def send_info(nodemanager):
hadoop = get_hadoop_base()
yarn = YARN(hadoop)
local_hostname = hookenv.local_unit().replace('/', '-')
port = hadoop.dist_config.port('resourcemanager')
hs_http = hadoop.dist_config.port('jh_webapp_http')
hs_ipc = hadoop.dist_config.port('jobhistory')
utils.update_kv_hosts(nodemanager.hosts_map())
utils.manage_etc_hosts()
nodemanager.send_spec(hadoop.spec())
nodemanager.send_resourcemanagers([local_hostname])
nodemanager.send_ports(port, hs_http, hs_ipc)
nodemanager.send_ssh_key(utils.get_ssh_key('yarn'))
nodemanager.send_hosts_map(utils.get_kv_hosts())
slaves = nodemanager.nodes()
if data_changed('resourcemanager.slaves', slaves):
unitdata.kv().set('resourcemanager.slaves', slaves)
yarn.register_slaves(slaves)
hookenv.status_set('active', 'Ready ({count} NodeManager{s})'.format(
count=len(slaves),
s='s' if len(slaves) > 1 else '',
))
set_state('resourcemanager.ready')
def configure_yarn(resourcemanager):
hadoop = get_hadoop_base()
yarn = YARN(hadoop)
utils.update_kv_hosts(resourcemanager.hosts_map())
utils.manage_etc_hosts()
if not resourcemanager.resourcemanagers():
data = yaml.dump({
'relation_name': resourcemanager.relation_name,
'conversations': {
conv.key: dict({'relation_ids': conv.relation_ids},
**conv.serialize(conv))
for conv in resourcemanager.conversations()
},
'relation_data': {
rid: {
unit: hookenv.relation_get(unit=unit, rid=rid)
for unit in hookenv.related_units(rid)
} for rid in hookenv.relation_ids(
resourcemanager.relation_name
)
},
}, default_flow_style=False)
for line in data.splitlines():
hookenv.log(line)
yarn.configure_yarn_base(
resourcemanager.resourcemanagers()[0], resourcemanager.port(),
resourcemanager.hs_http(), resourcemanager.hs_ipc())
set_state('hadoop.yarn.configured')
def configure_hdfs(hdfs_rel):
hadoop = get_hadoop_base()
yarn = YARN(hadoop)
yarn.start_resourcemanager()
yarn.start_jobhistory()
hadoop.open_ports('resourcemanager')
set_state('resourcemanager.started')
def configure_yarn(resourcemanager):
hadoop = get_hadoop_base()
yarn = YARN(hadoop)
utils.update_kv_hosts(resourcemanager.hosts_map())
utils.manage_etc_hosts()
if not resourcemanager.resourcemanagers():
data = yaml.dump(
{
'relation_name': resourcemanager.relation_name,
'conversations': {
conv.key: dict({'relation_ids': conv.relation_ids},
**conv.serialize(conv))
for conv in resourcemanager.conversations()
},
'relation_data': {
rid: {
unit: hookenv.relation_get(unit=unit, rid=rid)
for unit in hookenv.related_units(rid)
}
for rid in hookenv.relation_ids(
resourcemanager.relation_name)
},
},
default_flow_style=False)
for line in data.splitlines():
hookenv.log(line)
yarn.configure_yarn_base(resourcemanager.resourcemanagers()[0],
resourcemanager.port(),
resourcemanager.hs_http(),
resourcemanager.hs_ipc())
set_state('hadoop.yarn.configured')
def send_info(nodemanager):
hadoop = get_hadoop_base()
yarn = YARN(hadoop)
local_hostname = hookenv.local_unit().replace('/', '-')
port = hadoop.dist_config.port('resourcemanager')
hs_http = hadoop.dist_config.port('jh_webapp_http')
hs_ipc = hadoop.dist_config.port('jobhistory')
utils.update_kv_hosts({node['ip']: node['host'] for node in nodemanager.nodes()})
utils.manage_etc_hosts()
nodemanager.send_spec(hadoop.spec())
nodemanager.send_resourcemanagers([local_hostname])
nodemanager.send_ports(port, hs_http, hs_ipc)
nodemanager.send_ssh_key(utils.get_ssh_key('hdfs'))
nodemanager.send_hosts_map(utils.get_kv_hosts())
slaves = [node['host'] for node in nodemanager.nodes()]
if data_changed('resourcemanager.slaves', slaves):
unitdata.kv().set('resourcemanager.slaves', slaves)
yarn.register_slaves(slaves)
hookenv.status_set('active', 'Ready ({count} NodeManager{s})'.format(
count=len(slaves),
s='s' if len(slaves) > 1 else '',
))
set_state('resourcemanager.ready')
def hdfs_departed():
hadoop = get_hadoop_base()
yarn = YARN(hadoop)
hadoop.close_ports('resourcemanager')
yarn.stop_jobhistory()
yarn.stop_resourcemanager()
remove_state('resourcemanager.started')
remove_state('resourcemanager.ready')
def configure_resourcemanager():
local_hostname = hookenv.local_unit().replace('/', '-')
private_address = hookenv.unit_get('private-address')
ip_addr = utils.resolve_private_address(private_address)
hadoop = get_hadoop_base()
