def prepare_dataset(self, comb):
"""Prepare the dataset to be used in the next set of experiments.
Args:
comb (dict): The combination containing the dataset's parameters.
Returns:
dict: The dataset parameters.
"""
# Create ds_comb
(ds_class_name, ds_params) = self.comb_manager.get_ds_class_params(comb)
local_path = ds_params["local_path"]
remote_path = os.path.join(self.div_p2p.remote_dir,
os.path.basename(local_path))
ds_comb = {"ds.class.path": remote_path, "ds.class": ds_class_name}
# Copy dataset to host
logger.info(self._th_prefix() + "Prepare dataset with combination " +
str(self.comb_manager.get_ds_parameters(comb)))
copy_code = TaktukPut([self.div_p2p.host], [local_path], remote_path)
copy_code.run()
# Notify stats manager
self.stats_manager.add_ds(self.ds_id, comb)
return ds_comb
def _copy_conf(self, conf_dir, hosts=None):
if not hosts:
hosts = self.hosts
conf_files = [os.path.join(conf_dir, f) for f in os.listdir(conf_dir)]
action = TaktukPut(hosts, conf_files, self.conf_dir)
action.run()
if not action.finished_ok:
logger.warn("Error while copying configuration")
if not action.ended:
action.kill()
def setup(self):
"""Setup the cluster of hosts. Optionally deploy env and then copy the
executable jar to all the nodes.
"""
self.hosts = get_oar_job_nodes(self.oar_job_id, self.frontend)
if self.use_kadeploy:
(deployed, undeployed) = self.deploy_nodes()
return (len(deployed) != 0)
copy_code = TaktukPut(self.hosts, [self.jar_file], self.remote_dir)
copy_code.run()
return True
def bootstrap(self, tar_file):
# 0. Check that required packages are present
required_packages = "openjdk-7-jre openjdk-7-jdk"
check_packages = TaktukRemote("dpkg -s " + required_packages,
self.hosts)
for p in check_packages.processes:
p.nolog_exit_code = p.nolog_error = True
check_packages.run()
if not check_packages.ok:
logger.info("Packages not installed, trying to install")
install_packages = TaktukRemote(
"export DEBIAN_MASTER=noninteractive ; " +
"apt-get update && apt-get install -y --force-yes " +
required_packages, self.hosts).run()
if not install_packages.ok:
logger.error("Unable to install the packages")
get_java_home = SshProcess(
'echo $(readlink -f /usr/bin/javac | '
'sed "s:/bin/javac::")', self.master)
get_java_home.run()
self.java_home = get_java_home.stdout.strip()
logger.info("All required packages are present")
# 1. Copy Hive tar file and uncompress
logger.info("Copy " + tar_file + " to hosts and uncompress")
rm_dirs = TaktukRemote(
"rm -rf " + self.base_dir + " " + self.conf_dir + " " +
self.warehouse_dir + " " + self.logs_dir, self.hosts)
put_tar = TaktukPut(self.hosts, [tar_file], "/tmp")
tar_xf = TaktukRemote(
"tar xf /tmp/" + os.path.basename(tar_file) + " -C /tmp",
self.hosts)
SequentialActions([rm_dirs, put_tar, tar_xf]).run()
# 2. Move installation to base dir
logger.info("Create installation directories")
mv_base_dir = TaktukRemote(
"mv /tmp/" + os.path.basename(tar_file).replace(".tar.gz", "") +
" " + self.base_dir, self.hosts)
mkdirs = TaktukRemote(
"mkdir -p " + self.conf_dir + " && mkdir -p " + self.warehouse_dir,
self.hosts)
chmods = TaktukRemote(
"chmod g+w " + self.base_dir + " && chmod g+w " + self.conf_dir +
" && chmod g+w " + self.warehouse_dir, self.hosts)
SequentialActions([mv_base_dir, mkdirs, chmods]).run()
# 3. Specify environment variables
command = "cat >> " + self.conf_dir + "/hive-env.sh << EOF\n"
command += "JAVA_HOME=" + self.java_home + "\n"
command += "HIVE_HOME=" + self.base_dir + "\n"
command += "HIVE_CONF_DIR=" + self.conf_dir + "\n"
command += "HADOOP_HOME=" + self.hc.base_dir + "\n"
command += "EOF\n"
command += "chmod +x " + self.conf_dir + "/hive-env.sh"
action = Remote(command, self.hosts)
action.run()
def _copy_conf(self, conf_dir, hosts=None):
"""Copy configuration files from given dir to remote dir in cluster
hosts.
Args:
conf_dir (str):
The remote configuration dir.
hosts (list of Host, optional):
The list of hosts where the configuration is going to be copied. If
not specified, all the hosts of the Hadoop cluster are used.
