count(cluster, 'KEY_VALUE_STORE_INDEXER')
imp_count = v._get_inst_count(cluster, 'IMPALA_CATALOGSERVER')
core_site_safety_valve = ''
if c_helper.is_swift_enabled(cluster):
configs = swift_helper.get_swift_configs()
confs = dict((c['name'], c['value']) for c in configs)
core_site_safety_valve = xmlutils.create_elements_xml(confs)
all_confs = {
'HDFS': {
'zookeeper_service':
cu.ZOOKEEPER_SERVICE_NAME if zk_count else '',
'dfs_block_local_path_access_user':
'******' if imp_count else '',
'core_site_safety_valve': core_site_safety_valve
},
'HIVE': {
'mapreduce_yarn_service': cu.YARN_SERVICE_NAME,
'zookeeper_service':
cu.ZOOKEEPER_SERVICE_NAME if zk_count else ''
},
'OOZIE': {
'mapreduce_yarn_service': cu.YARN_SERVICE_NAME,
'zookeeper_service':
cu.ZOOKEEPER_SERVICE_NAME if zk_count else ''
},
'YARN': {
'hdfs_service': cu.HDFS_SERVICE_NAME,
'zookeeper_service':
cu.ZOOKEEPER_SERVICE_NAME if zk_count else ''
},
'HUE': {
'hive_service': cu.HIVE_SERVICE_NAME,
'oozie_service': cu.OOZIE_SERVICE_NAME,
'sentry_service': cu.SENTRY_SERVICE_NAME if snt_count else '',
'zookeeper_service':
cu.ZOOKEEPER_SERVICE_NAME if zk_count else ''
},
'SPARK_ON_YARN': {
'yarn_service': cu.YARN_SERVICE_NAME
},
'HBASE': {
'hdfs_service': cu.HDFS_SERVICE_NAME,
'zookeeper_service': cu.ZOOKEEPER_SERVICE_NAME,
'hbase_enable_indexing': 'true' if ks_count else 'false',
'hbase_enable_replication': 'true' if ks_count else 'false'
},
'FLUME': {
'hdfs_service': cu.HDFS_SERVICE_NAME,
'hbase_service': cu.HBASE_SERVICE_NAME if hbm_count else ''
},
'SENTRY': {
'hdfs_service': cu.HDFS_SERVICE_NAME
},
'SOLR': {
'hdfs_service': cu.HDFS_SERVICE_NAME,
'zookeeper_service': cu.ZOOKEEPER_SERVICE_NAME
},
'SQOOP': {
'mapreduce_yarn_service': cu.YARN_SERVICE_NAME
},
'KS_INDEXER': {
'hbase_service': cu.HBASE_SERVICE_NAME,
'solr_service': cu.SOLR_SERVICE_NAME
},
'IMPALA': {
'hdfs_service': cu.HDFS_SERVICE_NAME,
'hbase_service': cu.HBASE_SERVICE_NAME if hbm_count else '',
'hive_service': cu.HIVE_SERVICE_NAME
}
}
hive_confs = {
'HIVE': {
'hive_metastore_database_type':
'postgresql',
'hive_metastore_database_host':
pu.get_manager(cluster).internal_ip,
'hive_metastore_database_port':
'7432',
'hive_metastore_database_password':
db_helper.get_hive_db_password(cluster)
}
}
hue_confs = {
'HUE': {
'hue_webhdfs':
cu.get_role_name(pu.get_namenode(cluster), 'NAMENODE')
}
}
sentry_confs = {
'SENTRY': {
'sentry_server_database_type':
'postgresql',
'sentry_server_database_host':
pu.get_manager(cluster).internal_ip,
'sentry_server_database_port':
'7432',
'sentry_server_database_password':
db_helper.get_sentry_db_password(cluster)
}
}
all_confs = _merge_dicts(all_confs, hue_confs)
all_confs = _merge_dicts(all_confs, hive_confs)
all_confs = _merge_dicts(all_confs, sentry_confs)
