def start(self):
"""Starts the process on the MaxiNet node and returns the remote PID.
Returns:
True if the process started successfully, False otherwise.
"""
self.start_time = time.time()
# start command as daemonized process
logger.debug("Daemonizing \"%s\" on MaxiNet node %s" % (self._command,
self._node.nn))
result = self._node.cmd("%s %s %s" % (
transport_api.TransportAPI.get_binary_path("daemonize"),
configuration.get_worker_working_directory(),
self._command))
pid = result.splitlines()[0]
# daemonize is supposed to print ONLY the PID of the started process
if not pid.isdigit():
logger.error("Failed to start process:\n{}".format(
utils.indent(result, 2)))
return False
self.pid = int(pid)
return True
def start(self):
"""Starts the process on the MaxiNet node and returns the remote PID.
Returns:
True if the process started successfully, False otherwise.
"""
self.start_time = time.time()
# start command as daemonized process
logger.debug("Daemonizing \"%s\" on MaxiNet node %s" %
(self._command, self._node.nn))
result = self._node.cmd(
"%s %s %s" %
(transport_api.TransportAPI.get_binary_path("daemonize"),
configuration.get_worker_working_directory(), self._command))
pid = result.splitlines()[0]
# daemonize is supposed to print ONLY the PID of the started process
if not pid.isdigit():
logger.error("Failed to start process:\n{}".format(
utils.indent(result, 2)))
return False
self.pid = int(pid)
return True
def _get_remote_transport_bin_path(cls):
"""Returns path to transport binaries on worker nodes.
Returns:
Path to transport binaries on worker nodes.
"""
return os.path.join(configuration.get_worker_working_directory(),
"transport_bin")
def init(self, nodes):
"""Initialize the transport API.
Instantiate the process manager and all specified MaxiNet nodes.
Args:
nodes: List of MaxiNet nodes the transport API is used on.
"""
logger.debug("Initializing transport API")
# Copy binaries to workers
for worker in {node.worker for node in nodes}:
ssh_tools.copy_to_worker(
worker, os.path.abspath("./transport_bin"),
configuration.get_worker_working_directory())
self._process_manager.start()
self._nodes = nodes
for node in self._nodes:
self._init_node(node)
def run(self):
self.__stop.clear()
for worker in network_emulator.NetworkEmulator.get_instance(
).cluster.worker:
self.__running_processes[worker] = dict()
while not self.__stop.isSet():
for worker in network_emulator.NetworkEmulator.get_instance(
).cluster.worker:
successful_processes = []
try:
successful_processes = self.__worker_get_pids_from_file(
worker,
os.path.join(
configuration.get_worker_working_directory(),
"pids_successful"))
logger.debug("Successful processes {!s}".format(
successful_processes))
except subprocess.CalledProcessError:
# This possible, if file pids_successful does not yet exist
pass
failed_processes = []
try:
failed_processes = self.__worker_get_pids_from_file(
worker,
os.path.join(
configuration.get_worker_working_directory(),
"pids_failed"))
logger.debug(
"Failed processes {!s}".format(failed_processes))
except subprocess.CalledProcessError:
# This possible, if file pids_failed does not yet exist
pass
# For every failed process retrieve and print processes's output from worker
for pid in failed_processes:
try:
cat_cmd = "cat {1}".format(
worker.hn(),
os.path.join(
configuration.get_worker_working_directory(),
"processes", str(pid)))
logfile_content = ssh_tools.worker_ssh(worker, cat_cmd)
logfile_formatted = utils.indent(logfile_content, 2)
logger.error(
"Process with PID {0} failed:\n{1}".format(
pid, logfile_formatted))
except subprocess.CalledProcessError, err:
logger.error(
"Failed to retrieve logfile for process with PID %i"
% pid)
# Not allowed, as every daemonized process writes to a logfile
raise err
# post-process successful and failed processes
with self.__running_processes_lock:
# all successful transmissions
for pid in successful_processes:
if pid in self.__running_processes[worker]:
self.__running_processes[worker][
pid].call_terminated(
process.Process.SUCCESSFUL)
del self.__running_processes[worker][pid]
else:
logger.error(
"PID of successful transmission not found")
# all unsuccessful transmissions
for pid in failed_processes:
if pid in self.__running_processes[worker]:
self.__running_processes[worker][
pid].call_terminated(process.Process.FAILED)
del self.__running_processes[worker][pid]
time.sleep(self.__interval)
def run(self):
self.__stop.clear()
for worker in network_emulator.NetworkEmulator.get_instance().cluster.worker:
self.__running_processes[worker] = dict()
while not self.__stop.isSet():
for worker in network_emulator.NetworkEmulator.get_instance().cluster.worker:
successful_processes = []
try:
successful_processes = self.__worker_get_pids_from_file(
worker,
os.path.join(configuration.get_worker_working_directory(), "pids_successful"))
logger.debug("Successful processes {!s}".format(successful_processes))
except subprocess.CalledProcessError:
# This possible, if file pids_successful does not yet exist
pass
failed_processes = []
try:
failed_processes = self.__worker_get_pids_from_file(
worker,
os.path.join(configuration.get_worker_working_directory(), "pids_failed"))
logger.debug("Failed processes {!s}".format(failed_processes))
except subprocess.CalledProcessError:
# This possible, if file pids_failed does not yet exist
pass
# For every failed process retrieve and print processes's output from worker
for pid in failed_processes:
try:
cat_cmd = "cat {1}".format(
worker.hn(),
os.path.join(configuration.get_worker_working_directory(),
"processes", str(pid)))
logfile_content = ssh_tools.worker_ssh(worker, cat_cmd)
logfile_formatted = utils.indent(logfile_content, 2)
logger.error("Process with PID {0} failed:\n{1}".format(
pid, logfile_formatted))
except subprocess.CalledProcessError, err:
logger.error("Failed to retrieve logfile for process with PID %i" % pid)
# Not allowed, as every daemonized process writes to a logfile
raise err
# post-process successful and failed processes
with self.__running_processes_lock:
# all successful transmissions
for pid in successful_processes:
if pid in self.__running_processes[worker]:
self.__running_processes[worker][pid].call_terminated(
process.Process.SUCCESSFUL)
del self.__running_processes[worker][pid]
else:
logger.error("PID of successful transmission not found")
# all unsuccessful transmissions
for pid in failed_processes:
if pid in self.__running_processes[worker]:
self.__running_processes[worker][pid].call_terminated(
process.Process.FAILED)
del self.__running_processes[worker][pid]
time.sleep(self.__interval)