yarn = YARN(hadoop)
yarn.configure_resourcemanager()
yarn.configure_jobhistory()
utils.update_kv_hosts({ip_addr: local_hostname})
set_state('resourcemanager.configured')
def start_nodemanager(resourcemanager):
hadoop = get_hadoop_base()
yarn = YARN(hadoop)
yarn.configure_nodemanager(
resourcemanager.resourcemanagers()[0], resourcemanager.port(),
resourcemanager.hs_http(), resourcemanager.hs_ipc())
utils.install_ssh_key('yarn', resourcemanager.ssh_key())
utils.update_kv_hosts(resourcemanager.hosts_map())
utils.manage_etc_hosts()
yarn.start_nodemanager()
hadoop.open_ports('nodemanager')
set_state('nodemanager.started')
def ganglia_changed():
hadoop = get_hadoop_base()
hdfs = HDFS(hadoop)
yarn = YARN(hadoop)
if is_state('namenode.started'):
hdfs.restart_namenode()
if is_state('datanode.started'):
hdfs.restart_datanode()
if is_state('journalnode.started'):
hdfs.restart_journalnode()
if is_state('resourcemanager.started'):
yarn.restart_resourcemanager()
if is_state('nodemanager.started'):
yarn.restart_nodemanager()
def unregister_nodemanager(nodemanager):
hadoop = get_hadoop_base()
yarn = YARN(hadoop)
slaves = unitdata.kv().get('resourcemanager.slaves', [])
slaves_leaving = nodemanager.nodes()
hookenv.log('Slaves leaving: {}'.format(slaves_leaving))
slaves_remaining = list(set(slaves) - set(slaves_leaving))
unitdata.kv().set('resourcemanager.slaves', slaves_remaining)
yarn.register_slaves(slaves_remaining)
utils.remove_kv_hosts(slaves_leaving)
utils.manage_etc_hosts()
if not slaves_remaining:
remove_state('resourcemanager.ready')
nodemanager.dismiss()
def unregister_nodemanager(nodemanager):
hadoop = get_hadoop_base()
yarn = YARN(hadoop)
nodes_leaving = nodemanager.nodes() # only returns nodes in "leaving" state
slaves = unitdata.kv().get('resourcemanager.slaves', [])
slaves_leaving = [node['host'] for node in nodes_leaving]
hookenv.log('Slaves leaving: {}'.format(slaves_leaving))
slaves_remaining = list(set(slaves) - set(slaves_leaving))
unitdata.kv().set('resourcemanager.slaves', slaves_remaining)
yarn.register_slaves(slaves_remaining)
utils.remove_kv_hosts(slaves_leaving)
utils.manage_etc_hosts()
if not slaves_remaining:
remove_state('resourcemanager.ready')
nodemanager.dismiss()
def configure_hdfs(hdfs_rel):
hadoop = get_hadoop_base()
yarn = YARN(hadoop)
yarn.start_resourcemanager()
yarn.start_jobhistory()
hadoop.open_ports('resourcemanager')
set_state('resourcemanager.started')
def hdfs_departed():
hadoop = get_hadoop_base()
yarn = YARN(hadoop)
hadoop.close_ports('resourcemanager')
yarn.stop_jobhistory()
yarn.stop_resourcemanager()
remove_state('resourcemanager.started')
remove_state('resourcemanager.ready')
def configure_resourcemanager():
local_hostname = hookenv.local_unit().replace('/', '-')
private_address = hookenv.unit_get('private-address')
ip_addr = utils.resolve_private_address(private_address)
hadoop = get_hadoop_base()
yarn = YARN(hadoop)
yarn.configure_resourcemanager()
yarn.configure_jobhistory()
utils.update_kv_hosts({ip_addr: local_hostname})
set_state('resourcemanager.configured')
def start_nodemanager(resourcemanager):
hadoop = get_hadoop_base()
yarn = YARN(hadoop)
yarn.configure_nodemanager(resourcemanager.resourcemanagers()[0],
resourcemanager.port(),
resourcemanager.hs_http(),
resourcemanager.hs_ipc())
utils.install_ssh_key('yarn', resourcemanager.ssh_key())
utils.update_kv_hosts(resourcemanager.hosts_map())
utils.manage_etc_hosts()
yarn.start_nodemanager()
hadoop.open_ports('nodemanager')
set_state('nodemanager.started')
def ganglia_changed():
hadoop = get_hadoop_base()
hdfs = HDFS(hadoop)
yarn = YARN(hadoop)
if is_state('namenode.started'):
hdfs.restart_namenode()
if is_state('datanode.started'):
hdfs.restart_datanode()
if is_state('journalnode.started'):
hdfs.restart_journalnode()
if is_state('resourcemanager.started'):
yarn.restart_resourcemanager()
if is_state('nodemanager.started'):
yarn.restart_nodemanager()
def stop_nodemanager():
hadoop = get_hadoop_base()
yarn = YARN(hadoop)
yarn.stop_nodemanager()
hadoop.close_ports('nodemanager')
remove_state('nodemanager.started')
def stop_nodemanager():
hadoop = get_hadoop_base()
yarn = YARN(hadoop)
yarn.stop_nodemanager()
hadoop.close_ports('nodemanager')
remove_state('nodemanager.started')
