node.py
16.2 KB
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
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
# Copyright 2014 Budapest University of Technology and Economics (BME IK)
#
# This file is part of CIRCLE Cloud.
#
# CIRCLE is free software: you can redistribute it and/or modify it under
# the terms of the GNU General Public License as published by the Free
# Software Foundation, either version 3 of the License, or (at your option)
# any later version.
#
# CIRCLE is distributed in the hope that it will be useful, but WITHOUT ANY
# WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
# FOR A PARTICULAR PURPOSE. See the GNU General Public License for more
# details.
#
# You should have received a copy of the GNU General Public License along
# with CIRCLE. If not, see <http://www.gnu.org/licenses/>.
from __future__ import absolute_import, unicode_literals
from functools import update_wrapper
from glob import glob
from logging import getLogger
import os.path
from warnings import warn
import requests
from salt.client import LocalClient
from salt.exceptions import SaltClientError
import salt.utils
from time import time, sleep
from django.conf import settings
from django.db.models import (
CharField, IntegerField, ForeignKey, BooleanField, ManyToManyField,
FloatField, permalink, Sum
)
from django.utils import timezone
from django.utils.translation import ugettext_lazy as _
from celery.exceptions import TimeoutError
from model_utils.models import TimeStampedModel
from taggit.managers import TaggableManager
from common.models import method_cache, WorkerNotFound, HumanSortField
from common.operations import OperatedMixin
from firewall.models import Host
from ..tasks import vm_tasks
from .activity import NodeActivity
from .common import Trait
logger = getLogger(__name__)
class MyLocalClient(LocalClient):
def get_returns(self, jid, minions, timeout=None):
'''
Get the returns for the command line interface via the event system
'''
minions = set(minions)
if timeout is None:
timeout = self.opts['timeout']
jid_dir = salt.utils.jid_dir(jid,
self.opts['cachedir'],
self.opts['hash_type'])
start = time()
timeout_at = start + timeout
found = set()
ret = {}
wtag = os.path.join(jid_dir, 'wtag*')
# Check to see if the jid is real, if not return the empty dict
if not os.path.isdir(jid_dir):
logger.warning("jid_dir (%s) does not exist", jid_dir)
return ret
# Wait for the hosts to check in
while True:
time_left = timeout_at - time()
raw = self.event.get_event(time_left, jid)
if raw is not None and 'return' in raw:
found.add(raw['id'])
ret[raw['id']] = raw['return']
if len(found.intersection(minions)) >= len(minions):
# All minions have returned, break out of the loop
logger.debug("jid %s found all minions", jid)
break
continue
# Then event system timeout was reached and nothing was returned
if len(found.intersection(minions)) >= len(minions):
# All minions have returned, break out of the loop
logger.debug("jid %s found all minions", jid)
break
if glob(wtag) and time() <= timeout_at + 1:
# The timeout +1 has not been reached and there is still a
# write tag for the syndic
continue
if time() > timeout_at:
logger.info('jid %s minions %s did not return in time',
jid, (minions - found))
break
sleep(0.01)
return ret
def node_available(function):
"""Decorate methods to ignore disabled Nodes.
"""
def decorate(self, *args, **kwargs):
if self.enabled and self.online:
return function(self, *args, **kwargs)
else:
return None
update_wrapper(decorate, function)
decorate._original = function
return decorate
class Node(OperatedMixin, TimeStampedModel):
"""A VM host machine, a hypervisor.
"""
name = CharField(max_length=50, unique=True,
verbose_name=_('name'),
help_text=_('Human readable name of node.'))
normalized_name = HumanSortField(monitor='name', max_length=100)
priority = IntegerField(verbose_name=_('priority'),
help_text=_('Node usage priority.'))
host = ForeignKey(Host, verbose_name=_('host'),
help_text=_('Host in firewall.'))
enabled = BooleanField(verbose_name=_('enabled'), default=False,
help_text=_('Indicates whether the node can '
'be used for hosting.'))
schedule_enabled = BooleanField(verbose_name=_('schedule enabled'),
default=False, help_text=_(
'Indicates whether a vm can be '
'automatically scheduled to this '
'node.'))
traits = ManyToManyField(Trait, blank=True,
help_text=_("Declared traits."),
verbose_name=_('traits'))
tags = TaggableManager(blank=True, verbose_name=_("tags"))
overcommit = FloatField(default=1.0, verbose_name=_("overcommit ratio"),
help_text=_("The ratio of total memory with "
"to without overcommit."))
class Meta:
app_label = 'vm'
db_table = 'vm_node'
permissions = (
('view_statistics', _('Can view Node box and statistics.')),
)
ordering = ('-enabled', 'normalized_name')
def __unicode__(self):
return self.name
@method_cache(10)
def get_online(self):
"""Check if the node is online.
Check if node is online by queue is available.
"""
try:
self.get_remote_queue_name("vm", "fast")
except:
return False
else:
return True
online = property(get_online)
@method_cache(20)
def get_minion_online(self):
name = self.host.hostname
try:
client = MyLocalClient()
client.opts['timeout'] = 0.2
return bool(client.cmd(name, 'test.ping')[name])
except (KeyError, SaltClientError):
return False
minion_online = property(get_minion_online)
@node_available
@method_cache(300)
def get_info(self):
return self.remote_query(vm_tasks.get_info,
priority='fast',
default={'core_num': 0,
'ram_size': 0,
'architecture': ''})
info = property(get_info)
@property
def allocated_ram(self):
return (self.instance_set.aggregate(
r=Sum('ram_size'))['r'] or 0) * 1024 * 1024
@property
def ram_size(self):
warn('Use Node.info["ram_size"]', DeprecationWarning)
return self.info['ram_size']
@property
def num_cores(self):
warn('Use Node.info["core_num"]', DeprecationWarning)
return self.info['core_num']
STATES = {None: ({True: ('MISSING', _('missing')),
False: ('OFFLINE', _('offline'))}),
False: {False: ('DISABLED', _('disabled'))},
True: {False: ('PASSIVE', _('passive')),
True: ('ACTIVE', _('active'))}}
def _get_state(self):
"""The state tuple based on online and enabled attributes.
