1
|
|
|
from datetime import datetime |
2
|
|
|
|
3
|
|
|
from napalm import get_network_driver |
4
|
|
|
|
5
|
|
|
from st2reactor.sensor.base import PollingSensor |
6
|
|
|
|
7
|
|
|
BGP_PEER_INCREASE = 'napalm.BgpPeerIncrease' |
8
|
|
|
BGP_PEER_DECREASE = 'napalm.BgpPeerDecrease' |
9
|
|
|
|
10
|
|
|
|
11
|
|
|
class NapalmBGPSensor(PollingSensor): |
12
|
|
|
""" |
13
|
|
|
* self.sensor_service |
14
|
|
|
- provides utilities like |
15
|
|
|
get_logger() for writing to logs. |
16
|
|
|
dispatch() for dispatching triggers into the system. |
17
|
|
|
* self._config |
18
|
|
|
- contains configuration that was specified as |
19
|
|
|
config.yaml in the pack. |
20
|
|
|
* self._poll_interval |
21
|
|
|
- indicates the interval between two successive poll() calls. |
22
|
|
|
""" |
23
|
|
|
|
24
|
|
|
def __init__(self, sensor_service, config=None, poll_interval=5): |
25
|
|
|
super(NapalmBGPSensor, self).__init__(sensor_service=sensor_service, |
26
|
|
|
config=config, |
27
|
|
|
poll_interval=poll_interval) |
28
|
|
|
self._logger = self.sensor_service.get_logger( |
29
|
|
|
name=self.__class__.__name__ |
30
|
|
|
) |
31
|
|
|
|
32
|
|
|
# self._poll_interval = 30 |
33
|
|
|
|
34
|
|
|
def setup(self): |
35
|
|
|
# Setup stuff goes here. For example, you might establish connections |
36
|
|
|
# to external system once and reuse it. This is called only once by the system. |
37
|
|
|
|
38
|
|
|
# TODO Figure out how to use this and make a config.yaml |
39
|
|
|
# |
40
|
|
|
|
41
|
|
|
# database = database or self.config.get('default') |
42
|
|
|
# db_config = self.config.get(database, {}) |
43
|
|
|
# params = { |
44
|
|
|
# 'database': db_config.get('database') or database, |
45
|
|
|
# 'server': server or db_config.get('server'), |
46
|
|
|
# 'user': user or db_config.get('user'), |
47
|
|
|
# 'password': password or db_config.get('password') |
48
|
|
|
# } |
49
|
|
|
|
50
|
|
|
# Need to get initial BGP RIB, neighbor table, etc. Put into "self". |
51
|
|
|
# Then, write some logic within "poll" that checks again, and detects diffs |
52
|
|
|
# Detects: |
53
|
|
|
# - Diffs in BGP RIB (need to give a threshold like 100s or 1000s or routes different) |
54
|
|
|
# (may want to not only store the previous result, but also the previous 10 or so and do a stddev calc) |
55
|
|
|
|
56
|
|
|
# Stores number of BGP neighbors |
57
|
|
|
# self.bgp_neighbors = 0 |
58
|
|
|
|
59
|
|
|
# Stores RIB information in-between calls |
60
|
|
|
# self.bgp_rib = {} |
61
|
|
|
|
62
|
|
|
# Dictionary for tracking per-device known state |
63
|
|
|
# Top-level keys are the management IPs sent to NAPALM, and |
64
|
|
|
# information on each is contained below that |
65
|
|
|
self.device_state = {} |
66
|
|
|
|
67
|
|
|
napalm_config = self._config |
68
|
|
|
|
69
|
|
|
# Assign sane defaults for configuration |
70
|
|
|
default_opts = { |
71
|
|
|
"opt1": "val1" |
72
|
|
|
} |
73
|
|
|
for opt_name, opt_val in default_opts.items(): |
74
|
|
|
|
75
|
|
|
try: |
76
|
|
|
|
77
|
|
|
