def handle_process(self, task_id, process_id):
all_processes = {}
current_run = self._observer.process(task_id, process_id)
if not current_run:
HttpServer.abort(404, 'Invalid task/process combination: %s/%s' % (task_id, process_id))
process = self._observer.process_from_name(task_id, process_id)
if process is None:
msg = 'Could not recover process: %s/%s' % (task_id, process_id)
log.error(msg)
HttpServer.abort(404, msg)
current_run_number = current_run['process_run']
all_processes[current_run_number] = current_run
for run in range(current_run_number):
all_processes[run] = self._observer.process(task_id, process_id, run)
template = {
'task_id': task_id,
'process': {
'name': process_id,
'status': all_processes[current_run_number]["state"],
'cmdline': process.cmdline().get()
},
}
template['process'].update(**all_processes[current_run_number].get('used', {}))
template['runs'] = all_processes
log.debug('Rendering template is: %s' % template)
return template
def handle_process(self, task_id, process_id):
all_processes = {}
current_run = self._observer.process(task_id, process_id)
if not current_run:
HttpServer.abort(
404, 'Invalid task/process combination: %s/%s' %
(task_id, process_id))
process = self._observer.process_from_name(task_id, process_id)
if process is None:
msg = 'Could not recover process: %s/%s' % (task_id, process_id)
log.error(msg)
HttpServer.abort(404, msg)
current_run_number = current_run['process_run']
all_processes[current_run_number] = current_run
for run in range(current_run_number):
all_processes[run] = self._observer.process(
task_id, process_id, run)
template = {
'task_id': task_id,
'process': {
'name': process_id,
'status': all_processes[current_run_number]["state"],
'cmdline': process.cmdline().get()
},
}
template['process'].update(
**all_processes[current_run_number].get('used', {}))
template['runs'] = all_processes
log.debug('Rendering template is: %s', template)
return template
def __init__(
self, iface, zkport, request_handler, reply_handler=None, event_handler=None):
HttpServer.__init__(self)
config = SnifferConfig(iface=iface)
config.zookeeper_port = zkport
config.update_filter()
self._sniffer = Sniffer(config, request_handler, reply_handler, event_handler)
self._sniffer.start()
def handle_vars(self, var=None):
HttpServer.set_content_type('text/plain; charset=iso-8859-1')
samples = self._monitor.sample()
if var is None:
return '\n'.join(
'%s %s' % (key, val) for key, val in sorted(samples.items()))
else:
if var in samples:
return samples[var]
else:
HttpServer.abort(404, 'Unknown exported variable')
def handle_vars(self, var=None):
HttpServer.set_content_type('text/plain; charset=iso-8859-1')
samples = self._monitor.sample()
if var is None:
return '\n'.join('%s %s' % (key, val)
for key, val in sorted(samples.items()))
else:
if var in samples:
return samples[var]
else:
HttpServer.abort(404, 'Unknown exported variable')
def handle_vars(self, var = None):
samples = self._monitor.sample()
if var is None:
body='<br>'.join(
'%s %s' % (key, val) for key, val in samples.iteritems())
return '<html><body><pre>%s</pre></body></html>' % body
else:
if var in samples:
return samples[var]
else:
HttpServer.abort(404, 'Unknown exported variable')
def handle_vars(self, var=None):
samples = self._monitor.sample()
if var is None:
body = '<br>'.join('%s %s' % (key, val)
for key, val in samples.items())
return '<html><body><pre>%s</pre></body></html>' % body
else:
if var in samples:
return samples[var]
else:
HttpServer.abort(404, 'Unknown exported variable')
def handle_vars(self, var=None):
HttpServer.set_content_type('text/plain; charset=iso-8859-1')
filtered = self._parse_filtered_arg()
samples = self._monitor.sample()
if var is None and filtered and self._stats_filter:
return '\n'.join('%s %s' % (key, val)
for key, val in sorted(samples.items())
if not self._stats_filter.match(key))
elif var is None:
return '\n'.join('%s %s' % (key, val)
for key, val in sorted(samples.items()))
else:
if var in samples:
return samples[var]
else:
HttpServer.abort(404, 'Unknown exported variable')
def handle_vars(self, var=None):
HttpServer.set_content_type('text/plain; charset=iso-8859-1')
filtered = self._parse_filtered_arg()
samples = self._monitor.sample()
