Ping monitor ignores retry parameter
Now Tacker ping monitor do not retry even in 'retry' parameter is specified like described in [1]. In addition, description about ping monitor in [1] is incorrect. This patch fixes above problems. [1] https://docs.openstack.org/tacker/latest/contributor/vnfd_template_description.html Change-Id: I8e7cc4367002724f9ff00a623f7722e915ca1333 Closes-Bug: #1861774
This commit is contained in:
JO Hiroyuki
parent
55319282f3
commit
364ce6921e
@ -52,10 +52,13 @@ Following methods need to be overridden in the new driver:
|
||||
This method must return the url of vnf to monitor.
|
||||
|
||||
``def monitor_call(self, vnf, kwargs)``
|
||||
This method must either return boolean value 'True', if VNF is healthy.
|
||||
Otherwise it should return an event string like 'failure' or
|
||||
'calls-capacity-reached' based on specific VNF health condition. More
|
||||
details on these event is given in below section.
|
||||
This method is called cyclically each time a monitoring is
|
||||
triggered. **kwagrs** is a dict object given under **parameters** in
|
||||
the target VDU template. This method must either return boolean
|
||||
value 'True', if VNF is healthy. Otherwise it should return an event
|
||||
string like 'failure' or 'calls-capacity-reached' based on specific
|
||||
VNF health condition. More details on these event is given in below
|
||||
section.
|
||||
|
||||
Custom events
|
||||
--------------
|
||||
@ -67,13 +70,15 @@ For example:
|
||||
|
||||
::
|
||||
|
||||
vdu1:
|
||||
monitoring_policy:
|
||||
ping:
|
||||
VDU1:
|
||||
properties:
|
||||
...
|
||||
monitoring_policy:
|
||||
name: ping
|
||||
actions:
|
||||
failure: respawn
|
||||
|
||||
In this example, we have an event called 'failure'. So whenever monitor_call
|
||||
In this example, we have an event called 'failure'. So whenever monitor_call
|
||||
returns 'failure' tacker will respawn the VNF.
|
||||
|
||||
|
||||
@ -85,31 +90,32 @@ occurs.
|
||||
#. respawn
|
||||
In case of OpenStack VIM, when any VDU monitoring fails, it will delete
|
||||
the entire VNF and create a new one.
|
||||
#. log
|
||||
#. vdu_autoheal
|
||||
In case of OpenStack VIM, when any VDU monitoring fails, it will delete
|
||||
only that specific VDU resource and create a new one alone with it's
|
||||
dependent resources like CP.
|
||||
#. log
|
||||
#. log_and_kill
|
||||
|
||||
How to write TOSCA template to monitor VNF entities
|
||||
----------------------------------------------------
|
||||
|
||||
In the vdus section, under vdu you can specify the monitors details with
|
||||
corresponding actions and parameters.The syntax for writing monitor policy
|
||||
is as follows:
|
||||
In the vdus section, you can specify the monitor details with
|
||||
corresponding actions and parameters. The syntax for writing monitor
|
||||
policy is as follows:
|
||||
|
||||
::
|
||||
|
||||
vduN:
|
||||
monitoring_policy:
|
||||
<monitoring-driver-name>:
|
||||
monitoring_params:
|
||||
properties:
|
||||
...
|
||||
monitoring_policy:
|
||||
name: <monitoring-driver-name>:
|
||||
parameters:
|
||||
<param-name>: <param-value>
|
||||
...
|
||||
actions:
|
||||
<event>: <action-name>
|
||||
...
|
||||
...
|
||||
<event-name>: <action-name>
|
||||
|
||||
|
||||
Example Template
|
||||
@ -117,32 +123,32 @@ Example Template
|
||||
|
||||
::
|
||||
|
||||
vdu1:
|
||||
monitoring_policy:
|
||||
ping:
|
||||
VDU1:
|
||||
properties:
|
||||
...
|
||||
monitoring_policy:
|
||||
name: ping
|
||||
actions:
|
||||
failure: respawn
|
||||
|
||||
vdu2:
|
||||
monitoring_policy:
|
||||
http-ping:
|
||||
monitoring_params:
|
||||
VDU2:
|
||||
properties:
|
||||
...
|
||||
monitoring_policy:
|
||||
name: http-ping
|
||||
parameters:
|
||||
port: 8080
|
||||
url: ping.cgi
|
||||
actions:
|
||||
failure: respawn
|
||||
|
||||
vdu_scaling_driver:
|
||||
monitoring_params:
|
||||
resource: cpu
|
||||
threshold: 10000
|
||||
actions:
|
||||
max_foo_reached: scale_up
|
||||
min_foo_reached: scale_down
|
||||
|
||||
vdu3:
|
||||
monitoring_policy:
|
||||
ping:
|
||||
actions:
|
||||
failure: vdu_autoheal
|
||||
|
||||
VDU3:
|
||||
properties:
|
||||
...