# key exists but is set to nothing |
78
|
|
|
if not napalm_config[opt_name]: |
79
|
|
|
napalm_config[opt_name] == default_opts |
80
|
|
|
|
81
|
|
|
except KeyError: |
82
|
|
|
|
83
|
|
|
# key doesn't exist |
84
|
|
|
napalm_config[opt_name] == default_opts |
85
|
|
|
|
86
|
|
|
# Assign options to instance |
87
|
|
|
self._devices = napalm_config['devices'] |
88
|
|
|
|
89
|
|
|
# Generate dictionary of device objects per configuration |
90
|
|
|
# IP Address(key): Device Object(value) |
91
|
|
|
# |
92
|
|
|
# TODO(mierdin): Yes I know this looks terrible, I will fix it |
93
|
|
|
self.devices = { |
94
|
|
|
str(device['host']): get_network_driver(device['driver'])( |
95
|
|
|
hostname=str(device['host']), |
96
|
|
|
username=device['username'], |
97
|
|
|
password=device['password'], |
98
|
|
|
optional_args={ |
99
|
|
|
'port': str(device['port']) |
100
|
|
|
}) |
101
|
|
|
for device in self._devices |
102
|
|
|
} |
103
|
|
|
|
104
|
|
|
def poll(self): |
105
|
|
|
# This is where the crux of the sensor work goes. |
106
|
|
|
# This is called every self._poll_interval. |
107
|
|
|
# For example, let's assume you want to query ec2 and get |
108
|
|
|
# health information about your instances: |
109
|
|
|
# some_data = aws_client.get('') |
110
|
|
|
# payload = self._to_payload(some_data) |
111
|
|
|
# # _to_triggers is something you'd write to convert the data format you have |
112
|
|
|
# # into a standard python dictionary. This should follow the payload schema |
113
|
|
|
# # registered for the trigger. |
114
|
|
|
# self.sensor_service.dispatch(trigger, payload) |
115
|
|
|
# # You can refer to the trigger as dict |
116
|
|
|
# # { "name": ${trigger_name}, "pack": ${trigger_pack} } |
117
|
|
|
# # or just simply by reference as string. |
118
|
|
|
# # i.e. dispatch(${trigger_pack}.${trigger_name}, payload) |
119
|
|
|
# # E.g.: dispatch('examples.foo_sensor', {'k1': 'stuff', 'k2': 'foo'}) |
120
|
|
|
# # trace_tag is a tag you would like to associate with the dispacthed TriggerInstance |
121
|
|
|
# # Typically the trace_tag is unique and a reference to an external event. |
122
|
|
|
|
123
|
|
|
for hostname, device_obj in self.devices.items(): |
124
|
|
|
|
125
|
|
|
try: |
126
|
|
|
last_bgp_peers = self.device_state[hostname]["last_bgp_peers"] |
127
|
|
|
except KeyError: |
128
|
|
|
|
129
|
|
|
# Get current BGP peers (instead of setting to 0 and |
130
|
|
|
# triggering every time sensor starts initially) |
131
|
|
|
try: |
132
|
|
|
self.device_state[hostname] = { |
133
|
|
|
"last_bgp_peers": self.get_number_of_peers(device_obj) |
134
|
|
|
} |
135
|
|
|
continue |
136
|
|
|
# Any connection-related exception raised here is |
137
|
|
|
# driver-specific, so we have to catch "Exception" |
138
|
|
|
except Exception as e: |
139
|
|
|
self._logger.debug("Caught exception on connect: %s" % e) |
140
|
|
|
continue |
141
|
|
|
|
142
|
|
|
try: |
143
|
|
|
this_bgp_peers = self.get_number_of_peers(device_obj) |
144
|
|
|
# Any connection-related exception raised here is |
145
