def test_get_host_status(self):
"""
Verify that KuberContainerManager().test_get_host_status() works as expected.
"""
kube_container_mgr = KubeContainerManager(CONFIG)
status_struct = {'status': {'use': 'kube'}}
for test_data in (
(200, status_struct, False),
(200, status_struct['status'], True)):
kube_container_mgr.con.get = MagicMock(return_value=MagicMock(
status_code=test_data[0],
json=MagicMock(return_value=test_data[1])))
status_code, data = kube_container_mgr.get_host_status('10.2.0.2')
self.assertEquals(test_data[0], status_code)
self.assertEquals(test_data[1], data)
def test_node_registered(self):
"""
Verify that KubeContainerManager().node_registered() works as expected.
"""
kube_container_mgr = KubeContainerManager(CONFIG)
# First call should return True. The rest should be False.
kube_container_mgr.con = MagicMock()
kube_container_mgr.con.get = MagicMock(side_effect=(
MagicMock(status_code=200),
MagicMock(status_code=404),
MagicMock(status_code=500)))
self.assertTrue(kube_container_mgr.node_registered('test'))
self.assertFalse(kube_container_mgr.node_registered('test'))
self.assertFalse(kube_container_mgr.node_registered('test'))
def test_host_status_retrieve_with_container_manager(self):
"""
Verify retrieving Host status when it is in a container manager cluster.
"""
with mock.patch('cherrypy.engine.publish') as _publish:
manager = mock.MagicMock(StoreHandlerManager)
kube_container_mgr = KubeContainerManager({
'server_url': 'http://127.0.0.1:8080',
'token': 'token'
})
# A dummy requests.Response
response = requests.Response()
response.status_code = 200
response._content='{"use": "kube"}'
kube_container_mgr._get = mock.MagicMock(return_value=response)
manager.list_container_managers.return_value = [kube_container_mgr]
_publish.return_value = [manager]
test_host = make_new(HOST)
test_cluster = make_new(CLUSTER)
test_cluster.type = C.CLUSTER_TYPE_KUBERNETES
test_cluster.hostset = [test_host.address]
# Verify if the host exists the data is returned
manager.get.side_effect = (
test_host,
test_cluster)
manager.list.return_value = Clusters.new(clusters=[test_cluster])
body = self.simulate_request('/api/v0/host/10.2.0.2/status')
self.assertEqual(self.srmock.status, falcon.HTTP_200)
result = json.loads(body[0])
self.assertEquals(C.CLUSTER_TYPE_KUBERNETES, result['type'])
self.assertEquals('available', result['host']['status'])
self.assertEquals({'use': 'kube'}, result['container_manager'])
def test_node_registered(self):
"""
Verify that KuberContainerManager().node_registered() works as expected.
"""
config = Config(
etcd={
'uri': urlparse('http://127.0.0.1:2379'),
},
kubernetes={
'uri': urlparse('http://127.0.0.1:8080'),
'token': 'token',
}
)
kube_container_mgr = KubeContainerManager(config)
# First call should return True. The rest should be False.
kube_container_mgr.con = MagicMock()
kube_container_mgr.con.get = MagicMock(side_effect=(
MagicMock(status_code=200),
MagicMock(status_code=404),
MagicMock(status_code=500)))
self.assertTrue(kube_container_mgr.node_registered('test'))
self.assertFalse(kube_container_mgr.node_registered('test'))
self.assertFalse(kube_container_mgr.node_registered('test'))
def investigator(queue, config, run_once=False):
"""
Investigates new hosts to retrieve and store facts.
:param queue: Queue to pull work from.
:type queue: Queue.Queue
:param config: Configuration information.