if var is None and filtered and self._stats_filter:
return '\n'.join(
'%s %s' % (key, val) for key, val in sorted(samples.items())
if not self._stats_filter.match(key))
elif var is None:
return '\n'.join(
'%s %s' % (key, val) for key, val in sorted(samples.items()))
else:
if var in samples:
return samples[var]
else:
HttpServer.abort(404, 'Unknown exported variable')
def configure_server(task_observer):
bottle_wrapper = BottleObserver(task_observer)
root_metrics = RootMetrics()
server = HttpServer()
server.mount_routes(bottle_wrapper)
server.mount_routes(DiagnosticsEndpoints())
server.mount_routes(VarsEndpoint())
register_build_properties(root_metrics)
register_diagnostics(root_metrics)
return server
def handle_task(self, task_id):
task = self.get_task(task_id)
processes = self._observer.processes([task_id])
if not processes.get(task_id, None):
HttpServer.abort(404, "Unknown task_id: %s" % task_id)
processes = processes[task_id]
state = self._observer.state(task_id)
return dict(
task_id=task_id,
task=task,
statuses=self._observer.task_statuses(task_id),
user=task["user"],
ports=task["ports"],
processes=processes,
chroot=state.get("sandbox", ""),
launch_time=state.get("launch_time", 0),
hostname=state.get("hostname", "localhost"),
)
def handle_task(self, task_id):
task = self.get_task(task_id)
processes = self._observer.processes([task_id])
if not processes.get(task_id, None):
HttpServer.abort(404, 'Unknown task_id: %s' % task_id)
processes = processes[task_id]
state = self._observer.state(task_id)
return dict(
task_id=task_id,
task=task,
statuses=self._observer.task_statuses(task_id),
user=task['user'],
ports=task['ports'],
processes=processes,
chroot=state.get('sandbox', ''),
launch_time=state.get('launch_time', 0),
hostname=state.get('hostname', 'localhost'),
)
def handle_task(self, task_id):
task = self.get_task(task_id)
processes = self._observer.processes([task_id])
if not processes.get(task_id, None):
HttpServer.abort(404, 'Unknown task_id: %s' % task_id)
processes = processes[task_id]
state = self._observer.state(task_id)
return dict(
task_id = task_id,
task = task,
statuses = self._observer.task_statuses(task_id),
user = task['user'],
ports = task['ports'],
processes = processes,
chroot = state.get('sandbox', ''),
launch_time = state.get('launch_time', 0),
hostname = state.get('hostname', 'localhost'),
)
def handle_process(self, task_id, process_id):
all_processes = {}
current_run = self._observer.process(task_id, process_id)
if not current_run:
HttpServer.abort(
404, 'Invalid task/process combination: %s/%s' %
(task_id, process_id))
process = self._observer.process_from_name(task_id, process_id)
if process is None:
msg = 'Could not recover process: %s/%s' % (task_id, process_id)
log.error(msg)
HttpServer.abort(404, msg)
current_run_number = current_run['process_run']
all_processes[current_run_number] = current_run
for run in range(current_run_number):
all_processes[run] = self._observer.process(
task_id, process_id, run)
def convert_process_tuple(run_tuple):
process_tuple = dict(state=run_tuple['state'])
if 'start_time' in run_tuple:
process_tuple.update(start_time=run_tuple['start_time'])
if 'stop_time' in run_tuple:
process_tuple.update(stop_time=run_tuple['stop_time'])
return process_tuple
template = {
'task_id': task_id,
'process': {
'name': process_id,
'status': all_processes[current_run_number]["state"],
'cmdline': process.cmdline().get()
},
}
template['process'].update(
**all_processes[current_run_number].get('used', {}))
template['runs'] = dict((run, convert_process_tuple(run_tuple))
for run, run_tuple in all_processes.items())
log.info('Rendering template is: %s' % template)
return template
def configure_server(task_observer): bottle_wrapper = BottleObserver(task_observer) root_metrics = RootMetrics() server = HttpServer() server.mount_routes(bottle_wrapper) server.mount_routes(DiagnosticsEndpoints()) server.mount_routes(VarsEndpoint()) register_build_properties(root_metrics) register_diagnostics(root_metrics) return server
def test_basic_server_error_binding():
BREAKAGE = '*****breakage*****'
class MyServer(object):
@HttpServer.route('/broken')
def broken_handler(self):
raise Exception('unhandled exception!')
