def _set_config(cfg, gen_cfg, name=None):
if name in gen_cfg:
cfg.update(gen_cfg[name]["conf"])
if name is None:
for name in gen_cfg:
cfg.update(gen_cfg[name]["conf"])
return cfg
def _set_config(cfg, gen_cfg, name=None):
if name in gen_cfg:
cfg.update(gen_cfg[name]['conf'])
if name is None:
for name in gen_cfg:
cfg.update(gen_cfg[name]['conf'])
return cfg
def generate_xml_configs(configs, storage_path, nn_hostname, hadoop_port):
if hadoop_port is None:
hadoop_port = 8020
cfg = {
'fs.defaultFS': 'hdfs://%s:%s' % (nn_hostname, str(hadoop_port)),
'dfs.namenode.name.dir': extract_hadoop_path(storage_path, '/dfs/nn'),
'dfs.datanode.data.dir': extract_hadoop_path(storage_path, '/dfs/dn'),
'hadoop.tmp.dir': extract_hadoop_path(storage_path, '/dfs'),
}
# inserting user-defined configs
for key, value in extract_hadoop_xml_confs(configs):
cfg[key] = value
# invoking applied configs to appropriate xml files
core_all = CORE_DEFAULT
if CONF.enable_data_locality:
cfg.update(topology.TOPOLOGY_CONFIG)
# applying vm awareness configs
core_all += topology.vm_awareness_core_config()
xml_configs = {
'core-site': x.create_hadoop_xml(cfg, core_all),
'hdfs-site': x.create_hadoop_xml(cfg, HDFS_DEFAULT)
}
return xml_configs
def generate_xml_configs(configs, storage_path, nn_hostname, hadoop_port):
if hadoop_port is None:
hadoop_port = 8020
cfg = {
"fs.defaultFS": "hdfs://%s:%s" % (nn_hostname, str(hadoop_port)),
"dfs.namenode.name.dir": extract_hadoop_path(storage_path, "/dfs/nn"),
"dfs.datanode.data.dir": extract_hadoop_path(storage_path, "/dfs/dn"),
"hadoop.tmp.dir": extract_hadoop_path(storage_path, "/dfs"),
}
# inserting user-defined configs
for key, value in extract_hadoop_xml_confs(configs):
cfg[key] = value
# invoking applied configs to appropriate xml files
core_all = CORE_DEFAULT
if CONF.enable_data_locality:
cfg.update(topology.TOPOLOGY_CONFIG)
# applying vm awareness configs
core_all += topology.vm_awareness_core_config()
xml_configs = {"core-site": x.create_hadoop_xml(cfg, core_all), "hdfs-site": x.create_hadoop_xml(cfg, HDFS_DEFAULT)}
return xml_configs
def generate_sahara_configs(cluster, node_group=None):
nn_hostname = vu.get_instance_hostname(vu.get_namenode(cluster))
jt_hostname = vu.get_instance_hostname(vu.get_jobtracker(cluster))
oozie_hostname = vu.get_instance_hostname(vu.get_oozie(cluster))
hive_hostname = vu.get_instance_hostname(vu.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 _inject_swift_trust_info(cfg, cfg_filter, proxy_configs):
cfg = cfg.copy()
cfg.update({
swift.HADOOP_SWIFT_USERNAME: proxy_configs['proxy_username'],
swift.HADOOP_SWIFT_PASSWORD: proxy_configs['proxy_password'],
swift.HADOOP_SWIFT_TRUST_ID: proxy_configs['proxy_trust_id'],
swift.HADOOP_SWIFT_DOMAIN_NAME: CONF.proxy_user_domain_name
})
allow_swift_auth_filter = [
{'name': swift.HADOOP_SWIFT_USERNAME},
{'name': swift.HADOOP_SWIFT_PASSWORD},
{'name': swift.HADOOP_SWIFT_TRUST_ID},
{'name': swift.HADOOP_SWIFT_DOMAIN_NAME}
]
cfg_filter = cfg_filter + allow_swift_auth_filter
return cfg, cfg_filter
def generate_sahara_configs(cluster, node_group=None):
nn_hostname = vu.get_instance_hostname(vu.get_namenode(cluster))
jt_hostname = vu.get_instance_hostname(vu.get_jobtracker(cluster))
oozie_hostname = vu.get_instance_hostname(vu.get_oozie(cluster))
hive_hostname = vu.get_instance_hostname(vu.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 generate_xml_configs(configs, storage_path, nn_hostname, hadoop_port):