all_confs = _merge_dicts(all_confs, cluster.cluster_configs)
if node_group:
paths = node_group.storage_paths()
ng_default_confs = {
'NAMENODE': {
'dfs_name_dir_list': get_hadoop_dirs(paths, '/fs/nn')
},
'SECONDARYNAMENODE': {
'fs_checkpoint_dir_list': get_hadoop_dirs(paths, '/fs/snn')
},
'DATANODE': {
'dfs_data_dir_list': get_hadoop_dirs(paths, '/fs/dn'),
'dfs_datanode_data_dir_perm': 755,
'dfs_datanode_handler_count': 30
},
'NODEMANAGER': {
'yarn_nodemanager_local_dirs':
get_hadoop_dirs(paths, '/yarn/local')
},
'SERVER': {
'maxSessionTimeout': 60000
}
}
ng_user_confs = pu.convert_process_configs(node_group.node_configs)
all_confs = _merge_dicts(all_confs, ng_user_confs)
all_confs = _merge_dicts(all_confs, ng_default_confs)
return all_confs.get(service, {})
def test_get_role_name(self):
inst_mock = mock.Mock()
inst_mock.hostname.return_value = 'spam-host'
self.assertEqual('eggs_spam_host', cu.get_role_name(inst_mock, 'eggs'))
def _get_configs(service, cluster=None, node_group=None):
def get_hadoop_dirs(mount_points, suffix):
return ','.join([x + suffix for x in mount_points])
all_confs = {
'OOZIE': {
'mapreduce_yarn_service': cu.YARN_SERVICE_NAME
},
'YARN': {
'hdfs_service': cu.HDFS_SERVICE_NAME
},
'HUE': {
'hive_service': cu.HIVE_SERVICE_NAME,
'oozie_service': cu.OOZIE_SERVICE_NAME
},
'SPARK_ON_YARN': {
'yarn_service': cu.YARN_SERVICE_NAME
}
}
if node_group:
paths = node_group.storage_paths()
ng_default_confs = {
'NAMENODE': {
'dfs_name_dir_list': get_hadoop_dirs(paths, '/fs/nn')
},
'SECONDARYNAMENODE': {
'fs_checkpoint_dir_list': get_hadoop_dirs(paths, '/fs/snn')
},
'DATANODE': {
'dfs_data_dir_list': get_hadoop_dirs(paths, '/fs/dn')
},
'NODEMANAGER': {
'yarn_nodemanager_local_dirs': get_hadoop_dirs(paths,
'/yarn/local')
}
}
ng_user_confs = node_group.node_configs
all_confs = _merge_dicts(all_confs, ng_user_confs)
all_confs = _merge_dicts(all_confs, ng_default_confs)
if cluster:
hive_confs = {
'HIVE': {
'hive_metastore_database_type': 'postgresql',
'hive_metastore_database_host':
pu.get_manager(cluster).internal_ip,
'hive_metastore_database_port': '7432',
'hive_metastore_database_password':
db_helper.get_hive_db_password(cluster),
'mapreduce_yarn_service': cu.YARN_SERVICE_NAME
}
}
hue_confs = {
'HUE': {
'hue_webhdfs': cu.get_role_name(pu.get_namenode(cluster),
'NAMENODE')
}
}
all_confs = _merge_dicts(all_confs, hue_confs)
all_confs = _merge_dicts(all_confs, hive_confs)
all_confs = _merge_dicts(all_confs, cluster.cluster_configs)
return all_confs.get(service, {})
def _get_configs(service, cluster=None, node_group=None):
def get_hadoop_dirs(mount_points, suffix):
return ','.join([x + suffix for x in mount_points])
all_confs = {}
if cluster:
zk_count = v._get_inst_count(cluster, 'ZOOKEEPER_SERVER')
