class ContainerAppMetricsCollector(MetricsCollector):
'''
ContainerAppMetricsCollector is a metrics collector in app level,
it collects metrics via `docker exec`.
'''
def __init__(self, container_id, configs={}):
# initial backend.
self.backend = BackendFactory(configs["backend"]["name"])(
configs=configs
)
self.interval = configs["collect_interval"]
self.client = APIClient(configs)
self.configs = configs
# initial modules.
# query the CMonitor configured modules,
# and sync them into application container.
self.client.sync_modules(container_id)
super(ContainerAppMetricsCollector, self).__init__(container_id,
configs)
def execute(self):
'''
Execute the registered modules, and collect the returned metrics.
TODO: there could be some filters here.
'''
registered_modules = self.configs["registered_modules"]
data = {}
for reg_key, mod in registered_modules.iteritems():
ret = self.client.exec_module(self.container_id, mod["name"])
data[mod["name"]] = ret
return data
def collect(self):
try:
time.sleep(self.interval)
ret = self.execute()
# exec executor.
# we currently make it sync run.
# TODO: make a executor pattern to cocurrent it.
write_ok = self.backend.write(self.container_id, ret)
if write_ok:
print "Successfully collected container {} app metrics..".format(self.container_id)
else:
print "Failed to write metrics into backend, pls check backend write log for details."
except Exception, e:
import traceback
print "ERROR while collect container {} app metrics, error: {}\n\n".format(self.container_id, e)
print traceback.print_exc()
print "Error exiting current App Metrics Thread for {}".format(self.container_id)
# just exit, and main-Thread will help me spawn new one.
self.stop()
exit(-1)