@HttpServer.error(404)
@HttpServer.error(500)
def error_handler(self, error):
return BREAKAGE
server = HttpServer()
mserver = MyServer()
server.mount_routes(mserver)
# Test 404 error handling.
resp = server.app(make_request('/nonexistent_page'),
functools.partial(response_asserter, 404))
assert resp[0] == BREAKAGE
# Test 500 error handling.
resp = server.app(make_request('/broken'),
functools.partial(response_asserter, 500))
assert resp[0] == BREAKAGE
def test_basic_server_error_binding():
BREAKAGE = '*****breakage*****'
class MyServer(object):
@HttpServer.route('/broken')
def broken_handler(self):
raise Exception('unhandled exception!')
@HttpServer.error(404)
@HttpServer.error(500)
def error_handler(self, error):
return BREAKAGE
server = HttpServer()
mserver = MyServer()
server.mount_routes(mserver)
# Test 404 error handling.
resp = server.app(make_request('/nonexistent_page'), functools.partial(response_asserter, 404))
assert resp[0] == BREAKAGE
# Test 500 error handling.
resp = server.app(make_request('/broken'), functools.partial(response_asserter, 500))
assert resp[0] == BREAKAGE
def main(args, opts):
if args:
print("ERROR: unrecognized arguments: %s\n" % (" ".join(args)), file=sys.stderr)
app.help()
sys.exit(1)
root_server = HttpServer()
root_server.mount_routes(DiagnosticsEndpoints())
task_observer = TaskObserver(opts.root)
task_observer.start()
bottle_wrapper = BottleObserver(task_observer)
root_server.mount_routes(bottle_wrapper)
def run():
root_server.run('0.0.0.0', opts.port, 'cherrypy')
et = ExceptionalThread(target=run)
et.daemon = True
et.start()
et.join()
def handle_process(self, task_id, process_id):
all_processes = {}
current_run = self._observer.process(task_id, process_id)
if not current_run:
HttpServer.abort(404, "Invalid task/process combination: %s/%s" % (task_id, process_id))
process = self._observer.process_from_name(task_id, process_id)
if process is None:
msg = "Could not recover process: %s/%s" % (task_id, process_id)
log.error(msg)
HttpServer.abort(404, msg)
current_run_number = current_run["process_run"]
all_processes[current_run_number] = current_run
for run in range(current_run_number):
all_processes[run] = self._observer.process(task_id, process_id, run)
def convert_process_tuple(run_tuple):
process_tuple = dict(state=run_tuple["state"])
if "start_time" in run_tuple:
process_tuple.update(start_time=run_tuple["start_time"])
if "stop_time" in run_tuple:
process_tuple.update(stop_time=run_tuple["stop_time"])
return process_tuple
template = {
"task_id": task_id,
"process": {
"name": process_id,
"status": all_processes[current_run_number]["state"],
"cmdline": process.cmdline().get(),
},
}
template["process"].update(**all_processes[current_run_number].get("used", {}))
template["runs"] = dict((run, convert_process_tuple(run_tuple)) for run, run_tuple in all_processes.items())
log.info("Rendering template is: %s" % template)
return template
def handle_process(self, task_id, process_id):
all_processes = {}
current_run = self._observer.process(task_id, process_id)
if not current_run:
HttpServer.abort(404, 'Invalid task/process combination: %s/%s' % (task_id, process_id))
process = self._observer.process_from_name(task_id, process_id)
if process is None:
msg = 'Could not recover process: %s/%s' % (task_id, process_id)
log.error(msg)
HttpServer.abort(404, msg)
current_run_number = current_run['process_run']
all_processes[current_run_number] = current_run
for run in range(current_run_number):
all_processes[run] = self._observer.process(task_id, process_id, run)
def convert_process_tuple(run_tuple):
process_tuple = dict(state = run_tuple['state'])
