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 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 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(datanode.hosts_map())
    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 = datanode.nodes()
    if data_changed('namenode.slaves', slaves):
        unitdata.kv().set('namenode.slaves', slaves)
        hdfs.register_slaves(slaves)
        hdfs.refresh_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 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 stop_datanode():
    hadoop = get_hadoop_base()
    hdfs = HDFS(hadoop)
    hdfs.stop_datanode()
    hdfs.stop_journalnode()
    hadoop.close_ports('datanode')
    remove_state('datanode.started')
Esempio n. 6
0
 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 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_datanode(namenode):
    hadoop = get_hadoop_base()
    hdfs = HDFS(hadoop)
    hdfs.configure_datanode(namenode.namenodes()[0], namenode.port())
    utils.install_ssh_key('hdfs', namenode.ssh_key())
    utils.update_kv_hosts(namenode.hosts_map())
    utils.manage_etc_hosts()
    hdfs.start_datanode()
    hadoop.open_ports('datanode')
    set_state('datanode.started')
def accept_clients(clients):
    hadoop = get_hadoop_base()
    local_hostname = hookenv.local_unit().replace("/", "-")
    hdfs_port = hadoop.dist_config.port("namenode")
    webhdfs_port = hadoop.dist_config.port("nn_webapp_http")

    clients.send_spec(hadoop.spec())
    clients.send_namenodes([local_hostname])
    clients.send_ports(hdfs_port, webhdfs_port)
    clients.send_hosts_map(utils.get_kv_hosts())
    clients.send_ready(True)
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 accept_clients(clients):
    hadoop = get_hadoop_base()
    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')

    clients.send_spec(hadoop.spec())
    clients.send_resourcemanagers([local_hostname])
    clients.send_ports(port, hs_http, hs_ipc)
    clients.send_hosts_map(utils.get_kv_hosts())
    clients.send_ready(True)
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 configure_ha(cluster, datanode):
    hadoop = get_hadoop_base()
    hdfs = HDFS(hadoop)
    cluster_nodes = cluster.nodes()
    jn_nodes = datanode.nodes()
    jn_port = datanode.jn_port()
    if data_changed('namenode.ha', [cluster_nodes, jn_nodes, jn_port]):
        utils.update_kv_hosts(cluster.hosts_map())
        utils.manage_etc_hosts()
        hdfs.register_journalnodes(jn_nodes, jn_port)
        hdfs.restart_namenode()
        datanode.send_namenodes(cluster_nodes)
        if not is_state('namenode.shared-edits.init'):
            hdfs.init_sharededits()
            set_state('namenode.shared-edits.init')
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 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()
Esempio n. 17
0
 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 stop_nodemanager():
    hadoop = get_hadoop_base()
    yarn = YARN(hadoop)
    yarn.stop_nodemanager()
    hadoop.close_ports('nodemanager')
    remove_state('nodemanager.started')
Esempio n. 19
0
def handle_legacy_installed_flag():
    hadoop = get_hadoop_base()
    if hadoop.is_installed():
        set_state('hadoop.installed')
Esempio n. 20
0
def fetch_resources():
    hadoop = get_hadoop_base()
    if hadoop.verify_resources():
        set_state('resources.available')
Esempio n. 21
0
def install_hadoop():
    hadoop = get_hadoop_base()
    hadoop.install()
    set_state('hadoop.installed')
Esempio n. 22
0
 def set_hdfs_spec(namenode):
     hadoop = get_hadoop_base()
     namenode.set_local_spec(hadoop.spec())
Esempio n. 23
0
 def set_yarn_spec(resourcemanager):
     hadoop = get_hadoop_base()
     resourcemanager.set_local_spec(hadoop.spec())