|
||||
monitoring_policy:
|
||||
name: <your-driver-name>
|
||||
parameters:
|
||||
<param1>: <value1>
|
||||
<param2>: <value2>
|
||||
actions:
|
||||
<event1>: <action>
|
||||
<event2>: <action>
|
||||
|
||||
@ -110,10 +110,15 @@ a VNFD, **flavor** setting will take precedence.
|
||||
|
||||
Monitoring the VDU
|
||||
""""""""""""""""""
|
||||
A VDU can be monitored by pinging it on port 22 for 3 times at an interval of
|
||||
2 seconds every 20 seconds. Number of retries be 6 and timeout of 2 seconds.
|
||||
It can be re-spawned in case ping fails. This is described under
|
||||
**monitoring_policy**.
|
||||
|
||||
A VDU can be monitored by pinging it. The following VNFD pings VDU1 with
|
||||
3 ECHO packet counts, 0.2 second intervals and 2 second timeout options
|
||||
per monitoring trigger. Ping is attempted up to 6 times unless it
|
||||
returns an exit code 0. On VDU2, Tacker tries to open
|
||||
http://<VDU2_mgmt_ip>:80 with 2 second timeout and 6 retries. Both VDUs
|
||||
have 20 second delay time from creation or respawning to start
|
||||
monitoring. The VDUs can be re-spawned in case of failure. See
|
||||
:doc:`monitor-api` for more information.
|
||||
|
||||
::
|
||||
|
||||
@ -123,15 +128,27 @@ It can be re-spawned in case ping fails. This is described under
|
||||
properties:
|
||||
monitoring_policy:
|
||||
name: ping
|
||||
parameters:
|
||||
monitoring_delay: 20
|
||||
count: 3
|
||||
interval: 2
|
||||
timeout: 2
|
||||
actions:
|
||||
failure: respawn
|
||||
retry: 6
|
||||
port: 22
|
||||
parameters:
|
||||
monitoring_delay: 20
|
||||
count: 3
|
||||
interval: 0.2
|
||||
timeout: 2
|
||||
retry: 6
|
||||
actions:
|
||||
failure: respawn
|
||||
|
||||
VDU2:
|
||||
type: tosca.nodes.nfv.VDU.Tacker
|
||||
properties:
|
||||
monitoring_policy:
|
||||
name: http-ping
|
||||
parameters:
|
||||
monitoring_delay: 20
|
||||
timeout: 2
|
||||
retry: 6
|
||||
port: 80
|
||||
actions:
|
||||
failure: respawn
|
||||
|
||||
Providing user data
|
||||
"""""""""""""""""""
|
||||
|
||||
@ -32,8 +32,8 @@ class TestVNFMonitorPing(testtools.TestCase):
|
||||
}
|
||||
mock_ping_cmd = ['ping',
|
||||
'-c', 5,
|
||||
'-W', 1,
|
||||
'-i', '0.2',
|
||||
'-W', 5.0,
|
||||
'-i', 1.0,
|
||||
'a.b.c.d']
|
||||
self.monitor_ping.monitor_call(test_vnf,
|
||||
test_kwargs)
|
||||
@ -59,3 +59,39 @@ class TestVNFMonitorPing(testtools.TestCase):
|
||||
mock.ANY,
|
||||
test_vnf)
|
||||
self.assertEqual('a.b.c.d', test_monitor_url)
|
||||
|
||||
@mock.patch('tacker.agent.linux.utils.execute')
|
||||
def test_monitor_call_with_params(self, mock_utils_execute):
|
||||
check_ping_counts = 2
|
||||
check_ping_timeout = 5.0
|
||||
check_ping_interval = 0.5
|
||||
test_vnf = {}
|
||||
test_kwargs = {
|
||||
'mgmt_ip': 'a:b:c:d:e:f:1:2',
|
||||
'count': check_ping_counts,
|
||||
'timeout': check_ping_timeout,
|
||||
'interval': check_ping_interval
|
||||
}
|
||||
mock_ping_cmd = ['ping6',
|
||||
'-c', check_ping_counts,
|
||||
'-W', check_ping_timeout,
|
||||
'-i', check_ping_interval,
|
||||
'a:b:c:d:e:f:1:2']
|
||||
self.monitor_ping.monitor_call(test_vnf,
|
||||
test_kwargs)
|
||||
mock_utils_execute.assert_called_once_with(mock_ping_cmd,
|
||||
check_exit_code=True)
|
||||
|
||||