if 'start_time' in run_tuple:
process_tuple.update(start_time = run_tuple['start_time'])
if 'stop_time' in run_tuple:
process_tuple.update(stop_time = run_tuple['stop_time'])
return process_tuple
template = {
'task_id': task_id,
'process': {
'name': process_id,
'status': all_processes[current_run_number]["state"],
'cmdline': process.cmdline().get()
},
}
template['process'].update(**all_processes[current_run_number].get('used', {}))
template['runs'] = dict((run, convert_process_tuple(run_tuple))
for run, run_tuple in all_processes.items())
log.info('Rendering template is: %s' % template)
return template
def handle_main(self, type=None, offset=None, num=None):
if type not in (None, 'all', 'finished', 'active'):
HttpServer.abort(404, 'Invalid task type: %s' % type)
if offset is not None:
try:
offset = int(offset)
except ValueError:
HttpServer.abort(404, 'Invalid offset: %s' % offset)
if num is not None:
try:
num = int(num)
except ValueError:
HttpServer.abort(404, 'Invalid count: %s' % num)
return self._observer.main(type, offset, num)
def handle_main(self, type=None, offset=None, num=None):
if type not in (None, 'all', 'finished', 'active'):
HttpServer.abort(404, 'Invalid task type: %s' % type)
if offset is not None:
try:
offset = int(offset)
except ValueError:
HttpServer.abort(404, 'Invalid offset: %s' % offset)
if num is not None:
try:
num = int(num)
except ValueError:
HttpServer.abort(404, 'Invalid count: %s' % num)
return self._observer.main(type, offset, num)
def main(args, opts):
if args:
print("ERROR: unrecognized arguments: %s\n" % (" ".join(args)),
file=sys.stderr)
app.help()
sys.exit(1)
root_server = HttpServer()
root_server.mount_routes(DiagnosticsEndpoints())
task_observer = TaskObserver(opts.root)
task_observer.start()
bottle_wrapper = BottleObserver(task_observer)
root_server.mount_routes(bottle_wrapper)
def run():
root_server.run('0.0.0.0', opts.port, 'cherrypy')
et = ExceptionalThread(target=run)
et.daemon = True
et.start()
et.join()
def main(args, options):
log.info("Options in use: %s", options)
if not options.api_port:
app.error('Must specify --port')
if not options.mesos_master:
app.error('Must specify --mesos_master')
if not options.framework_user:
app.error('Must specify --framework_user')
if not options.executor_uri:
app.error('Must specify --executor_uri')
if not options.executor_cmd:
app.error('Must specify --executor_cmd')
if not options.zk_url:
app.error('Must specify --zk_url')
if not options.admin_keypath:
app.error('Must specify --admin_keypath')
try:
election_timeout = parse_time(options.election_timeout)
framework_failover_timeout = parse_time(
options.framework_failover_timeout)
except InvalidTime as e:
app.error(e.message)
try:
_, zk_servers, zk_root = zookeeper.parse(options.zk_url)
except Exception as e:
app.error("Invalid --zk_url: %s" % e.message)
web_assets_dir = os.path.join(options.work_dir, "web")
pkgutil.unpack_assets(web_assets_dir, MYSOS_MODULE, ASSET_RELPATH)
log.info("Extracted web assets into %s" % options.work_dir)
fw_principal = None
fw_secret = None
if options.framework_authentication_file:
try:
with open(options.framework_authentication_file, "r") as f:
cred = yaml.load(f)
fw_principal = cred["principal"]
fw_secret = cred["secret"]
log.info(
"Loaded credential (principal=%s) for framework authentication"
% fw_principal)
except IOError as e:
app.error(
"Unable to read the framework authentication key file: %s"
% e)
except (KeyError, yaml.YAMLError) as e:
app.error(
"Invalid framework authentication key file format %s" % e)
log.info("Starting Mysos scheduler")
kazoo = KazooClient(zk_servers)
