def configure_hdfs(namenode):
hadoop = get_hadoop_base()
hdfs = HDFS(hadoop)
utils.update_kv_hosts(namenode.hosts_map())
utils.manage_etc_hosts()
if not namenode.namenodes():
data = yaml.dump(
{
'relation_name': namenode.relation_name,
'conversations': {
conv.key: dict({'relation_ids': conv.relation_ids},
**conv.serialize(conv))
for conv in namenode.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(namenode.relation_name)
},
},
default_flow_style=False)
for line in data.splitlines():
hookenv.log(line)
hdfs.configure_hdfs_base(namenode.namenodes()[0], namenode.port())
set_state('hadoop.hdfs.configured')
def send_info(datanode):
hadoop = get_hadoop_base()
hdfs = HDFS(hadoop)
local_hostname = hookenv.local_unit().replace('/', '-')
hdfs_port = hadoop.dist_config.port('namenode')
webhdfs_port = hadoop.dist_config.port('nn_webapp_http')
utils.update_kv_hosts(
{node['ip']: node['host']
for node in datanode.nodes()})
utils.manage_etc_hosts()
datanode.send_spec(hadoop.spec())
datanode.send_namenodes([local_hostname])
datanode.send_ports(hdfs_port, webhdfs_port)
datanode.send_ssh_key(utils.get_ssh_key('hdfs'))
datanode.send_hosts_map(utils.get_kv_hosts())
slaves = [node['host'] for node in datanode.nodes()]
if data_changed('namenode.slaves', slaves):
unitdata.kv().set('namenode.slaves', slaves)
hdfs.register_slaves(slaves)
hookenv.status_set(
'active', 'Ready ({count} DataNode{s})'.format(
count=len(slaves),
s='s' if len(slaves) > 1 else '',
))
set_state('namenode.ready')
def configure_namenode():
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()
hdfs = HDFS(hadoop)
hdfs.configure_namenode()
hdfs.format_namenode()
hdfs.start_namenode()
hdfs.create_hdfs_dirs()
hadoop.open_ports('namenode')
utils.update_kv_hosts({ip_addr: local_hostname})
set_state('namenode.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_datanode(datanode):
hadoop = get_hadoop_base()
hdfs = HDFS(hadoop)
nodes_leaving = datanode.nodes() # only returns nodes in "leaving" state
slaves = unitdata.kv().get('namenode.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('namenode.slaves', slaves_remaining)
hdfs.register_slaves(slaves_remaining)
utils.remove_kv_hosts(slaves_leaving)
utils.manage_etc_hosts()
if not slaves_remaining:
hookenv.status_set('blocked', 'Waiting for relation to DataNodes')
remove_state('namenode.ready')
datanode.dismiss()
