def scale_cluster(self, cluster, instances):
master = utils.get_instance(cluster, "master")
r_master = remote.get_remote(master)
run.stop_spark(r_master, self._spark_home(cluster))
self._setup_instances(cluster, instances)
nn = utils.get_instance(cluster, "namenode")
run.refresh_nodes(remote.get_remote(nn), "dfsadmin")
self._start_slave_datanode_processes(instances)
run.start_spark_master(r_master, self._spark_home(cluster))
LOG.info(_LI("Spark master service at '%s' has been restarted"), master.hostname())
def scale_cluster(self, cluster, instances):
master = utils.get_instance(cluster, "master")
r_master = remote.get_remote(master)
run.stop_spark(r_master, self._spark_home(cluster))
self._setup_instances(cluster, instances)
nn = utils.get_instance(cluster, "namenode")
run.refresh_nodes(remote.get_remote(nn), "dfsadmin")
self._start_slave_datanode_processes(instances)
run.start_spark_master(r_master, self._spark_home(cluster))
LOG.info(_LI("Spark master service at '%s' has been restarted"),
master.hostname())
def scale_cluster(self, cluster, instances):
master = utils.get_instance(cluster, "master")
r_master = remote.get_remote(master)
run.stop_spark(r_master, self._spark_home(cluster))
self._setup_instances(cluster, instances)
nn = utils.get_instance(cluster, "namenode")
run.refresh_nodes(remote.get_remote(nn), "dfsadmin")
dn_instances = [instance for instance in instances if "datanode" in instance.node_group.node_processes]
self._start_datanode_processes(dn_instances)
run.start_spark_master(r_master, self._spark_home(cluster))
LOG.info(_LI("Spark master service has been restarted"))
def scale_cluster(self, cluster, instances):
master = utils.get_instance(cluster, "master")
r_master = remote.get_remote(master)
run.stop_spark(r_master, self._spark_home(cluster))
self._setup_instances(cluster, instances)
nn = utils.get_instance(cluster, "namenode")
run.refresh_nodes(remote.get_remote(nn), "dfsadmin")
dn_instances = [instance for instance in instances if
'datanode' in instance.node_group.node_processes]
self._start_datanode_processes(dn_instances)
swift_helper.install_ssl_certs(instances)
run.start_spark_master(r_master, self._spark_home(cluster))
LOG.info("Spark master service has been restarted")
def decommission_dn(nn, inst_to_be_deleted, survived_inst):
with remote.get_remote(nn) as r:
r.write_file_to('/etc/hadoop/dn.excl',
utils.generate_fqdn_host_names(
inst_to_be_deleted))
run.refresh_nodes(remote.get_remote(nn), "dfsadmin")
context.sleep(3)
poll_utils.plugin_option_poll(
nn.cluster, _is_decommissioned, c_helper.DECOMMISSIONING_TIMEOUT,
_("Decommission %s") % "DataNodes", 3, {
'r': r, 'inst_to_be_deleted': inst_to_be_deleted})
r.write_files_to({
'/etc/hadoop/dn.incl': utils.
generate_fqdn_host_names(survived_inst),
'/etc/hadoop/dn.excl': ""})
def scale_cluster(self, cluster, instances):
master = utils.get_instance(cluster, "master")
r_master = remote.get_remote(master)
run.stop_spark(r_master, self._spark_home(cluster))
self._setup_instances(cluster, instances)
nn = utils.get_instance(cluster, "namenode")
run.refresh_nodes(remote.get_remote(nn), "dfsadmin")
dn_instances = [
instance for instance in instances
if 'datanode' in instance.node_group.node_processes
]
self._start_datanode_processes(dn_instances)
run.start_spark_master(r_master, self._spark_home(cluster))
LOG.info(
_LI("Spark master service at {host} has been restarted").format(
host=master.hostname()))
def decommission_dn(nn, inst_to_be_deleted, survived_inst):
with remote.get_remote(nn) as r:
r.write_file_to('/etc/hadoop/dn.excl',
utils.generate_fqdn_host_names(
inst_to_be_deleted))
run.refresh_nodes(remote.get_remote(nn), "dfsadmin")
context.sleep(3)
timeout = c_helper.get_decommissioning_timeout(
nn.node_group.cluster)
s_time = timeutils.utcnow()
all_found = False
while timeutils.delta_seconds(s_time, timeutils.utcnow()) < timeout:
cmd = r.execute_command(
"sudo -u hdfs hadoop dfsadmin -report")
all_found = True
datanodes_info = parse_dfs_report(cmd[1])
for i in inst_to_be_deleted:
for dn in datanodes_info:
if (dn["Name"].startswith(i.internal_ip)) and (
dn["Decommission Status"] != "Decommissioned"):
all_found = False
break
if all_found:
r.write_files_to({'/etc/hadoop/dn.incl':
utils.
generate_fqdn_host_names(survived_inst),
'/etc/hadoop/dn.excl': "",
})
break
context.sleep(3)
if not all_found:
ex.DecommissionError(
_("Cannot finish decommission of cluster %(cluster)s in "
"%(seconds)d seconds") %
{"cluster": nn.node_group.cluster,
"seconds": timeout})
