def _extract_configs_to_extra(self, cluster): nn = utils.get_namenode(cluster) jt = utils.get_jobtracker(cluster) oozie = utils.get_oozie(cluster) hive = utils.get_hiveserver(cluster) extra = dict() for ng in cluster.node_groups: extra[ng.id] = { 'xml': c_helper.generate_xml_configs(ng.configuration, ng.storage_paths, nn.hostname, jt.hostname if jt else None, oozie.hostname if oozie else None, hive.hostname if hive else None), 'setup_script': c_helper.generate_setup_script( ng.storage_paths, c_helper.extract_environment_confs(ng.configuration), append_oozie=( oozie is not None and oozie.node_group.id == ng.id) ) } return extra
def generate_savanna_configs(cluster, node_group=None): nn_hostname = _get_hostname(utils.get_namenode(cluster)) jt_hostname = _get_hostname(utils.get_jobtracker(cluster)) oozie_hostname = _get_hostname(utils.get_oozie(cluster)) hive_hostname = _get_hostname(utils.get_hiveserver(cluster)) storage_path = node_group.storage_paths() if node_group else None # inserting common configs depends on provisioned VMs and HDFS placement # TODO(aignatov): should be moved to cluster context cfg = { 'fs.default.name': 'hdfs://%s:8020' % nn_hostname, 'dfs.name.dir': extract_hadoop_path(storage_path, '/lib/hadoop/hdfs/namenode'), 'dfs.data.dir': extract_hadoop_path(storage_path, '/lib/hadoop/hdfs/datanode'), 'dfs.hosts': '/etc/hadoop/dn.incl', 'dfs.hosts.exclude': '/etc/hadoop/dn.excl', } if jt_hostname: mr_cfg = { 'mapred.job.tracker': '%s:8021' % jt_hostname, 'mapred.system.dir': extract_hadoop_path(storage_path, '/mapred/mapredsystem'), 'mapred.local.dir': extract_hadoop_path(storage_path, '/lib/hadoop/mapred'), 'mapred.hosts': '/etc/hadoop/tt.incl', 'mapred.hosts.exclude': '/etc/hadoop/tt.excl', } cfg.update(mr_cfg) if oozie_hostname: o_cfg = { 'hadoop.proxyuser.hadoop.hosts': "localhost," + oozie_hostname, 'hadoop.proxyuser.hadoop.groups': 'hadoop', } cfg.update(o_cfg) LOG.debug('Applied Oozie configs for core-site.xml') cfg.update(o_h.get_oozie_required_xml_configs()) LOG.debug('Applied Oozie configs for oozie-site.xml') if hive_hostname: h_cfg = { 'hive.warehouse.subdir.inherit.perms': True, 'javax.jdo.option.ConnectionURL': 'jdbc:derby:;databaseName=/opt/hive/metastore_db;create=true' } cfg.update(h_cfg) LOG.debug('Applied Hive config for hive metastore server') return cfg
def _extract_configs_to_extra(self, cluster): oozie = utils.get_oozie(cluster) hive = utils.get_hiveserver(cluster) extra = dict() if hive: extra['hive_mysql_passwd'] = six.text_type(uuid.uuid4()) for ng in cluster.node_groups: extra[ng.id] = { 'xml': c_helper.generate_xml_configs( cluster, ng, extra['hive_mysql_passwd'] if hive else None), 'setup_script': c_helper.generate_setup_script( ng.storage_paths(), c_helper.extract_environment_confs(ng.configuration()), append_oozie=(oozie and oozie.node_group.id == ng.id)) } if c_helper.is_data_locality_enabled(cluster): topology_data = th.generate_topology_map( cluster, CONF.enable_hypervisor_awareness) extra['topology_data'] = "\n".join( [k + " " + v for k, v in topology_data.items()]) + "\n" return extra
def _extract_configs_to_extra(self, cluster): oozie = utils.get_oozie(cluster) hive = utils.get_hiveserver(cluster) extra = dict() if hive: extra['hive_mysql_passwd'] = six.text_type(uuid.uuid4()) for ng in cluster.node_groups: extra[ng.id] = { 'xml': c_helper.generate_xml_configs( cluster, ng, extra['hive_mysql_passwd'] if hive else None), 'setup_script': c_helper.generate_setup_script( ng.storage_paths(), c_helper.extract_environment_confs(ng.configuration()), append_oozie=( oozie and oozie.node_group.id == ng.id) ) } if c_helper.is_data_locality_enabled(cluster): topology_data = th.generate_topology_map( cluster, CONF.enable_hypervisor_awareness) extra['topology_data'] = "\n".join( [k + " " + v for k, v in topology_data.items()]) + "\n" return extra