hbm_count = v._get_inst_count(cluster, 'HBASE_MASTER')
snt_count = v._get_inst_count(cluster, 'SENTRY_SERVER')
ks_count = v._get_inst_count(cluster, 'KEY_VALUE_STORE_INDEXER')
imp_count = v._get_inst_count(cluster, 'IMPALA_CATALOGSERVER')
core_site_safety_valve = ''
if c_helper.is_swift_enabled(cluster):
configs = swift_helper.get_swift_configs()
confs = dict((c['name'], c['value']) for c in configs)
core_site_safety_valve = xmlutils.create_elements_xml(confs)
all_confs = {
'HDFS': {
'zookeeper_service':
cu.ZOOKEEPER_SERVICE_NAME if zk_count else '',
'dfs_block_local_path_access_user':
'******' if imp_count else '',
'core_site_safety_valve': core_site_safety_valve
},
'HIVE': {
'mapreduce_yarn_service': cu.YARN_SERVICE_NAME,
'zookeeper_service':
cu.ZOOKEEPER_SERVICE_NAME if zk_count else ''
},
'OOZIE': {
'mapreduce_yarn_service': cu.YARN_SERVICE_NAME,
'zookeeper_service':
cu.ZOOKEEPER_SERVICE_NAME if zk_count else ''
},
'YARN': {
'hdfs_service': cu.HDFS_SERVICE_NAME,
'zookeeper_service':
cu.ZOOKEEPER_SERVICE_NAME if zk_count else ''
},
'HUE': {
'hive_service': cu.HIVE_SERVICE_NAME,
'oozie_service': cu.OOZIE_SERVICE_NAME,
'sentry_service': cu.SENTRY_SERVICE_NAME if snt_count else '',
'zookeeper_service':
cu.ZOOKEEPER_SERVICE_NAME if zk_count else ''
},
'SPARK_ON_YARN': {
'yarn_service': cu.YARN_SERVICE_NAME
},
'HBASE': {
'hdfs_service': cu.HDFS_SERVICE_NAME,
'zookeeper_service': cu.ZOOKEEPER_SERVICE_NAME,
'hbase_enable_indexing': 'true' if ks_count else 'false',
'hbase_enable_replication': 'true' if ks_count else 'false'
},
'FLUME': {
'hdfs_service': cu.HDFS_SERVICE_NAME,
'hbase_service': cu.HBASE_SERVICE_NAME if hbm_count else ''
},
'SENTRY': {
'hdfs_service': cu.HDFS_SERVICE_NAME
},
'SOLR': {
'hdfs_service': cu.HDFS_SERVICE_NAME,
'zookeeper_service': cu.ZOOKEEPER_SERVICE_NAME
},
'SQOOP': {
'mapreduce_yarn_service': cu.YARN_SERVICE_NAME
},
'KS_INDEXER': {
'hbase_service': cu.HBASE_SERVICE_NAME,
'solr_service': cu.SOLR_SERVICE_NAME
},
'IMPALA': {
'hdfs_service': cu.HDFS_SERVICE_NAME,
'hbase_service': cu.HBASE_SERVICE_NAME if hbm_count else '',
'hive_service': cu.HIVE_SERVICE_NAME
}
}
hive_confs = {
'HIVE': {
'hive_metastore_database_type': 'postgresql',
'hive_metastore_database_host':
pu.get_manager(cluster).internal_ip,
'hive_metastore_database_port': '7432',
'hive_metastore_database_password':
db_helper.get_hive_db_password(cluster)
}
}
hue_confs = {
'HUE': {
'hue_webhdfs': cu.get_role_name(pu.get_namenode(cluster),
'NAMENODE')
}
}
sentry_confs = {
'SENTRY': {
'sentry_server_database_type': 'postgresql',
'sentry_server_database_host':
pu.get_manager(cluster).internal_ip,
'sentry_server_database_port': '7432',
'sentry_server_database_password':
db_helper.get_sentry_db_password(cluster)
}
}
all_confs = _merge_dicts(all_confs, hue_confs)