|
|
|
# driver-specific, so we have to catch "Exception" |
146
|
|
|
except Exception as e: |
147
|
|
|
self._logger.debug("Caught exception on get peers: %s" % e) |
148
|
|
|
continue |
149
|
|
|
|
150
|
|
|
if this_bgp_peers > last_bgp_peers: |
151
|
|
|
self._logger.info( |
152
|
|
|
"Peer count went UP to %s" % str(this_bgp_peers) |
153
|
|
|
) |
154
|
|
|
self._bgp_peer_trigger(BGP_PEER_INCREASE, hostname, |
155
|
|
|
last_bgp_peers, this_bgp_peers) |
156
|
|
|
|
157
|
|
|
elif this_bgp_peers < last_bgp_peers: |
158
|
|
|
self._logger.info( |
159
|
|
|
"BGP neighbors went DOWN to %s" % str(this_bgp_peers) |
160
|
|
|
) |
161
|
|
|
|
162
|
|
|
self._bgp_peer_trigger(BGP_PEER_DECREASE, hostname, |
163
|
|
|
last_bgp_peers, this_bgp_peers) |
164
|
|
|
|
165
|
|
|
elif this_bgp_peers == last_bgp_peers: |
166
|
|
|
self._logger.info( |
167
|
|
|
"BGP neighbors STAYED at %s" % str(this_bgp_peers) |
168
|
|
|
) |
169
|
|
|
|
170
|
|
|
# Save this state for the next poll |
171
|
|
|
self.device_state[hostname]["last_bgp_peers"] = this_bgp_peers |
172
|
|
|
|
173
|
|
|
def cleanup(self): |
174
|
|
|
# This is called when the st2 system goes down. You can perform |
175
|
|
|
# cleanup operations like closing the connections to external |
176
|
|
|
# system here. |
177
|
|
|
pass |
178
|
|
|
|
179
|
|
|
def add_trigger(self, trigger): |
180
|
|
|
# This method is called when trigger is created |
181
|
|
|
pass |
182
|
|
|
|
183
|
|
|
def update_trigger(self, trigger): |
184
|
|
|
# This method is called when trigger is updated |
185
|
|
|
pass |
186
|
|
|
|
187
|
|
|
def remove_trigger(self, trigger): |
188
|
|
|
# This method is called when trigger is deleted |
189
|
|
|
pass |
190
|
|
|
|
191
|
|
|
def get_number_of_peers(self, device_obj): |
192
|
|
|
|
193
|
|
|
vrfs = {} |
194
|
|
|
|
195
|
|
|
# Retrieve full BGP peer info at the VRF level |
196
|
|
|
try: |
197
|
|
|
with device_obj: |
198
|
|
|
vrfs = device_obj.get_bgp_neighbors() |
199
|
|
|
except Exception: # TODO(mierdin): convert to ConnectionException |
200
|
|
|
raise |
201
|
|
|
|
202
|
|
|
this_bgp_peers = 0 |
203
|
|
|
|
204
|
|
|
for vrf_id, vrf_details in vrfs.items(): |
205
|
|
|
|
206
|
|
|
for peer_id, peer_details in vrf_details['peers'].items(): |
207
|
|
|
|
208
|
|
|
# TODO(mierdin): This isn't always the fastest method when |
209
|
|
|
# peers go down. Try to improve on this. |
210
|
|
|
if peer_details['is_up']: |
211
|
|
|
this_bgp_peers += 1 |
212
|
|
|
|
213
|
|
|
return this_bgp_peers |
214
|
|
|
|
215
|
|
|
def _bgp_peer_trigger(self, trigger, hostname, oldpeers, newpeers): |
216
|
|
|
trigger = 'napalm.BgpPeerDecrease' |
217
|
|
|
payload = { |
218
|
|
|
'device': hostname, |
219
|
|
|
'oldpeers': int(oldpeers), |
220
|
|
|
'newpeers': int(newpeers), |
221
|
|
|
'timestamp': str(datetime.now()), |
222
|
|
|
} |
223
|
|
|
self._logger.debug("DISPATCHING TRIGGER %s" % trigger) |
224
|
|
|
self._sensor_service.dispatch(trigger=trigger, payload=payload) |
225
|
|
|
|