def _extract_configs_to_extra(self, cluster): nn = utils.get_namenode(cluster) jt = utils.get_jobtracker(cluster) oozie = utils.get_oozie(cluster) hive = utils.get_hiveserver(cluster) extra = dict() if hive: extra['hive_mysql_passwd'] = uuidutils.generate_uuid() for ng in cluster.node_groups: extra[ng.id] = { 'xml': c_helper.generate_xml_configs( ng.configuration, ng.storage_paths, nn.hostname, jt.hostname if jt else None, oozie.hostname if oozie else None, hive.hostname if hive else None, extra['hive_mysql_passwd'] if hive else None), 'setup_script': c_helper.generate_setup_script( ng.storage_paths, c_helper.extract_environment_confs(ng.configuration), append_oozie=(oozie is not None and oozie.node_group.id == ng.id)) } if c_helper.is_data_locality_enabled(cluster): topology_data = th.generate_topology_map( cluster, CONF.enable_hypervisor_awareness) extra['topology_data'] = "\n".join( [k + " " + v for k, v in topology_data.items()]) + "\n" return extra
def _configure_services(client, cluster): nn_host = u.get_namenode(cluster).fqdn() snn = u.get_secondarynamenodes(cluster) snn_host = snn[0].fqdn() if snn else None jt_host = u.get_jobtracker(cluster).fqdn() if u.get_jobtracker( cluster) else None dn_hosts = [dn.fqdn() for dn in u.get_datanodes(cluster)] tt_hosts = [tt.fqdn() for tt in u.get_tasktrackers(cluster)] oozie_host = u.get_oozie(cluster).fqdn() if u.get_oozie(cluster) else None hive_host = u.get_hiveserver(cluster).fqdn() if u.get_hiveserver( cluster) else None services = [] if u.get_namenode(cluster): services += ['hdfs'] if u.get_jobtracker(cluster): services += ['mapred'] if oozie_host: services += ['oozie'] services += ['pig'] if hive_host: services += ['hive'] LOG.debug("Add services: %s" % ', '.join(services)) client.services.add(services) LOG.debug("Assign roles to hosts") client.services.hdfs.add_nodes('PrimaryNameNode', [nn_host]) client.services.hdfs.add_nodes('DataNode', dn_hosts) if snn: client.services.hdfs.add_nodes('SecondaryNameNode', [snn_host]) if oozie_host: client.services.oozie.add_nodes('Oozie', [oozie_host]) if hive_host: client.services.hive.add_nodes('HiveServer', [hive_host]) if jt_host: client.services.mapred.add_nodes('JobTracker', [jt_host]) client.services.mapred.add_nodes('TaskTracker', tt_hosts)
def _configure_services(client, cluster): nn_host = u.get_namenode(cluster).fqdn() snn = u.get_secondarynamenodes(cluster) snn_host = snn[0].fqdn() if snn else None jt_host = u.get_jobtracker(cluster).fqdn() dn_hosts = [dn.fqdn() for dn in u.get_datanodes(cluster)] tt_hosts = [tt.fqdn() for tt in u.get_tasktrackers(cluster)] oozie_host = u.get_oozie(cluster).fqdn() if u.get_oozie( cluster) else None hive_host = u.get_hiveserver(cluster).fqdn() if u.get_hiveserver( cluster) else None services = [] if u.get_namenode(cluster): services += ['hdfs'] if u.get_jobtracker(cluster): services += ['mapred'] if oozie_host: services += ['oozie'] services += ['pig'] if hive_host: services += ['hive'] LOG.debug("Add services: %s" % ', '.join(services)) client.services.add(services) LOG.debug("Assign roles to hosts") client.services.hdfs.add_nodes('PrimaryNameNode', [nn_host]) client.services.hdfs.add_nodes('DataNode', dn_hosts) if snn: client.services.hdfs.add_nodes('SecondaryNameNode', [snn_host]) if oozie_host: client.services.oozie.add_nodes('Oozie', [oozie_host]) if hive_host: client.services.hive.add_nodes('HiveServer', [hive_host]) client.services.mapred.add_nodes('JobTracker', [jt_host]) client.services.mapred.add_nodes('TaskTracker', tt_hosts)
