def _copy_base_conf(self):
"""Copy base configuration files to tmp dir."""
self.temp_conf_dir = tempfile.mkdtemp("", "hadoop-", "/tmp")
if os.path.exists(self.local_base_conf_dir):
base_conf_files = [
os.path.join(self.local_base_conf_dir, f)
for f in os.listdir(self.local_base_conf_dir)
]
for f in base_conf_files:
shutil.copy(f, self.temp_conf_dir)
else:
logger.warn(
"Local conf dir does not exist. Using default configuration")
base_conf_files = []
mandatory_files = [
CORE_CONF_FILE, HDFS_CONF_FILE, MR_CONF_FILE, YARN_CONF_FILE
]
missing_conf_files = mandatory_files
for f in base_conf_files:
f_base_name = os.path.basename(f)
if f_base_name in missing_conf_files:
missing_conf_files.remove(f_base_name)
logger.info("Copying missing conf files from master: " +
str(missing_conf_files))
remote_missing_files = [
os.path.join(self.conf_dir, f) for f in missing_conf_files
]
action = Get([self.master], remote_missing_files, self.temp_conf_dir)
action.run()
def _copy_base_conf(self):
"""Copy base configuration files to tmp dir."""
self.temp_conf_dir = tempfile.mkdtemp("", "hadoop-", "/tmp")
if os.path.exists(self.local_base_conf_dir):
base_conf_files = [os.path.join(self.local_base_conf_dir, f)
for f in os.listdir(self.local_base_conf_dir)]
for f in base_conf_files:
shutil.copy(f, self.temp_conf_dir)
else:
logger.warn(
"Local conf dir does not exist. Using default configuration")
base_conf_files = []
mandatory_files = [CORE_CONF_FILE, HDFS_CONF_FILE, MR_CONF_FILE,
YARN_CONF_FILE]
missing_conf_files = mandatory_files
for f in base_conf_files:
f_base_name = os.path.basename(f)
if f_base_name in missing_conf_files:
missing_conf_files.remove(f_base_name)
logger.info("Copying missing conf files from master: " + str(
missing_conf_files))
remote_missing_files = [os.path.join(self.conf_dir, f)
for f in missing_conf_files]
action = Get([self.master], remote_missing_files, self.temp_conf_dir)
action.run()
def copy_history(self, dest, job_ids=None):
"""Copy history logs from dfs.
Args:
dest (str):
The path of the local dir where the logs will be copied.
job_ids (list of str, optional):
A list with the ids of the jobs for which the history should be
copied. If nothing is passed, the history of all jobs is copied.
"""
if not os.path.exists(dest):
logger.warning("Destination directory " + dest +
" does not exist. It will be created")
os.makedirs(dest)
# Dirs used
user_login = getpass.getuser()
hist_dfs_dir = "/tmp/hadoop-yarn/staging/history/done_intermediate/" + \
user_login
hist_tmp_dir = "/tmp/hadoop_hist"
# Remove file in tmp dir if exists
proc = SshProcess("rm -rf " + hist_tmp_dir, self.master)
proc.run()
# Get files in master
if job_ids:
proc = SshProcess("mkdir " + hist_tmp_dir, self.master)
proc.run()
for jid in job_ids:
self.execute("fs -get " + hist_dfs_dir + "/" + jid + "* " +
hist_tmp_dir,
verbose=False)
else:
self.execute("fs -get " + hist_dfs_dir + " " + hist_tmp_dir,
verbose=False)
# Copy files from master
action = Get([self.master], [hist_tmp_dir], dest)
action.run()
def copy_history(self, dest, job_ids=None):
"""Copy history logs from dfs.
Args:
dest (str):
The path of the local dir where the logs will be copied.
job_ids (list of str, optional):
A list with the ids of the jobs for which the history should be
copied. If nothing is passed, the history of all jobs is copied.
"""
if not os.path.exists(dest):
logger.warning("Destination directory " + dest +
" does not exist. It will be created")
os.makedirs(dest)
# Dirs used
user_login = getpass.getuser()
hist_dfs_dir = "/tmp/hadoop-yarn/staging/history/done_intermediate/" + \
user_login
hist_tmp_dir = "/tmp/hadoop_hist"
# Remove file in tmp dir if exists
proc = SshProcess("rm -rf " + hist_tmp_dir, self.master)
proc.run()
# Get files in master
if job_ids:
proc = SshProcess("mkdir " + hist_tmp_dir, self.master)
proc.run()
for jid in job_ids:
self.execute("fs -get " + hist_dfs_dir + "/" + jid + "* " +
hist_tmp_dir, verbose=False)
else:
self.execute("fs -get " + hist_dfs_dir + " " + hist_tmp_dir,
verbose=False)
# Copy files from master
action = Get([self.master], [hist_tmp_dir], dest)
action.run()
def _initialize_conf(self):
"""Merge locally-specified configuration files with default files
from the distribution."""
if os.path.exists(self.local_base_conf_dir):
base_conf_files = [os.path.join(self.local_base_conf_dir, f) for f in os.listdir(self.local_base_conf_dir)]
for f in base_conf_files:
shutil.copy(f, self.init_conf_dir)
else:
logger.warn("Local conf dir does not exist. Using default configuration")
base_conf_files = []
missing_conf_files = self.conf_mandatory_files
for f in base_conf_files:
f_base_name = os.path.basename(f)
if f_base_name in missing_conf_files:
missing_conf_files.remove(f_base_name)
logger.info("Copying missing conf files from master: " + str(missing_conf_files))
remote_missing_files = [os.path.join(self.conf_dir, f) for f in missing_conf_files]
action = Get([self.hosts[0]], remote_missing_files, self.init_conf_dir)
action.run()