:type config: commissaire.config.Config
"""
logger = logging.getLogger('investigator')
logger.info('Investigator started')
while True:
# Statuses follow:
# http://commissaire.readthedocs.org/en/latest/enums.html#host-statuses
transport = ansibleapi.Transport()
to_investigate, ssh_priv_key = queue.get()
address = to_investigate['address']
logger.info('{0} is now in investigating.'.format(address))
logger.debug('Investigation details: key={0}, data={1}'.format(
to_investigate, ssh_priv_key))
f = tempfile.NamedTemporaryFile(prefix='key', delete=False)
key_file = f.name
logger.debug('Using {0} as the temporary key location for {1}'.format(
key_file, address))
f.write(base64.decodestring(ssh_priv_key))
logger.info('Wrote key for {0}'.format(address))
f.close()
key = '/commissaire/hosts/{0}'.format(address)
etcd_resp, error = cherrypy.engine.publish('store-get', key)[0]
if error:
logger.warn('Unable to continue for {0} due to '
'{1}: {2}. Returning...'.format(
address, type(error), error))
clean_up_key(key_file)
continue
data = json.loads(etcd_resp.value)
try:
result, facts = transport.get_info(address, key_file)
data.update(facts)
data['last_check'] = datetime.datetime.utcnow().isoformat()
data['status'] = 'bootstrapping'
logger.info('Facts for {0} retrieved'.format(address))
logger.debug('Data: {0}'.format(data))
except:
exc_type, exc_msg, tb = sys.exc_info()
logger.warn('Getting info failed for {0}: {1}'.format(
address, exc_msg))
data['status'] = 'failed'
cherrypy.engine.publish('store-save', key, json.dumps(data))[0]
clean_up_key(key_file)
if run_once:
break
continue
cherrypy.engine.publish('store-save', key, json.dumps(data))[0]
logger.info(
'Finished and stored investigation data for {0}'.format(address))
logger.debug('Finished investigation update for {0}: {1}'.format(
address, data))
logger.info('{0} is now in bootstrapping'.format(address))
oscmd = get_oscmd(data['os'])
try:
result, facts = transport.bootstrap(address, key_file, config,
oscmd)
data['status'] = 'inactive'
cherrypy.engine.publish('store-save', key, json.dumps(data))[0]
except:
exc_type, exc_msg, tb = sys.exc_info()
logger.warn('Unable to start bootstraping for {0}: {1}'.format(
address, exc_msg))
data['status'] = 'disassociated'
cherrypy.engine.publish('store-save', key, json.dumps(data))[0]
clean_up_key(key_file)
if run_once:
break
continue
# Verify association with the container manager
try:
container_mgr = KubeContainerManager(config)
# Try 3 times waiting 5 seconds each time before giving up
for cnt in range(0, 3):
if container_mgr.node_registered(address):
logger.info(
'{0} has been registered with the container manager.')
data['status'] = 'active'
break
if cnt == 3:
msg = 'Could not register with the container manager'
logger.warn(msg)
raise Exception(msg)
logger.debug(
'{0} has not been registered with the container manager. '
'Checking again in 5 seconds...'.format(address))
sleep(5)
except:
_, exc_msg, _ = sys.exc_info()
logger.warn(
'Unable to finish bootstrap for {0} while associating with '
'the container manager: {1}'.format(address, exc_msg))
data['status'] = 'inactive'
cherrypy.engine.publish('store-save', key, json.dumps(data))[0]
logger.info('Finished bootstrapping for {0}'.format(address))
logging.debug('Finished bootstrapping for {0}: {1}'.format(
address, data))
clean_up_key(key_file)
if run_once:
logger.info('Exiting due to run_once request.')
break
logger.info('Investigator stopping')
def investigator(queue, config, store_kwargs={}, run_once=False):
"""
Investigates new hosts to retrieve and store facts.
:param queue: Queue to pull work from.
:type queue: Queue.Queue
:param config: Configuration information.
:type config: commissaire.config.Config
:param store_kwargs: Keyword arguments used to make the etcd client.
:type store_kwargs: dict
"""
logger = logging.getLogger('investigator')
logger.info('Investigator started')
store = etcd.Client(**store_kwargs)
while True:
# Statuses follow:
# http://commissaire.readthedocs.org/en/latest/enums.html#host-statuses
to_investigate, ssh_priv_key, remote_user = queue.get()
address = to_investigate['address']
logger.info('{0} is now in investigating.'.format(address))
logger.debug(
'Investigation details: key={0}, data={1}, remote_user={2}'.format(
to_investigate, ssh_priv_key, remote_user))
transport = ansibleapi.Transport(remote_user)
f = tempfile.NamedTemporaryFile(prefix='key', delete=False)
key_file = f.name
logger.debug(
'Using {0} as the temporary key location for {1}'.format(
key_file, address))
f.write(base64.decodestring(ssh_priv_key))
logger.info('Wrote key for {0}'.format(address))
f.close()
try:
key = '/commissaire/hosts/{0}'.format(address)
etcd_resp = store.get(key)
except Exception as error:
logger.warn(
'Unable to continue for {0} due to '
'{1}: {2}. Returning...'.format(address, type(error), error))
clean_up_key(key_file)
continue
data = json.loads(etcd_resp.value)
try:
result, facts = transport.get_info(address, key_file)
data.update(facts)
data['last_check'] = datetime.datetime.utcnow().isoformat()
data['status'] = 'bootstrapping'
logger.info('Facts for {0} retrieved'.format(address))
logger.debug('Data: {0}'.format(data))
except:
exc_type, exc_msg, tb = sys.exc_info()
logger.warn('Getting info failed for {0}: {1}'.format(
address, exc_msg))
data['status'] = 'failed'
store.write(key, json.dumps(data))
clean_up_key(key_file)
if run_once:
break
continue
store.write(key, json.dumps(data))
logger.info(
'Finished and stored investigation data for {0}'.format(address))
logger.debug('Finished investigation update for {0}: {1}'.format(
address, data))
logger.info('{0} is now in bootstrapping'.format(address))
oscmd = get_oscmd(data['os'])
try:
result, facts = transport.bootstrap(
address, key_file, config, oscmd)
data['status'] = 'inactive'
store.write(key, json.dumps(data))
except:
exc_type, exc_msg, tb = sys.exc_info()
logger.warn('Unable to start bootstraping for {0}: {1}'.format(
address, exc_msg))
data['status'] = 'disassociated'
store.write(key, json.dumps(data))
clean_up_key(key_file)
if run_once:
break
continue
# Verify association with the container manager
try:
container_mgr = KubeContainerManager(config)
# Try 3 times waiting 5 seconds each time before giving up
for cnt in range(0, 3):
if container_mgr.node_registered(address):
logger.info(
'{0} has been registered with the container manager.')