def start_cluster(self, cluster): nn_instance = utils.get_namenode(cluster) datanodes = utils.get_datanodes(cluster) jt_instance = utils.get_jobtracker(cluster) tasktrackers = utils.get_tasktrackers(cluster) oozie = utils.get_oozie(cluster) hive_server = utils.get_hiveserver(cluster) with remote.get_remote(nn_instance) as r: run.format_namenode(r) run.start_process(r, "namenode") snns = utils.get_secondarynamenodes(cluster) if snns: for snn in snns: run.start_process(remote.get_remote(snn), "secondarynamenode") for dn in datanodes: run.start_process(remote.get_remote(dn), "datanode") LOG.info("HDFS service at '%s' has been started", nn_instance.hostname) if jt_instance: run.start_process(remote.get_remote(jt_instance), "jobtracker") for tt in tasktrackers: run.start_process(remote.get_remote(tt), "tasktracker") LOG.info("MapReduce service at '%s' has been started", jt_instance.hostname) if oozie: with remote.get_remote(oozie) as r: if c_helper.is_mysql_enable(cluster): run.mysql_start(r, oozie) run.oozie_create_db(r) run.oozie_share_lib(r, nn_instance.hostname) run.start_oozie(r) LOG.info("Oozie service at '%s' has been started", nn_instance.hostname) if hive_server: with remote.get_remote(nn_instance) as r: run.hive_create_warehouse_dir(r) if c_helper.is_mysql_enable(cluster): with remote.get_remote(hive_server) as h: if not oozie or hive_server.hostname != oozie.hostname: run.mysql_start(h, hive_server) run.hive_create_db(h) run.hive_metastore_start(h) LOG.info("Hive Metastore server at %s has been started", hive_server.hostname) LOG.info('Cluster %s has been started successfully' % cluster.name) self._set_cluster_info(cluster)
def start_cluster(cluster): client = c.IntelClient( u.get_instance(cluster, 'manager').management_ip, cluster.name) LOG.debug("Starting hadoop services") client.services.hdfs.start() client.services.mapred.start() if u.get_hiveserver(cluster): client.services.hive.start() if u.get_oozie(cluster): LOG.info("Setup oozie") _setup_oozie(cluster) client.services.oozie.start()
def start_cluster(self, cluster): nn_instance = utils.get_namenode(cluster) with remote.get_remote(nn_instance) as r: run.format_namenode(r) run.start_processes(r, "namenode") for snn in utils.get_secondarynamenodes(cluster): run.start_processes(remote.get_remote(snn), "secondarynamenode") jt_instance = utils.get_jobtracker(cluster) if jt_instance: run.start_processes(remote.get_remote(jt_instance), "jobtracker") self._start_tt_dn_processes(utils.get_instances(cluster)) self._await_datanodes(cluster) LOG.info("Hadoop services in cluster %s have been started" % cluster.name) oozie = utils.get_oozie(cluster) if oozie: with remote.get_remote(oozie) as r: if c_helper.is_mysql_enable(cluster): run.mysql_start(r, oozie) run.oozie_create_db(r) run.oozie_share_lib(r, nn_instance.hostname()) run.start_oozie(r) LOG.info("Oozie service at '%s' has been started", nn_instance.hostname()) hive_server = utils.get_hiveserver(cluster) if hive_server: with remote.get_remote(nn_instance) as r: run.hive_create_warehouse_dir(r) if c_helper.is_mysql_enable(cluster): with remote.get_remote(hive_server) as h: if not oozie or hive_server.hostname() != oozie.hostname(): run.mysql_start(h, hive_server) run.hive_create_db(h) run.hive_metastore_start(h) LOG.info("Hive Metastore server at %s has been started", hive_server.hostname()) LOG.info('Cluster %s has been started successfully' % cluster.name) self._set_cluster_info(cluster)
def generate_xml_configs(cluster, node_group, hive_mysql_passwd): oozie_hostname = _get_hostname(utils.get_oozie(cluster)) hive_hostname = _get_hostname(utils.get_hiveserver(cluster)) ng_configs = node_group.configuration() general_cfg = get_general_configs(hive_hostname, hive_mysql_passwd) all_cfg = generate_savanna_configs(cluster, node_group) # inserting user-defined configs for key, value in extract_xml_confs(ng_configs): all_cfg[key] = value # applying swift configs if user enabled it swift_xml_confs = swift.get_swift_configs() all_cfg = generate_cfg_from_general(all_cfg, ng_configs, general_cfg) # invoking applied configs to appropriate xml files core_all = CORE_DEFAULT + swift_xml_confs mapred_all = MAPRED_DEFAULT if