"""
if not hosts:
hosts = self.hosts
conf_files = [os.path.join(conf_dir, f) for f in os.listdir(conf_dir)]
action = TaktukPut(hosts, conf_files, self.conf_dir)
action.run()
if not action.finished_ok:
logger.warn("Error while copying configuration")
if not action.ended:
action.kill()
def _copy_conf(self, conf_dir, hosts=None):
"""Copy configuration files from given dir to remote dir in cluster
hosts.
Args:
conf_dir (str):
The remote configuration dir.
hosts (list of Host, optional):
The list of hosts where the configuration is going to be copied. If
not specified, all the hosts of the Spark cluster are used.
"""
if not hosts:
hosts = self.hosts
conf_files = [os.path.join(conf_dir, f) for f in os.listdir(conf_dir)]
action = TaktukPut(hosts, conf_files, self.conf_dir)
action.run()
if not action.finished_ok:
logger.warn("Error while copying configuration")
if not action.ended:
action.kill()
def bootstrap(self, tar_file):
# 0. Check requirements
java_major_version = 7
if not check_java_version(java_major_version, self.hosts):
msg = "Java 1.%d+ required" % java_major_version
logger.error(msg)
raise SparkException(msg)
self.java_home = get_java_home(self.master)
# 1. Copy hadoop tar file and uncompress
logger.info("Copy " + tar_file + " to hosts and uncompress")
rm_dirs = TaktukRemote("rm -rf " + self.base_dir + " " + self.conf_dir,
self.hosts)
put_tar = TaktukPut(self.hosts, [tar_file], "/tmp")
tar_xf = TaktukRemote(
"tar xf /tmp/" + os.path.basename(tar_file) + " -C /tmp",
self.hosts)
rm_tar = TaktukRemote("rm /tmp/" + os.path.basename(tar_file),
self.hosts)
SequentialActions([rm_dirs, put_tar, tar_xf, rm_tar]).run()
# 2. Move installation to base dir
logger.info("Create installation directories")
mv_base_dir = TaktukRemote(
"mv /tmp/" + os.path.basename(tar_file).replace(".tgz", "") + " " +
self.base_dir, self.hosts)
mkdirs = TaktukRemote(
"mkdir -p " + self.conf_dir + " && mkdir -p " + self.logs_dir,
self.hosts)
chmods = TaktukRemote(
"chmod g+w " + self.base_dir + " && chmod g+w " + self.conf_dir +
" && chmod g+w " + self.logs_dir, self.hosts)
SequentialActions([mv_base_dir, mkdirs, chmods]).run()
# 2.1. Create spark-events dir
if self.evs_log_dir:
if self.evs_log_dir.startswith("file://") or \
"://" not in self.evs_log_dir:
mk_evs_dir = TaktukRemote(
"mkdir -p " + self.evs_log_dir + " && chmod g+w " +
self.evs_log_dir, self.hosts)
mk_evs_dir.run()
elif self.evs_log_dir.startswith("hdfs://"):
self.hc.execute("fs -mkdir -p " + self.evs_log_dir)
# 3. Specify environment variables
env_file = self.conf_dir + "/spark-env.sh"
command = "cat >> " + env_file + " << EOF\n"
command += "JAVA_HOME=" + self.java_home + "\n"
command += "SPARK_LOG_DIR=" + self.logs_dir + "\n"
if self.hc:
command += "HADOOP_CONF_DIR=" + self.hc.conf_dir + "\n"
if self.mode == YARN_MODE:
command += "YARN_CONF_DIR=" + self.hc.conf_dir + "\n"
command += "EOF\n"
command += "echo SPARK_PUBLIC_DNS=$(hostname) >> " + env_file
command += " && chmod +x " + env_file
action = Remote(command, self.hosts)
action.run()
# 4. Generate initial configuration
self._initialize_conf()
def bootstrap(self, tar_file):
"""Install Hadoop in all cluster nodes from the specified tar.gz file.
Args:
tar_file (str):
The file containing Hadoop binaries.
