def sync_spark(self):
print("Syncing Spark directory to the slaves")
path = os.path.abspath(self.spark_home)
make_parent_dir = [(make_ssh_cmd("mkdir -p %s" % path, s), True) for s in self.slaves]
run_cmds_parallel(make_parent_dir)
copy_spark = [(make_rsync_cmd(path, s), True) for s in self.slaves]
run_cmds_parallel(copy_spark)
def sync_spark(self):
print("Syncing Spark directory to the slaves")
path = os.path.abspath(self.spark_home)
make_parent_dir = [(make_ssh_cmd("mkdir -p %s" % path, s), True)
for s in self.slaves]
run_cmds_parallel(make_parent_dir)
copy_spark = [(make_rsync_cmd(path, s), True) for s in self.slaves]
run_cmds_parallel(copy_spark)
def warmup_disks(self, bytes_to_write, disk_warmup_files):
"""
Warm up local disks (this is only necessary on EC2).
:param bytes_to_write: Total number of bytes used to warm up each local directory.
:param disk_warmup_files: Number of files to create when warming up each local directory.
Bytes will be evenly divided across files.
"""
# Search for 'spark.local.dir' in spark-env.sh.
path_to_env_file = "%s/spark-env.sh" % self.spark_conf_dir
env_file_content = open(path_to_env_file, "r").read()
re_result = re.search(r"SPARK_LOCAL_DIRS=(.*)", env_file_content)
if re_result:
spark_local_dirs = re_result.group(1).split(",")
else:
err_msg = (
"ERROR: These scripts require you to explicitly set SPARK_LOCAL_DIRS "
"in spark-env.sh so that it can be cleaned. The way we check this is pretty "
"picky, specifically we try to find the following string in spark-env.sh: "
'SPARK_LOCAL_DIRS=ONE_OR_MORE_DIRNAMES" so you will want a line like this: '
)
sys.exit(err_msg)
for local_dir in spark_local_dirs:
# Strip off any trailing whitespace(s) so that the clear commands can work properly:
local_dir = local_dir.rstrip()
bytes_per_file = bytes_to_write / disk_warmup_files
gen_command = "dd if=/dev/urandom bs=%s count=1 | split -a 5 -b %s - %s/random" % (
bytes_to_write,
bytes_per_file,
local_dir,
)
# Ensures the directory exists.
dir_command = "mkdir -p %s" % local_dir
print ("Generating test data for %s, this may take some time" % local_dir)
all_hosts = self.slaves + ["localhost"]
run_cmds_parallel([(make_ssh_cmd(dir_command, host), True) for host in all_hosts])
run_cmds_parallel([(make_ssh_cmd(gen_command, host), True) for host in all_hosts])
clear_dir(local_dir, all_hosts, prompt_for_deletes=False)
def warmup_disks(self, bytes_to_write, disk_warmup_files):
"""
Warm up local disks (this is only necessary on EC2).
:param bytes_to_write: Total number of bytes used to warm up each local directory.
:param disk_warmup_files: Number of files to create when warming up each local directory.
Bytes will be evenly divided across files.
"""
# Search for 'spark.local.dir' in spark-env.sh.
path_to_env_file = "%s/spark-env.sh" % self.spark_conf_dir
env_file_content = open(path_to_env_file, 'r').read()
re_result = re.search(r'SPARK_LOCAL_DIRS=(.*)', env_file_content)
if re_result:
spark_local_dirs = re_result.group(1).split(",")
else:
err_msg = \
("ERROR: These scripts require you to explicitly set SPARK_LOCAL_DIRS "
"in spark-env.sh so that it can be cleaned. The way we check this is pretty "
"picky, specifically we try to find the following string in spark-env.sh: "
"SPARK_LOCAL_DIRS=ONE_OR_MORE_DIRNAMES\" so you will want a line like this: ")
sys.exit(err_msg)
for local_dir in spark_local_dirs:
# Strip off any trailing whitespace(s) so that the clear commands can work properly:
local_dir = local_dir.rstrip()
bytes_per_file = bytes_to_write / disk_warmup_files
gen_command = "dd if=/dev/urandom bs=%s count=1 | split -a 5 -b %s - %s/random" % (
bytes_to_write, bytes_per_file, local_dir)
# Ensures the directory exists.
dir_command = "mkdir -p %s" % local_dir
print("Generating test data for %s, this may take some time" %
local_dir)
all_hosts = self.slaves + ["localhost"]
run_cmds_parallel([(make_ssh_cmd(dir_command, host), True)
for host in all_hosts])
run_cmds_parallel([(make_ssh_cmd(gen_command, host), True)
for host in all_hosts])
clear_dir(local_dir, all_hosts, prompt_for_deletes=False)
def ensure_spark_stopped_on_slaves(self):
"""
Ensures that no executors are running on Spark slaves. Executors can continue to run for some
time after a shutdown signal is given due to cleaning up temporary files.
"""
stop = False
while not stop:
cmd = "ps -ef | grep -v grep | grep ExecutorBackend"
ret_vals = run_cmds_parallel([(make_ssh_cmd(cmd, s), False) for s in self.slaves])
if 0 in ret_vals:
print "Spark is still running on some slaves ... sleeping for 10 seconds"
time.sleep(10)
else:
stop = True
# Allow some extra time for slaves to fully terminate.
time.sleep(5)
def ensure_spark_stopped_on_slaves(self):
"""
Ensures that no executors are running on Spark slaves. Executors can continue to run for some
time after a shutdown signal is given due to cleaning up temporary files.
"""
stop = False
while not stop:
cmd = "ps -ef | grep -v grep | grep ExecutorBackend"
ret_vals = run_cmds_parallel([(make_ssh_cmd(cmd, s), False)
for s in self.slaves])
if 0 in ret_vals:
print "Spark is still running on some slaves ... sleeping for 10 seconds"
time.sleep(10)
else:
stop = True
# Allow some extra time for slaves to fully terminate.
time.sleep(5)