data['status'] = 'active'
break
if cnt == 3:
msg = 'Could not register with the container manager'
logger.warn(msg)
raise Exception(msg)
logger.debug(
'{0} has not been registered with the container manager. '
'Checking again in 5 seconds...'.format(address))
sleep(5)
except:
_, exc_msg, _ = sys.exc_info()
logger.warn(
'Unable to finish bootstrap for {0} while associating with '
'the container manager: {1}'.format(address, exc_msg))
data['status'] = 'inactive'
store.write(key, json.dumps(data))
logger.info(
'Finished bootstrapping for {0}'.format(address))
logging.debug('Finished bootstrapping for {0}: {1}'.format(
address, data))
clean_up_key(key_file)
if run_once:
logger.info('Exiting due to run_once request.')
break
logger.info('Investigator stopping')
def investigator(queue, config, store, run_once=False):
"""
Investigates new hosts to retrieve and store facts.
:param queue: Queue to pull work from.
:type queue: gevent.queue.Queue
:param config: Configuration information.
:type config: commissaire.config.Config
:param store: Data store to place results.
:type store: etcd.Client
"""
# TODO: Change this to be watch and etcd "queue" and kick off a function
# similar to clusterpoolexec
logger = logging.getLogger("investigator")
logger.info("Investigator started")
transport = ansibleapi.Transport()
while True:
# Statuses follow:
# http://commissaire.readthedocs.org/en/latest/enums.html#host-statuses
to_investigate, ssh_priv_key = queue.get()
address = to_investigate["address"]
logger.info("{0} is now in investigating.".format(address))
logger.debug("Investigation details: key={0}, data={1}".format(to_investigate, ssh_priv_key))
f = tempfile.NamedTemporaryFile(prefix="key", delete=False)
key_file = f.name
logger.debug("Using {0} as the temporary key location for {1}".format(key_file, address))
f.write(base64.decodestring(ssh_priv_key))
logger.debug("Wrote key for {0}".format(address))
f.close()
key = "/commissaire/hosts/{0}".format(address)
data = json.loads(store.get(key).value)
try:
result, facts = transport.get_info(address, key_file)
data.update(facts)
data["last_check"] = datetime.datetime.utcnow().isoformat()
data["status"] = "bootstrapping"
logger.info("Facts for {0} retrieved".format(address))
except:
logger.warn("Getting info failed for {0}".format(address))
data["status"] = "failed"
store.set(key, json.dumps(data))
exc_type, exc_msg, tb = sys.exc_info()
logger.debug("{0} Exception: {1}".format(address, exc_msg))
clean_up_key(key_file)
if run_once:
break
continue
store.set(key, json.dumps(data))
logger.info("Finished and stored investigation data for {0}".format(address))
logger.debug("Finished investigation update for {0}: {1}".format(address, data))
# --
logger.info("{0} is now in bootstrapping".format(address))
oscmd = get_oscmd(data["os"])()
try:
result, facts = transport.bootstrap(address, key_file, config, oscmd)
data["status"] = "inactive"
store.set(key, json.dumps(data))
except:
logger.warn("Unable to bootstrap {0}".format(address))
exc_type, exc_msg, tb = sys.exc_info()
logger.debug("{0} Exception: {1}".format(address, exc_msg))
data["status"] = "disassociated"
store.set(key, json.dumps(data))
clean_up_key(key_file)
if run_once:
break
continue
# Verify association with the container manager
try:
container_mgr = KubeContainerManager(config)
# Try 3 times waiting 5 seconds each time before giving up
for cnt in range(0, 3):
if container_mgr.node_registered(address):
logger.info("{0} has been registered with the container manager.")