kazoo.start()
if options.state_storage == 'zk':
log.info("Using ZooKeeper (path: %s) for state storage" % zk_root)
state_provider = ZooKeeperStateProvider(kazoo, zk_root)
else:
log.info("Using local disk for state storage")
state_provider = LocalStateProvider(options.work_dir)
try:
state = state_provider.load_scheduler_state()
except StateProvider.Error as e:
app.error(e.message)
if state:
log.info("Successfully restored scheduler state")
framework_info = state.framework_info
if framework_info.HasField('id'):
log.info("Recovered scheduler's FrameworkID is %s" %
framework_info.id.value)
else:
log.info("No scheduler state to restore")
framework_info = FrameworkInfo(
user=options.framework_user,
name=FRAMEWORK_NAME,
checkpoint=True,
failover_timeout=framework_failover_timeout.as_(Time.SECONDS),
role=options.framework_role)
if fw_principal:
framework_info.principal = fw_principal
state = Scheduler(framework_info)
state_provider.dump_scheduler_state(state)
scheduler = MysosScheduler(state,
state_provider,
options.framework_user,
options.executor_uri,
options.executor_cmd,
kazoo,
options.zk_url,
election_timeout,
options.admin_keypath,
installer_args=options.installer_args,
backup_store_args=options.backup_store_args,
executor_environ=options.executor_environ,
framework_role=options.framework_role)
if fw_principal and fw_secret:
cred = Credential(principal=fw_principal, secret=fw_secret)
scheduler_driver = mesos.native.MesosSchedulerDriver(
scheduler, framework_info, options.mesos_master, cred)
else:
scheduler_driver = mesos.native.MesosSchedulerDriver(
scheduler, framework_info, options.mesos_master)
scheduler_driver.start()
server = HttpServer()
server.mount_routes(MysosServer(scheduler, web_assets_dir))
et = ExceptionalThread(target=server.run,
args=('0.0.0.0', options.api_port, 'cherrypy'))
et.daemon = True
et.start()
try:
# Wait for the scheduler to stop.
# The use of 'stopped' event instead of scheduler_driver.join() is necessary to stop the
# process with SIGINT.
while not scheduler.stopped.wait(timeout=0.5):
pass
except KeyboardInterrupt:
log.info('Interrupted, exiting.')
else:
log.info('Scheduler exited.')
app.shutdown(
1
) # Mysos scheduler is supposed to be long-running thus the use of exit status 1.
def get_task(self, task_id):
task = self._observer._task(task_id)
if not task:
HttpServer.abort(
404, "Failed to find task %s. Try again shortly." % task_id)
return task
def __init__(self):
HttpServer.__init__(self)
def __init__(self, observer):
self._observer = observer
StaticAssets.__init__(self)
TaskObserverFileBrowser.__init__(self)
TaskObserverJSONBindings.__init__(self)
HttpServer.__init__(self)
def __init__(self):
self._name = BaseServer.NAME
HttpServer.__init__(self)
def __init__(self): self._name = BaseServer.NAME HttpServer.__init__(self)
def __init__(self): HttpServer.__init__(self)
def __init__(self, observer): self._observer = observer StaticAssets.__init__(self) TaskObserverFileBrowser.__init__(self) TaskObserverJSONBindings.__init__(self) HttpServer.__init__(self)
def main(args, options):
log.info("Options in use: %s", options)
if not options.api_port:
app.error('Must specify --port')
if not options.mesos_master:
app.error('Must specify --mesos_master')
if not options.framework_user:
app.error('Must specify --framework_user')
if not options.executor_uri:
app.error('Must specify --executor_uri')
if not options.executor_cmd:
app.error('Must specify --executor_cmd')
if not options.zk_url:
app.error('Must specify --zk_url')
if not options.admin_keypath:
app.error('Must specify --admin_keypath')
try:
election_timeout = parse_time(options.election_timeout)