"""
# 0. Check that required packages are present
required_packages = "openjdk-7-jre openjdk-7-jdk"
check_packages = TaktukRemote("dpkg -s " + required_packages,
self.hosts)
for p in check_packages.processes:
p.nolog_exit_code = p.nolog_error = True
check_packages.run()
if not check_packages.ok:
logger.info("Packages not installed, trying to install")
install_packages = TaktukRemote(
"export DEBIAN_MASTER=noninteractive ; " +
"apt-get update && apt-get install -y --force-yes " +
required_packages, self.hosts).run()
if not install_packages.ok:
logger.error("Unable to install the packages")
get_java_home = SshProcess('echo $(readlink -f /usr/bin/javac | '
'sed "s:/bin/javac::")', self.master)
get_java_home.run()
self.java_home = get_java_home.stdout.strip()
logger.info("All required packages are present")
# 1. Copy hadoop tar file and uncompress
logger.info("Copy " + tar_file + " to hosts and uncompress")
rm_dirs = Remote("rm -rf " + self.base_dir +
" " + self.conf_dir +
" " + self.logs_dir +
" " + self.hadoop_temp_dir,
self.hosts)
put_tar = TaktukPut(self.hosts, [tar_file], "/tmp")
tar_xf = TaktukRemote(
"tar xf /tmp/" + os.path.basename(tar_file) + " -C /tmp",
self.hosts)
SequentialActions([rm_dirs, put_tar, tar_xf]).run()
# 2. Move installation to base dir and create other dirs
logger.info("Create installation directories")
mv_base_dir = TaktukRemote(
"mv /tmp/" +
os.path.basename(tar_file).replace(".tar.gz", "") + " " +
self.base_dir,
self.hosts)
mkdirs = TaktukRemote("mkdir -p " + self.conf_dir +
" && mkdir -p " + self.logs_dir +
" && mkdir -p " + self.hadoop_temp_dir,
self.hosts)
chmods = TaktukRemote("chmod g+w " + self.base_dir +
" && chmod g+w " + self.conf_dir +
" && chmod g+w " + self.logs_dir +
" && chmod g+w " + self.hadoop_temp_dir,
self.hosts)
SequentialActions([mv_base_dir, mkdirs, chmods]).run()
# 4. Specify environment variables
command = "cat >> " + self.conf_dir + "/hadoop-env.sh << EOF\n"
command += "export JAVA_HOME=" + self.java_home + "\n"
command += "export HADOOP_LOG_DIR=" + self.logs_dir + "\n"
command += "HADOOP_HOME_WARN_SUPPRESS=\"TRUE\"\n"
command += "EOF"
action = Remote(command, self.hosts)
action.run()
# 5. Check version
return self._check_version_compliance()
def bootstrap(self, tar_file):
"""Install Hadoop in all cluster nodes from the specified tar.gz file.
Args:
tar_file (str):
The file containing Hadoop binaries.
"""
# 0. Check requirements
java_major_version = 7
if not check_java_version(java_major_version, self.hosts):
msg = "Java 1.%d+ required" % java_major_version
logger.error(msg)
raise HadoopException(msg)
self.java_home = get_java_home(self.master)
# 1. Copy hadoop tar file and uncompress
logger.info("Copy " + tar_file + " to hosts and uncompress")
rm_dirs = TaktukRemote(
"rm -rf " + self.base_dir + " " + self.conf_dir + " " +
self.logs_dir + " " + self.hadoop_temp_dir, self.hosts)
put_tar = TaktukPut(self.hosts, [tar_file], "/tmp")
tar_xf = TaktukRemote(
"tar xf /tmp/" + os.path.basename(tar_file) + " -C /tmp",
self.hosts)
rm_tar = TaktukRemote("rm /tmp/" + os.path.basename(tar_file),
self.hosts)
SequentialActions([rm_dirs, put_tar, tar_xf, rm_tar]).run()
# 2. Move installation to base dir and create other dirs
logger.info("Create installation directories")
mv_base_dir = TaktukRemote(
"mv /tmp/" + os.path.basename(tar_file).replace(".tar.gz", "") +
" " + self.base_dir, self.hosts)
mkdirs = TaktukRemote(
"mkdir -p " + self.conf_dir + " && mkdir -p " + self.logs_dir +
" && mkdir -p " + self.hadoop_temp_dir, self.hosts)
chmods = TaktukRemote(
"chmod g+w " + self.base_dir + " && chmod g+w " + self.conf_dir +
" && chmod g+w " + self.logs_dir + " && chmod g+w " +
self.hadoop_temp_dir, self.hosts)
SequentialActions([mv_base_dir, mkdirs, chmods]).run()
# 4. Specify environment variables
command = "cat >> " + self.conf_dir + "/hadoop-env.sh << EOF\n"
command += "export JAVA_HOME=" + self.java_home + "\n"
command += "export HADOOP_LOG_DIR=" + self.logs_dir + "\n"
command += "HADOOP_HOME_WARN_SUPPRESS=\"TRUE\"\n"
command += "EOF"
action = Remote(command, self.hosts)
action.run()
# 5. Check version (cannot do it before)
if not self._check_version_compliance():
return False
# 6. Generate initial configuration
self._initialize_conf()
return True