@mock.patch('tacker.agent.linux.utils.execute')
|
||||
def test_monitor_call_for_counts(self, mock_utils_execute):
|
||||
check_retury_counts = 5
|
||||
mock_utils_execute.side_effect = RuntimeError()
|
||||
test_vnf = {}
|
||||
test_kwargs = {
|
||||
'mgmt_ip': 'a:b:c:d:e:f:1:2',
|
||||
'retry': check_retury_counts
|
||||
}
|
||||
self.monitor_ping.monitor_call(test_vnf,
|
||||
test_kwargs)
|
||||
self.assertEqual(check_retury_counts,
|
||||
mock_utils_execute.call_count)
|
||||
|
||||
@ -25,12 +25,14 @@ from tacker.vnfm.monitor_drivers import abstract_driver
|
||||
|
||||
LOG = logging.getLogger(__name__)
|
||||
OPTS = [
|
||||
cfg.StrOpt('count', default='1',
|
||||
cfg.IntOpt('count', default=5,
|
||||
help=_('Number of ICMP packets to send')),
|
||||
cfg.StrOpt('timeout', default='1',
|
||||
cfg.FloatOpt('timeout', default=5,
|
||||
help=_('Number of seconds to wait for a response')),
|
||||
cfg.StrOpt('interval', default='1',
|
||||
help=_('Number of seconds to wait between packets'))
|
||||
cfg.FloatOpt('interval', default=1,
|
||||
help=_('Number of seconds to wait between packets')),
|
||||
cfg.IntOpt('retry', default=1,
|
||||
help=_('Number of ping retries'))
|
||||
]
|
||||
cfg.CONF.register_opts(OPTS, 'monitor_ping')
|
||||
|
||||
@ -53,12 +55,12 @@ class VNFMonitorPing(abstract_driver.VNFMonitorAbstractDriver):
|
||||
LOG.debug('monitor_url %s', vnf)
|
||||
return vnf.get('monitor_url', '')
|
||||
|
||||
def _is_pingable(self, mgmt_ip="", count=5, timeout=1, interval='0.2',
|
||||
**kwargs):
|
||||
def _is_pingable(self, mgmt_ip="", count=None, timeout=None,
|
||||
interval=None, retry=None, **kwargs):
|
||||
"""Checks whether an IP address is reachable by pinging.
|
||||
|
||||
Use linux utils to execute the ping (ICMP ECHO) command.
|
||||
Sends 5 packets with an interval of 0.2 seconds and timeout of 1
|
||||
Sends 5 packets with an interval of 1 seconds and timeout of 1
|
||||
seconds. Runtime error implies unreachability else IP is pingable.
|
||||
:param ip: IP to check
|
||||
:return: bool - True or string 'failure' depending on pingability.
|
||||
@ -67,18 +69,28 @@ class VNFMonitorPing(abstract_driver.VNFMonitorAbstractDriver):
|
||||
if netaddr.valid_ipv6(mgmt_ip):
|
||||
cmd_ping = 'ping6'
|
||||
|
||||
if not count:
|
||||
count = cfg.CONF.monitor_ping.count
|
||||
if not timeout:
|
||||
timeout = cfg.CONF.monitor_ping.timeout
|
||||
if not interval:
|
||||
interval = cfg.CONF.monitor_ping.interval
|
||||
if not retry:
|
||||
retry = cfg.CONF.monitor_ping.retry
|
||||
|
||||
ping_cmd = [cmd_ping,
|
||||
'-c', count,
|
||||
'-W', timeout,
|
||||
'-i', interval,
|
||||
mgmt_ip]
|
||||
|
||||
try:
|
||||
linux_utils.execute(ping_cmd, check_exit_code=True)
|
||||
return True
|
||||
except RuntimeError:
|
||||
LOG.warning("Cannot ping ip address: %s", mgmt_ip)
|
||||
return 'failure'
|
||||
for retry_range in range(int(retry)):
|
||||
try:
|
||||
linux_utils.execute(ping_cmd, check_exit_code=True)
|
||||
return True
|
||||
except RuntimeError:
|
||||
LOG.warning("Cannot ping ip address: %s", mgmt_ip)
|
||||
return 'failure'
|
||||
|
||||
@log.log
|
||||
def monitor_call(self, vnf, kwargs):
|
||||
|
||||
Loading…
x
Reference in New Issue
Block a user