framework_failover_timeout = parse_time(options.framework_failover_timeout)
except InvalidTime as e:
app.error(e.message)
try:
_, zk_servers, zk_root = zookeeper.parse(options.zk_url)
except Exception as e:
app.error("Invalid --zk_url: %s" % e.message)
web_assets_dir = os.path.join(options.work_dir, "web")
pkgutil.unpack_assets(web_assets_dir, MYSOS_MODULE, ASSET_RELPATH)
log.info("Extracted web assets into %s" % options.work_dir)
fw_principal = None
fw_secret = None
if options.framework_authentication_file:
try:
with open(options.framework_authentication_file, "r") as f:
cred = yaml.load(f)
fw_principal = cred["principal"]
fw_secret = cred["secret"]
log.info("Loaded credential (principal=%s) for framework authentication" % fw_principal)
except IOError as e:
app.error("Unable to read the framework authentication key file: %s" % e)
except (KeyError, yaml.YAMLError) as e:
app.error("Invalid framework authentication key file format %s" % e)
log.info("Starting Mysos scheduler")
kazoo = KazooClient(zk_servers)
kazoo.start()
if options.state_storage == 'zk':
log.info("Using ZooKeeper (path: %s) for state storage" % zk_root)
state_provider = ZooKeeperStateProvider(kazoo, zk_root)
else:
log.info("Using local disk for state storage")
state_provider = LocalStateProvider(options.work_dir)
try:
state = state_provider.load_scheduler_state()
except StateProvider.Error as e:
app.error(e.message)
if state:
log.info("Successfully restored scheduler state")
framework_info = state.framework_info
if framework_info.HasField('id'):
log.info("Recovered scheduler's FrameworkID is %s" % framework_info.id.value)
else:
log.info("No scheduler state to restore")
framework_info = FrameworkInfo(
user=options.framework_user,
name=FRAMEWORK_NAME,
checkpoint=True,
failover_timeout=framework_failover_timeout.as_(Time.SECONDS),
role=options.framework_role)
if fw_principal:
framework_info.principal = fw_principal
state = Scheduler(framework_info)
state_provider.dump_scheduler_state(state)
scheduler = MysosScheduler(
state,
state_provider,
options.framework_user,
options.executor_uri,
options.executor_cmd,
kazoo,
options.zk_url,
election_timeout,
options.admin_keypath,
installer_args=options.installer_args,
backup_store_args=options.backup_store_args,
executor_environ=options.executor_environ,
framework_role=options.framework_role)
if fw_principal and fw_secret:
cred = Credential(principal=fw_principal, secret=fw_secret)
scheduler_driver = mesos.native.MesosSchedulerDriver(
scheduler,
framework_info,
options.mesos_master,
cred)
else:
scheduler_driver = mesos.native.MesosSchedulerDriver(
scheduler,
framework_info,
options.mesos_master)
scheduler_driver.start()
server = HttpServer()
server.mount_routes(MysosServer(scheduler, web_assets_dir))
et = ExceptionalThread(
target=server.run, args=('0.0.0.0', options.api_port, 'cherrypy'))
et.daemon = True
et.start()
try:
# Wait for the scheduler to stop.
# The use of 'stopped' event instead of scheduler_driver.join() is necessary to stop the
# process with SIGINT.
while not scheduler.stopped.wait(timeout=0.5):
pass
except KeyboardInterrupt:
log.info('Interrupted, exiting.')
else:
log.info('Scheduler exited.')
app.shutdown(1) # Mysos scheduler is supposed to be long-running thus the use of exit status 1.
def get_task(self, task_id):
task = self._observer._task(task_id)
if not task:
HttpServer.abort(404, "Failed to find task %s. Try again shortly." % task_id)
return task
def main(args, options):
tracker = Tracker()
server = HttpServer()
server.mount_routes(tracker)
server.run('0.0.0.0', port=options.port, server='tornado')