CONF.enable_data_locality: all_cfg.update(topology.TOPOLOGY_CONFIG) # applying vm awareness configs core_all += topology.vm_awareness_core_config() mapred_all += topology.vm_awareness_mapred_config() xml_configs = { 'core-site': x.create_hadoop_xml(all_cfg, core_all), 'mapred-site': x.create_hadoop_xml(all_cfg, mapred_all), 'hdfs-site': x.create_hadoop_xml(all_cfg, HDFS_DEFAULT) } if hive_hostname: xml_configs.update( {'hive-site': x.create_hadoop_xml(all_cfg, HIVE_DEFAULT)}) LOG.debug('Generated hive-site.xml for hive % s', hive_hostname) if oozie_hostname: xml_configs.update( {'oozie-site': x.create_hadoop_xml(all_cfg, o_h.OOZIE_DEFAULT)}) LOG.debug('Generated oozie-site.xml for oozie % s', oozie_hostname) return xml_configs
def start_cluster(cluster): client = c.IntelClient(u.get_instance(cluster, 'manager'), cluster.name) LOG.debug("Starting hadoop services") client.services.hdfs.start() if u.get_jobtracker(cluster): client.services.mapred.start() if u.get_hiveserver(cluster): client.services.hive.start() if u.get_oozie(cluster): LOG.info("Setup oozie") _setup_oozie(cluster) client.services.oozie.start()
def start_cluster(self, cluster): instances = utils.get_instances(cluster) nn_instance = utils.get_namenode(cluster) jt_instance = utils.get_jobtracker(cluster) oozie = utils.get_oozie(cluster) hive_server = utils.get_hiveserver(cluster) with remote.get_remote(nn_instance) as r: run.format_namenode(r) run.start_processes(r, "namenode") for snn in utils.get_secondarynamenodes(cluster): run.start_processes(remote.get_remote(snn), "secondarynamenode") if jt_instance: run.start_processes(remote.get_remote(jt_instance), "jobtracker") self._start_tt_dn_processes(instances) LOG.info("Hadoop services in cluster %s have been started" % cluster.name) if oozie: with remote.get_remote(oozie) as r: if c_helper.is_mysql_enable(cluster): run.mysql_start(r, oozie) run.oozie_create_db(r) run.oozie_share_lib(r, nn_instance.hostname) run.start_oozie(r) LOG.info("Oozie service at '%s' has been started", nn_instance.hostname) if hive_server: with remote.get_remote(nn_instance) as r: run.hive_create_warehouse_dir(r) if c_helper.is_mysql_enable(cluster): with remote.get_remote(hive_server) as h: if not oozie or hive_server.hostname != oozie.hostname: run.mysql_start(h, hive_server) run.hive_create_db(h) run.hive_metastore_start(h) LOG.info("Hive Metastore server at %s has been started", hive_server.hostname) LOG.info('Cluster %s has been started successfully' % cluster.name) self._set_cluster_info(cluster)
def generate_xml_configs(cluster, node_group, hive_mysql_passwd): oozie_hostname = _get_hostname(utils.get_oozie(cluster)) hive_hostname = _get_hostname(utils.get_hiveserver(cluster)) ng_configs = node_group.configuration() general_cfg = get_general_configs(hive_hostname, hive_mysql_passwd) all_cfg = generate_savanna_configs(cluster, node_group) # inserting user-defined configs for key, value in extract_xml_confs(ng_configs): all_cfg[key] = value # applying swift configs if user enabled it swift_xml_confs = swift.get_swift_configs() all_cfg = generate_cfg_from_general(all_cfg, ng_configs, general_cfg) # invoking applied configs to appropriate xml files core_all = CORE_DEFAULT + swift_xml_confs mapred_all = MAPRED_DEFAULT if CONF.enable_data_locality: all_cfg.update(topology.TOPOLOGY_CONFIG) # applying vm awareness configs core_all += topology.vm_awareness_core_config() mapred_all += topology.vm_awareness_mapred_config() xml_configs = { 'core-site': x.create_hadoop_xml(all_cfg, core_all), 'mapred-site': x.create_hadoop_xml(all_cfg, mapred_all), 'hdfs-site': x.create_hadoop_xml(all_cfg, HDFS_DEFAULT) } if hive_hostname: xml_configs.update({'hive-site': x.create_hadoop_xml(all_cfg, HIVE_DEFAULT)}) LOG.debug('Generated hive-site.xml for hive % s', hive_hostname) if oozie_hostname: xml_configs.update({'oozie-site': x.create_hadoop_xml(all_cfg, o_h.OOZIE_DEFAULT)}) LOG.debug('Generated oozie-site.xml for oozie % s', oozie_hostname) return xml_configs
