1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
|
"""
diskprediction with local predictor
"""
import json
import datetime
from threading import Event
import time
from typing import Any, Dict, List, Optional, Tuple, TYPE_CHECKING
from mgr_module import CommandResult, MgrModule, Option
# Importing scipy early appears to avoid a future deadlock when
# we try to do
#
# from .predictor import get_diskfailurepredictor_path
#
# in a command thread. See https://tracker.ceph.com/issues/42764
import scipy # noqa: ignore=F401
from .predictor import DevSmartT, Predictor, get_diskfailurepredictor_path
TIME_FORMAT = '%Y%m%d-%H%M%S'
TIME_DAYS = 24 * 60 * 60
TIME_WEEK = TIME_DAYS * 7
class Module(MgrModule):
MODULE_OPTIONS = [
Option(name='sleep_interval',
default=600),
Option(name='predict_interval',
default=86400),
Option(name='predictor_model',
default='prophetstor')
]
def __init__(self, *args: Any, **kwargs: Any) -> None:
super(Module, self).__init__(*args, **kwargs)
# options
for opt in self.MODULE_OPTIONS:
setattr(self, opt['name'], opt['default'])
# other
self._run = True
self._event = Event()
# for mypy which does not run the code
if TYPE_CHECKING:
self.sleep_interval = 0
self.predict_interval = 0
self.predictor_model = ''
def config_notify(self) -> None:
for opt in self.MODULE_OPTIONS:
setattr(self,
opt['name'],
self.get_module_option(opt['name']))
self.log.debug(' %s = %s', opt['name'], getattr(self, opt['name']))
if self.get_ceph_option('device_failure_prediction_mode') == 'local':
self._event.set()
def refresh_config(self) -> None:
for opt in self.MODULE_OPTIONS:
setattr(self,
opt['name'],
self.get_module_option(opt['name']))
self.log.debug(' %s = %s', opt['name'], getattr(self, opt['name']))
def self_test(self) -> None:
self.log.debug('self_test enter')
ret, out, err = self.predict_all_devices()
assert ret == 0
def serve(self) -> None:
self.log.info('Starting diskprediction local module')
self.config_notify()
last_predicted = None
ls = self.get_store('last_predicted')
if ls:
try:
last_predicted = datetime.datetime.strptime(ls, TIME_FORMAT)
except ValueError:
pass
self.log.debug('Last predicted %s', last_predicted)
while self._run:
self.refresh_config()
mode = self.get_ceph_option('device_failure_prediction_mode')
if mode == 'local':
now = datetime.datetime.utcnow()
if not last_predicted:
next_predicted = now
else:
predicted_frequency = self.predict_interval or 86400
seconds = (last_predicted - datetime.datetime.utcfromtimestamp(0)).total_seconds()
seconds -= seconds % predicted_frequency
seconds += predicted_frequency
next_predicted = datetime.datetime.utcfromtimestamp(seconds)
self.log.debug('Last scrape %s, next scrape due %s',
last_predicted.strftime(TIME_FORMAT),
next_predicted.strftime(TIME_FORMAT))
if now >= next_predicted:
self.predict_all_devices()
last_predicted = now
self.set_store('last_predicted', last_predicted.strftime(TIME_FORMAT))
sleep_interval = self.sleep_interval or 60
self.log.debug('Sleeping for %d seconds', sleep_interval)
self._event.wait(sleep_interval)
self._event.clear()
def shutdown(self) -> None:
self.log.info('Stopping')
self._run = False
self._event.set()
@staticmethod
def _convert_timestamp(predicted_timestamp: int, life_expectancy_day: int) -> str:
"""
:param predicted_timestamp: unit is nanoseconds
:param life_expectancy_day: unit is seconds
:return:
date format '%Y-%m-%d' ex. 2018-01-01
"""
return datetime.datetime.fromtimestamp(
predicted_timestamp / (1000 ** 3) + life_expectancy_day).strftime('%Y-%m-%d')
def _predict_life_expectancy(self, devid: str) -> str:
predicted_result = ''
health_data: Dict[str, Dict[str, Any]] = {}
predict_datas: List[DevSmartT] = []
try:
r, outb, outs = self.remote(
'devicehealth', 'show_device_metrics', devid=devid, sample='')
if r != 0:
self.log.error('failed to get device %s health', devid)
health_data = {}
else:
health_data = json.loads(outb)
except Exception as e:
self.log.error('failed to get device %s health data due to %s', devid, str(e))
# initialize appropriate disk failure predictor model
obj_predictor = Predictor.create(self.predictor_model)
if obj_predictor is None:
self.log.error('invalid value received for MODULE_OPTIONS.predictor_model')
return predicted_result
try:
obj_predictor.initialize(
"{}/models/{}".format(get_diskfailurepredictor_path(), self.predictor_model))
except Exception as e:
self.log.error('Error initializing predictor: %s', e)
return predicted_result
if len(health_data) >= 6:
o_keys = sorted(health_data.keys(), reverse=True)
for o_key in o_keys:
# get values for current day (?)
dev_smart = {}
s_val = health_data[o_key]
# add all smart attributes
ata_smart = s_val.get('ata_smart_attributes', {})
for attr in ata_smart.get('table', []):
# get raw smart values
if attr.get('raw', {}).get('string') is not None:
if str(attr.get('raw', {}).get('string', '0')).isdigit():
dev_smart['smart_%s_raw' % attr.get('id')] = \
int(attr.get('raw', {}).get('string', '0'))
else:
if str(attr.get('raw', {}).get('string', '0')).split(' ')[0].isdigit():
dev_smart['smart_%s_raw' % attr.get('id')] = \
int(attr.get('raw', {}).get('string',
'0').split(' ')[0])
else:
dev_smart['smart_%s_raw' % attr.get('id')] = \
attr.get('raw', {}).get('value', 0)
# get normalized smart values
if attr.get('value') is not None:
dev_smart['smart_%s_normalized' % attr.get('id')] = \
attr.get('value')
# add power on hours manually if not available in smart attributes
power_on_time = s_val.get('power_on_time', {}).get('hours')
if power_on_time is not None:
dev_smart['smart_9_raw'] = int(power_on_time)
# add device capacity
user_capacity = s_val.get('user_capacity', {}).get('bytes')
if user_capacity is not None:
dev_smart['user_capacity'] = user_capacity
else:
self.log.debug('user_capacity not found in smart attributes list')
# add device model
model_name = s_val.get('model_name')
if model_name is not None:
dev_smart['model_name'] = model_name
# add vendor
vendor = s_val.get('vendor')
if vendor is not None:
dev_smart['vendor'] = vendor
# if smart data was found, then add that to list
if dev_smart:
predict_datas.append(dev_smart)
if len(predict_datas) >= 12:
break
else:
self.log.error('unable to predict device due to health data records less than 6 days')
if len(predict_datas) >= 6:
predicted_result = obj_predictor.predict(predict_datas)
return predicted_result
def predict_life_expectancy(self, devid: str) -> Tuple[int, str, str]:
result = self._predict_life_expectancy(devid)
if result.lower() == 'good':
return 0, '>6w', ''
elif result.lower() == 'warning':
return 0, '>=2w and <=6w', ''
elif result.lower() == 'bad':
return 0, '<2w', ''
else:
return 0, 'unknown', ''
def _reset_device_life_expectancy(self, device_id: str) -> int:
result = CommandResult('')
self.send_command(result, 'mon', '', json.dumps({
'prefix': 'device rm-life-expectancy',
'devid': device_id
}), '')
ret, _, outs = result.wait()
if ret != 0:
self.log.error(
'failed to reset device life expectancy, %s' % outs)
return ret
def _set_device_life_expectancy(self,
device_id: str,
from_date: str,
to_date: Optional[str] = None) -> int:
result = CommandResult('')
if to_date is None:
self.send_command(result, 'mon', '', json.dumps({
'prefix': 'device set-life-expectancy',
'devid': device_id,
'from': from_date
}), '')
else:
self.send_command(result, 'mon', '', json.dumps({
'prefix': 'device set-life-expectancy',
'devid': device_id,
'from': from_date,
'to': to_date
}), '')
ret, _, outs = result.wait()
if ret != 0:
self.log.error(
'failed to set device life expectancy, %s' % outs)
return ret
def predict_all_devices(self) -> Tuple[int, str, str]:
self.log.debug('predict_all_devices')
devices = self.get('devices').get('devices', [])
for devInfo in devices:
if not devInfo.get('daemons'):
continue
if not devInfo.get('devid'):
continue
self.log.debug('%s' % devInfo)
result = self._predict_life_expectancy(devInfo['devid'])
if result == 'unknown':
self._reset_device_life_expectancy(devInfo['devid'])
continue
predicted = int(time.time() * (1000 ** 3))
if result.lower() == 'good':
life_expectancy_day_min = (TIME_WEEK * 6) + TIME_DAYS
life_expectancy_day_max = 0
elif result.lower() == 'warning':
life_expectancy_day_min = (TIME_WEEK * 2)
life_expectancy_day_max = (TIME_WEEK * 6)
elif result.lower() == 'bad':
life_expectancy_day_min = 0
life_expectancy_day_max = (TIME_WEEK * 2) - TIME_DAYS
else:
predicted = 0
life_expectancy_day_min = 0
life_expectancy_day_max = 0
if predicted and devInfo['devid'] and life_expectancy_day_min:
from_date = None
to_date = None
try:
assert life_expectancy_day_min
from_date = self._convert_timestamp(predicted, life_expectancy_day_min)
if life_expectancy_day_max:
to_date = self._convert_timestamp(predicted, life_expectancy_day_max)
self._set_device_life_expectancy(devInfo['devid'], from_date, to_date)
self._logger.info(
'succeed to set device {} life expectancy from: {}, to: {}'.format(
devInfo['devid'], from_date, to_date))
except Exception as e:
self._logger.error(
'failed to set device {} life expectancy from: {}, to: {}, {}'.format(
devInfo['devid'], from_date, to_date, str(e)))
else:
self._reset_device_life_expectancy(devInfo['devid'])
return 0, 'succeed to predicted all devices', ''
|