data["status"] = "active"
break
if cnt == 3:
raise Exception("Could not register with the container manager")
logger.debug(
"{0} has not been registered with the container manager. "
"Checking again in 5 seconds...".format(address)
)
gevent.sleep(5)
except:
logger.warn("Unable to bootstrap {0}".format(address))
exc = sys.exc_info()[0]
logger.debug("{0} Exception: {1}".format(address, exc))
data["status"] = "inactive"
store.set(key, json.dumps(data))
logger.info("Finished bootstrapping for {0}".format(address))
logging.debug("Finished bootstrapping for {0}: {1}".format(address, data))
clean_up_key(key_file)
if run_once:
logger.info("Exiting due to run_once request.")
break
logger.info("Investigator stopping")
def investigator(queue, config, run_once=False):
"""
Investigates new hosts to retrieve and store facts.
:param queue: Queue to pull work from.
:type queue: Queue.Queue
:param config: Configuration information.
:type config: commissaire.config.Config
"""
logger = logging.getLogger('investigator')
logger.info('Investigator started')
while True:
# Statuses follow:
# http://commissaire.readthedocs.org/en/latest/enums.html#host-statuses
store_manager, to_investigate, ssh_priv_key, remote_user = queue.get()
address = to_investigate['address']
logger.info('{0} is now in investigating.'.format(address))
logger.debug(
'Investigation details: key={0}, data={1}, remote_user={2}'.format(
to_investigate, ssh_priv_key, remote_user))
transport = ansibleapi.Transport(remote_user)
try:
host = store_manager.get(
Host(
address=address,
status='',
os='',
cpus=0,
memory=0,
space=0,
last_check='',
ssh_priv_key='',
remote_user=''))
key = TemporarySSHKey(host, logger)
key.create()
except Exception as error:
logger.warn(
'Unable to continue for {0} due to '
'{1}: {2}. Returning...'.format(address, type(error), error))
key.remove()
continue
try:
result, facts = transport.get_info(address, key.path)
# recreate the host instance with new data
data = json.loads(host.to_json(secure=True))
data.update(facts)
host = Host(**data)
host.last_check = datetime.datetime.utcnow().isoformat()
host.status = 'bootstrapping'
logger.info('Facts for {0} retrieved'.format(address))
logger.debug('Data: {0}'.format(host.to_json()))
except:
exc_type, exc_msg, tb = sys.exc_info()
logger.warn('Getting info failed for {0}: {1}'.format(
address, exc_msg))
host.status = 'failed'
store_manager.save(host)
key.remove()
if run_once:
break
continue
store_manager.save(host)
logger.info(
'Finished and stored investigation data for {0}'.format(address))
logger.debug('Finished investigation update for {0}: {1}'.format(
address, host.to_json()))
logger.info('{0} is now in bootstrapping'.format(address))
oscmd = get_oscmd(host.os)
try:
result, facts = transport.bootstrap(
address, key.path, config, oscmd, store_manager)
host.status = 'inactive'
store_manager.save(host)
except:
exc_type, exc_msg, tb = sys.exc_info()
logger.warn('Unable to start bootstraping for {0}: {1}'.format(
address, exc_msg))
host.status = 'disassociated'
store_manager.save(host)
key.remove()
if run_once:
break
continue
host.status = cluster_type = C.CLUSTER_TYPE_HOST
try:
cluster = util.cluster_for_host(address, store_manager)
cluster_type = cluster.type
except KeyError:
# Not part of a cluster
pass
# Verify association with the container manager
if cluster_type == C.CLUSTER_TYPE_KUBERNETES:
try:
container_mgr = KubeContainerManager(config)
# Try 3 times waiting 5 seconds each time before giving up
for cnt in range(0, 3):
if container_mgr.node_registered(address):
logger.info(
'{0} has been registered with the '
'container manager.'.format(address))
host.status = 'active'
break
if cnt == 3:
msg = 'Could not register with the container manager'
logger.warn(msg)
raise Exception(msg)
logger.debug(
'{0} has not been registered with the container '
' manager. Checking again in 5 seconds...'.format(
address))
sleep(5)
except:
_, exc_msg, _ = sys.exc_info()
logger.warn(
'Unable to finish bootstrap for {0} while associating '
'with the container manager: {1}'.format(
address, exc_msg))
host.status = 'inactive'
store_manager.save(host)
logger.info(
'Finished bootstrapping for {0}'.format(address))
logging.debug('Finished bootstrapping for {0}: {1}'.format(
address, host.to_json()))
key.remove()
if run_once:
logger.info('Exiting due to run_once request.')
break
logger.info('Investigator stopping')