all_confs = _merge_dicts(all_confs, hive_confs)
all_confs = _merge_dicts(all_confs, sentry_confs)
all_confs = _merge_dicts(all_confs, cluster.cluster_configs)
if node_group:
paths = node_group.storage_paths()
ng_default_confs = {
'NAMENODE': {
'dfs_name_dir_list': get_hadoop_dirs(paths, '/fs/nn')
},
'SECONDARYNAMENODE': {
'fs_checkpoint_dir_list': get_hadoop_dirs(paths, '/fs/snn')
},
'DATANODE': {
'dfs_data_dir_list': get_hadoop_dirs(paths, '/fs/dn'),
'dfs_datanode_data_dir_perm': 755,
'dfs_datanode_handler_count': 30
},
'NODEMANAGER': {
'yarn_nodemanager_local_dirs': get_hadoop_dirs(paths,
'/yarn/local')
},
'SERVER': {
'maxSessionTimeout': 60000
}
}
ng_user_confs = pu.convert_process_configs(node_group.node_configs)
all_confs = _merge_dicts(all_confs, ng_user_confs)
all_confs = _merge_dicts(all_confs, ng_default_confs)
return all_confs.get(service, {})
def _get_configs(service, cluster=None, node_group=None):
def get_hadoop_dirs(mount_points, suffix):
return ','.join([x + suffix for x in mount_points])
all_confs = {
'OOZIE': {
'mapreduce_yarn_service': cu.YARN_SERVICE_NAME
},
'YARN': {
'hdfs_service': cu.HDFS_SERVICE_NAME
},
'HUE': {
'hive_service': cu.HIVE_SERVICE_NAME,
'oozie_service': cu.OOZIE_SERVICE_NAME
},
'SPARK_ON_YARN': {
'yarn_service': cu.YARN_SERVICE_NAME
},
'HBASE': {
'hdfs_service': cu.HDFS_SERVICE_NAME,
'zookeeper_service': cu.ZOOKEEPER_SERVICE_NAME
}
}
if node_group:
paths = node_group.storage_paths()
ng_default_confs = {
'NAMENODE': {
'dfs_name_dir_list': get_hadoop_dirs(paths, '/fs/nn')
},
'SECONDARYNAMENODE': {
'fs_checkpoint_dir_list': get_hadoop_dirs(paths, '/fs/snn')
},
'DATANODE': {
'dfs_data_dir_list': get_hadoop_dirs(paths, '/fs/dn')
},
'NODEMANAGER': {
'yarn_nodemanager_local_dirs':
get_hadoop_dirs(paths, '/yarn/local')
}
}
ng_user_confs = pu.convert_process_configs(node_group.node_configs)
all_confs = _merge_dicts(all_confs, ng_user_confs)
all_confs = _merge_dicts(all_confs, ng_default_confs)
if cluster:
hive_confs = {
'HIVE': {
'hive_metastore_database_type':
'postgresql',
'hive_metastore_database_host':
pu.get_manager(cluster).internal_ip,
'hive_metastore_database_port':
'7432',
'hive_metastore_database_password':
db_helper.get_hive_db_password(cluster),
'mapreduce_yarn_service':
cu.YARN_SERVICE_NAME
}
}
hue_confs = {
'HUE': {
'hue_webhdfs':
cu.get_role_name(pu.get_namenode(cluster), 'NAMENODE')
}
}
all_confs = _merge_dicts(all_confs, hue_confs)
all_confs = _merge_dicts(all_confs, hive_confs)
all_confs = _merge_dicts(all_confs, cluster.cluster_configs)
return all_confs.get(service, {})
def test_get_role_name(self):
inst_mock = mock.Mock()
inst_mock.hostname.return_value = 'spam-host'
self.assertEqual('eggs_spam_host', cu.get_role_name(inst_mock, 'eggs'))
