deb-sahara/sahara/plugins/cdh/cloudera_utils.py
Ken Chen 086ebb7658 Enable HDFS HA in Cloudera plugin
If HDFS_JOURNALNODE roles are selected, HDFS is assumed enabled.
enable_nn_ha will be called to do the work. We add the involved
implementation and cluster validation codes. We also modified the
CM client codes, for we put enable_nn_ha method in wrong class
before.

Partial-implements: blueprint cdh-ha-support
Change-Id: Id5c47d485e53b867d93ea9b4c94367587dc93c2a
2015-07-19 12:32:24 +08:00

291 lines
11 KiB
Python

# Copyright (c) 2014 Mirantis Inc.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
# implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import functools
from oslo_log import log as logging
import six
from sahara import context
from sahara.i18n import _
from sahara.plugins.cdh.client import api_client
from sahara.plugins.cdh.client import services
from sahara.plugins.cdh import db_helper
from sahara.plugins import exceptions as ex
from sahara.utils import cluster_progress_ops as cpo
from sahara.utils import poll_utils
LOG = logging.getLogger(__name__)
def cloudera_cmd(f):
@functools.wraps(f)
def wrapper(*args, **kwargs):
for cmd in f(*args, **kwargs):
result = cmd.wait()
if not result.success:
if result.children is not None:
for c in result.children:
if not c.success:
raise ex.HadoopProvisionError(c.resultMessage)
else:
raise ex.HadoopProvisionError(result.resultMessage)
return wrapper
class ClouderaUtils(object):
CM_DEFAULT_USERNAME = 'admin'
CM_DEFAULT_PASSWD = 'admin'
CM_API_VERSION = 6
HDFS_SERVICE_NAME = 'hdfs01'
YARN_SERVICE_NAME = 'yarn01'
OOZIE_SERVICE_NAME = 'oozie01'
HIVE_SERVICE_NAME = 'hive01'
HUE_SERVICE_NAME = 'hue01'
SPARK_SERVICE_NAME = 'spark_on_yarn01'
ZOOKEEPER_SERVICE_NAME = 'zookeeper01'
HBASE_SERVICE_NAME = 'hbase01'
def __init__(self):
# pu will be defined in derived class.
self.pu = None
def get_api_client_by_default_password(self, cluster):
manager_ip = self.pu.get_manager(cluster).management_ip
return api_client.ApiResource(manager_ip,
username=self.CM_DEFAULT_USERNAME,
password=self.CM_DEFAULT_PASSWD,
version=self.CM_API_VERSION)
def get_api_client(self, cluster):
manager_ip = self.pu.get_manager(cluster).management_ip
cm_password = db_helper.get_cm_password(cluster)
return api_client.ApiResource(manager_ip,
username=self.CM_DEFAULT_USERNAME,
password=cm_password,
version=self.CM_API_VERSION)
def update_cloudera_password(self, cluster):
api = self.get_api_client_by_default_password(cluster)
user = api.get_user(self.CM_DEFAULT_USERNAME)
user.password = db_helper.get_cm_password(cluster)
api.update_user(user)
def get_cloudera_cluster(self, cluster):
api = self.get_api_client(cluster)
return api.get_cluster(cluster.name)
@cloudera_cmd
def start_instances(self, cluster):
cm_cluster = self.get_cloudera_cluster(cluster)
yield cm_cluster.start()
@cpo.event_wrapper(True, step=_("Delete instances"), param=('cluster', 1))
def delete_instances(self, cluster, instances):
api = self.get_api_client(cluster)
cm_cluster = self.get_cloudera_cluster(cluster)
hosts = api.get_all_hosts(view='full')
hostsnames_to_deleted = [i.fqdn() for i in instances]
for host in hosts:
if host.hostname in hostsnames_to_deleted:
cm_cluster.remove_host(host.hostId)
api.delete_host(host.hostId)
@cpo.event_wrapper(
True, step=_("Decommission nodes"), param=('cluster', 1))
def decommission_nodes(self, cluster, process, role_names):
service = self.get_service_by_role(process, cluster)
service.decommission(*role_names).wait()
for role_name in role_names:
service.delete_role(role_name)
@cpo.event_wrapper(
True, step=_("Refresh DataNodes"), param=('cluster', 1))
def refresh_datanodes(self, cluster):
self._refresh_nodes(cluster, 'DATANODE', self.HDFS_SERVICE_NAME)
@cpo.event_wrapper(
True, step=_("Refresh YARNNodes"), param=('cluster', 1))
def refresh_yarn_nodes(self, cluster):
self._refresh_nodes(cluster, 'NODEMANAGER', self.YARN_SERVICE_NAME)
@cloudera_cmd
def _refresh_nodes(self, cluster, process, service_name):
cm_cluster = self.get_cloudera_cluster(cluster)
service = cm_cluster.get_service(service_name)
nds = [n.name for n in service.get_roles_by_type(process)]
for nd in nds:
for st in service.refresh(nd):
yield st
@cpo.event_wrapper(True, step=_("Deploy configs"), param=('cluster', 1))
@cloudera_cmd
def deploy_configs(self, cluster):
cm_cluster = self.get_cloudera_cluster(cluster)
yield cm_cluster.deploy_client_config()
def update_configs(self, instances):
# instances non-empty
cpo.add_provisioning_step(
instances[0].cluster_id, _("Update configs"), len(instances))
with context.ThreadGroup as tg:
for instance in instances:
tg.spawn("update-configs-%s" % instances.instance_name,
self._update_configs, instance)
@cpo.event_wrapper(True)
@cloudera_cmd
def _update_configs(self, instance):
for process in instance.node_group.node_processes:
process = self.pu.convert_role_showname(process)
service = self.get_service_by_role(process, instance=instance)
yield service.deploy_client_config(self.pu.get_role_name(instance,
process))
@cloudera_cmd
def restart_mgmt_service(self, cluster):
api = self.get_api_client(cluster)
cm = api.get_cloudera_manager()
mgmt_service = cm.get_service()
yield mgmt_service.restart()
@cloudera_cmd
def start_service(self, service):
yield service.start()
@cloudera_cmd
def start_roles(self, service, *role_names):
for role in service.start_roles(*role_names):
yield role
@cpo.event_wrapper(
True, step=_("Create mgmt service"), param=('cluster', 1))
def create_mgmt_service(self, cluster):
api = self.get_api_client(cluster)
cm = api.get_cloudera_manager()
setup_info = services.ApiServiceSetupInfo()
manager = self.pu.get_manager(cluster)
hostname = manager.fqdn()
processes = ['SERVICEMONITOR', 'HOSTMONITOR',
'EVENTSERVER', 'ALERTPUBLISHER']
for proc in processes:
setup_info.add_role_info(self.pu.get_role_name(manager, proc),
proc, hostname)
cm.create_mgmt_service(setup_info)
cm.hosts_start_roles([hostname])
def get_service_by_role(self, role, cluster=None, instance=None):
cm_cluster = None
if cluster:
cm_cluster = self.get_cloudera_cluster(cluster)
elif instance:
cm_cluster = self.get_cloudera_cluster(instance.cluster)
else:
raise ValueError(_("'cluster' or 'instance' argument missed"))
if role in ['NAMENODE', 'DATANODE', 'SECONDARYNAMENODE',
'HDFS_GATEWAY']:
return cm_cluster.get_service(self.HDFS_SERVICE_NAME)
elif role in ['RESOURCEMANAGER', 'NODEMANAGER', 'JOBHISTORY',
'YARN_GATEWAY']:
return cm_cluster.get_service(self.YARN_SERVICE_NAME)
elif role in ['OOZIE_SERVER']:
return cm_cluster.get_service(self.OOZIE_SERVICE_NAME)
elif role in ['HIVESERVER2', 'HIVEMETASTORE', 'WEBHCAT']:
return cm_cluster.get_service(self.HIVE_SERVICE_NAME)
elif role in ['HUE_SERVER']:
return cm_cluster.get_service(self.HUE_SERVICE_NAME)
elif role in ['SPARK_YARN_HISTORY_SERVER']:
return cm_cluster.get_service(self.SPARK_SERVICE_NAME)
elif role in ['SERVER']:
return cm_cluster.get_service(self.ZOOKEEPER_SERVICE_NAME)
elif role in ['MASTER', 'REGIONSERVER']:
return cm_cluster.get_service(self.HBASE_SERVICE_NAME)
else:
raise ValueError(
_("Process %(process)s is not supported by CDH plugin") %
{'process': role})
def _agents_connected(self, instances, api):
hostnames = [i.fqdn() for i in instances]
hostnames_to_manager = [h.hostname for h in
api.get_all_hosts('full')]
for hostname in hostnames:
if hostname not in hostnames_to_manager:
return False
return True
@cpo.event_wrapper(True, step=_("Await agents"), param=('cluster', 1))
def _await_agents(self, cluster, instances, timeout_config):
api = self.get_api_client(instances[0].cluster)
poll_utils.plugin_option_poll(
cluster, self._agents_connected, timeout_config,
_("Await Cloudera agents"), 5, {
'instances': instances, 'api': api})
def configure_instances(self, instances, cluster=None):
# instances non-empty
cpo.add_provisioning_step(
instances[0].cluster_id, _("Configure instances"), len(instances))
for inst in instances:
self.configure_instance(inst, cluster)
def get_roles_list(self, node_processes):
current = set(node_processes)
extra_roles = {
'YARN_GATEWAY': ["YARN_NODEMANAGER"],
'HDFS_GATEWAY': ['HDFS_NAMENODE', 'HDFS_DATANODE',
"HDFS_SECONDARYNAMENODE"]
}
for extra_role in six.iterkeys(extra_roles):
valid_processes = extra_roles[extra_role]
for valid in valid_processes:
if valid in current:
current.add(extra_role)
break
return list(current)
def get_role_type(self, process):
mapper = {
'YARN_GATEWAY': 'GATEWAY',
'HDFS_GATEWAY': 'GATEWAY',
}
return mapper.get(process, process)
@cpo.event_wrapper(True)
def configure_instance(self, instance, cluster=None):
roles_list = self.get_roles_list(instance.node_group.node_processes)
for role in roles_list:
self._add_role(instance, role, cluster)
def _add_role(self, instance, process, cluster):
if process in ['CLOUDERA_MANAGER', 'HDFS_JOURNALNODE']:
return
process = self.pu.convert_role_showname(process)
service = self.get_service_by_role(process, instance=instance)
role_type = self.get_role_type(process)
role = service.create_role(self.pu.get_role_name(instance, process),
role_type, instance.fqdn())
role.update_config(self._get_configs(process, cluster,
node_group=instance.node_group))
def _get_configs(self, service, cluster=None, node_group=None):
# Defined in derived class.
return