"""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.name.dir': storage_path + 'hdfs/name',
'dfs.data.dir': storage_path + 'hdfs/data',
'dfs.hosts': '/etc/hadoop/dn.incl',
'dfs.hosts.exclude': '/etc/hadoop/dn.excl',
"""
if hadoop_port is None:
hadoop_port = 8020
cfg = {
'fs.defaultFS': 'hdfs://%s:%s' % (nn_hostname, str(hadoop_port)),
'dfs.namenode.name.dir': extract_hadoop_path(storage_path,
'/dfs/nn'),
'dfs.datanode.data.dir': extract_hadoop_path(storage_path,
'/dfs/dn'),
'hadoop.tmp.dir': extract_hadoop_path(storage_path,
'/dfs'),
}
# inserting user-defined configs
for key, value in extract_hadoop_xml_confs(configs):
cfg[key] = value
# invoking applied configs to appropriate xml files
core_all = CORE_DEFAULT
if CONF.enable_data_locality:
cfg.update(topology.TOPOLOGY_CONFIG)
# applying vm awareness configs
core_all += topology.vm_awareness_core_config()
xml_configs = {
'core-site': x.create_hadoop_xml(cfg, core_all),
'hdfs-site': x.create_hadoop_xml(cfg, HDFS_DEFAULT)
}
return xml_configs
def _inject_swift_trust_info(cfg, cfg_filter, proxy_configs):
cfg = cfg.copy()
cfg.update({
swift.HADOOP_SWIFT_USERNAME: proxy_configs['proxy_username'],
swift.HADOOP_SWIFT_PASSWORD: proxy_configs['proxy_password'],
swift.HADOOP_SWIFT_TRUST_ID: proxy_configs['proxy_trust_id'],
swift.HADOOP_SWIFT_DOMAIN_NAME: CONF.proxy_user_domain_name
})
allow_swift_auth_filter = [{
'name': swift.HADOOP_SWIFT_USERNAME
}, {
'name': swift.HADOOP_SWIFT_PASSWORD
}, {
'name': swift.HADOOP_SWIFT_TRUST_ID
}, {
'name': swift.HADOOP_SWIFT_DOMAIN_NAME
}]
cfg_filter = cfg_filter + allow_swift_auth_filter
return cfg, cfg_filter
def generate_xml_configs(configs, storage_path, nn_hostname, jt_hostname,
oozie_hostname, hive_hostname, passwd_hive_mysql):
general_cfg = get_general_configs(hive_hostname, passwd_hive_mysql)
# 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')
# inserting user-defined configs
for key, value in extract_xml_confs(configs):
cfg[key] = value
# applying swift configs if user enabled it
swift_xml_confs = swift.get_swift_configs()
cfg = generate_cfg_from_general(cfg, 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:
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(cfg, core_all),
'mapred-site': x.create_hadoop_xml(cfg, mapred_all),
'hdfs-site': x.create_hadoop_xml(cfg, HDFS_DEFAULT)
}
if hive_hostname:
xml_configs.update(
{'hive-site': x.create_hadoop_xml(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(cfg, o_h.OOZIE_DEFAULT)})
LOG.debug('Generated oozie-site.xml for oozie % s', oozie_hostname)
return xml_configs
def generate_xml_configs(configs, storage_path, nn_hostname, jt_hostname,
oozie_hostname, hive_hostname, passwd_hive_mysql):
general_cfg = get_general_configs(hive_hostname, passwd_hive_mysql)
# 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')
# inserting user-defined configs
for key, value in extract_xml_confs(configs):
cfg[key] = value
# applying swift configs if user enabled it
swift_xml_confs = swift.get_swift_configs()
cfg = generate_cfg_from_general(cfg, 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:
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(cfg, core_all),
'mapred-site': x.create_hadoop_xml(cfg, mapred_all),
'hdfs-site': x.create_hadoop_xml(cfg, HDFS_DEFAULT)
}
if hive_hostname:
xml_configs.update({'hive-site':
x.create_hadoop_xml(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(cfg, o_h.OOZIE_DEFAULT)})
LOG.debug('Generated oozie-site.xml for oozie % s', oozie_hostname)
return xml_configs