def _extract_configs_to_extra(self, cluster): nn = utils.get_namenode(cluster) jt = utils.get_jobtracker(cluster) oozie = utils.get_oozie(cluster) hive = utils.get_hiveserver(cluster) extra = dict() if hive: extra['hive_mysql_passwd'] = uuidutils.generate_uuid() for ng in cluster.node_groups: extra[ng.id] = { 'xml': c_helper.generate_xml_configs( ng.configuration, ng.storage_paths, nn.hostname, jt.hostname if jt else None, oozie.hostname if oozie else None, hive.hostname if hive else None, extra['hive_mysql_passwd'] if hive else None), 'setup_script': c_helper.generate_setup_script( ng.storage_paths, c_helper.extract_environment_confs(ng.configuration), append_oozie=( oozie is not None and oozie.node_group.id == ng.id) ) } if c_helper.is_data_locality_enabled(cluster): topology_data = th.generate_topology_map( cluster, CONF.enable_hypervisor_awareness) extra['topology_data'] = "\n".join( [k + " " + v for k, v in topology_data.items()]) + "\n" return extra
def configure_workflow_if_needed(self, cluster, wf_dir): h_s = u.get_hiveserver(cluster) plugin = plugin_base.PLUGINS.get_plugin(cluster.plugin_name) h.copy_from_local(remote.get_remote(h_s), plugin.get_hive_config_path(), wf_dir)
def upload_hive_site(cluster, wf_dir): h_s = u.get_hiveserver(cluster) plugin = plugin_base.PLUGINS.get_plugin(cluster.plugin_name) h.copy_from_local(remote.get_remote(h_s), plugin.get_hive_config_path(), wf_dir)
def _push_configs_to_nodes(self, cluster, instances=None): extra = self._extract_configs_to_extra(cluster) if instances is None: instances = utils.get_instances(cluster) for inst in instances: ng_extra = extra[inst.node_group.id] files = { '/etc/hadoop/core-site.xml': ng_extra['xml']['core-site'], '/etc/hadoop/mapred-site.xml': ng_extra['xml']['mapred-site'], '/etc/hadoop/hdfs-site.xml': ng_extra['xml']['hdfs-site'], '/tmp/savanna-hadoop-init.sh': ng_extra['setup_script'] } with remote.get_remote(inst) as r: # TODO(aignatov): sudo chown is wrong solution. But it works. r.execute_command( 'sudo chown -R $USER:$USER /etc/hadoop' ) r.execute_command( 'sudo chown -R $USER:$USER /opt/oozie/conf' ) r.write_files_to(files) r.execute_command( 'sudo chmod 0500 /tmp/savanna-hadoop-init.sh' ) r.execute_command( 'sudo /tmp/savanna-hadoop-init.sh ' '>> /tmp/savanna-hadoop-init.log 2>&1') nn = utils.get_namenode(cluster) jt = utils.get_jobtracker(cluster) with remote.get_remote(nn) as r: r.write_file_to('/etc/hadoop/dn.incl', utils. generate_fqdn_host_names( utils.get_datanodes(cluster))) if jt: with remote.get_remote(jt) as r: r.write_file_to('/etc/hadoop/tt.incl', utils. generate_fqdn_host_names( utils.get_tasktrackers(cluster))) oozie = utils.get_oozie(cluster) if oozie: with remote.get_remote(oozie) as r: r.write_file_to('/opt/oozie/conf/oozie-site.xml', extra[oozie.node_group.id] ['xml']['oozie-site']) if c_helper.is_mysql_enable(cluster): sql_script = f.get_file_text( 'plugins/vanilla/resources/create_oozie_db.sql') files = { '/tmp/create_oozie_db.sql': sql_script } remote.get_remote(oozie).write_files_to(files) hive_server = utils.get_hiveserver(cluster) if hive_server: ng_extra = extra[hive_server.node_group.id] files = { '/opt/hive/conf/hive-site.xml': ng_extra['xml']['hive-site'] } if c_helper.is_mysql_enable(cluster): sql_script = f.get_file_text( 'plugins/vanilla/resources/create_hive_db.sql' ) files.update({'/tmp/create_hive_db.sql': sql_script}) remote.get_remote(hive_server).write_files_to(files)
def configure_workflow_if_needed(self, cluster, wf_dir): h_s = u.get_hiveserver(cluster) plugin = plugin_base.PLUGINS.get_plugin(cluster.plugin_name) hdfs_user = plugin.get_hdfs_user() h.copy_from_local(remote.get_remote(h_s), plugin.get_hive_config_path(), wf_dir, hdfs_user)