"""
if self.online:
return self.STATES[self.enabled][self.schedule_enabled]
else:
return self.STATES[None][self.enabled]
def get_status_display(self):
return self._get_state()[1]
def get_state(self):
return self._get_state()[0]
state = property(get_state)
def enable(self, user=None, base_activity=None):
raise NotImplementedError("Use activate or passivate instead.")
@property
@node_available
def ram_size_with_overcommit(self):
"""Bytes of total memory including overcommit margin.
"""
return self.ram_size * self.overcommit
@method_cache(30)
def get_remote_queue_name(self, queue_id, priority=None):
"""Returns the name of the remote celery queue for this node.
Throws Exception if there is no worker on the queue.
The result may include dead queues because of caching.
"""
if vm_tasks.check_queue(self.host.hostname, queue_id, priority):
queue_name = self.host.hostname + "." + queue_id
if priority is not None:
queue_name = queue_name + "." + priority
self.node_online()
return queue_name
else:
if self.enabled:
self.node_offline()
raise WorkerNotFound()
def node_online(self):
"""Create activity and log entry when node reappears.
"""
try:
act = self.activity_log.order_by('-pk')[0]
except IndexError:
pass # no monitoring activity at all
else:
logger.debug("The last activity was %s" % act)
if act.activity_code.endswith("offline"):
act = NodeActivity.create(code_suffix='monitor_success_online',
node=self, user=None)
act.started = timezone.now()
act.finished = timezone.now()
act.succeeded = True
act.save()
logger.info("Node %s is ONLINE." % self.name)
self.get_info(invalidate_cache=True)
def node_offline(self):
"""Called when a node disappears.
If the node is not already offline, record an activity and a log entry.
"""
try:
act = self.activity_log.order_by('-pk')[0]
except IndexError:
pass # no activity at all
else:
logger.debug("The last activity was %s" % act)
if act.activity_code.endswith("offline"):
return
act = NodeActivity.create(code_suffix='monitor_failed_offline',
node=self, user=None)
act.started = timezone.now()
act.finished = timezone.now()
act.succeeded = False
act.save()
logger.critical("Node %s is OFFLINE%s.", self.name,
", but enabled" if self.enabled else "")
# TODO: check if we should reschedule any VMs?
def remote_query(self, task, timeout=30, priority=None, raise_=False,
default=None):
"""Query the given task, and get the result.
If the result is not ready or worker not reachable
in timeout secs, return default value or raise a
TimeoutError or WorkerNotFound exception.
"""
try:
r = task.apply_async(
queue=self.get_remote_queue_name('vm', priority),
expires=timeout + 60)
return r.get(timeout=timeout)
except (TimeoutError, WorkerNotFound):
if raise_:
raise
else:
return default
@property
@node_available
@method_cache(10)
def monitor_info(self):
metrics = ('cpu.percent', 'memory.usage')
prefix = 'circle.%s.' % self.host.hostname
params = [('target', '%s%s' % (prefix, metric))
for metric in metrics]
params.append(('from', '-5min'))
params.append(('format', 'json'))
try:
logger.info('%s %s', settings.GRAPHITE_URL, params)
response = requests.get(settings.GRAPHITE_URL, params=params)
retval = {}
for target in response.json():
# Example:
# {"target": "circle.szianode.cpu.usage",
# "datapoints": [[0.6, 1403045700], [0.5, 1403045760]
try:
metric = target['target']
if metric.startswith(prefix):
metric = metric[len(prefix):]
else:
continue
value = target['datapoints'][-2][0]
retval[metric] = float(value)
except (KeyError, IndexError, ValueError):
continue
return retval
except:
logger.exception('Unhandled exception: ')
return self.remote_query(vm_tasks.get_node_metrics, timeout=30,
priority="fast")
@property
@node_available
def driver_version(self):
return self.info.get('driver_version')
@property
@node_available
def cpu_usage(self):
return self.monitor_info.get('cpu.percent') / 100
@property
@node_available
def ram_usage(self):
return self.monitor_info.get('memory.usage') / 100
@property
@node_available
def byte_ram_usage(self):
return self.ram_usage * self.ram_size
def get_status_icon(self):
return {
'DISABLED': 'fa-times-circle-o',
'OFFLINE': 'fa-times-circle',
'MISSING': 'fa-warning',
'PASSIVE': 'fa-play-circle-o',
'ACTIVE': 'fa-play-circle'}.get(self.get_state(),
'fa-question-circle')
def get_status_label(self):
return {
'OFFLINE': 'label-warning',
'DISABLED': 'label-danger',
'MISSING': 'label-danger',
'ACTIVE': 'label-success',
'PASSIVE': 'label-warning',
}.get(self.get_state(), 'label-danger')
@node_available
def update_vm_states(self):
"""Update state of Instances running on this Node.
Query state of all libvirt domains, and notify Instances by their
vm_state_changed hook.
"""
domains = {}
domain_list = self.remote_query(vm_tasks.list_domains_info, timeout=5,
priority="fast")
if domain_list is None:
logger.info("Monitoring failed at: %s", self.name)
return
for i in domain_list:
# [{'name': 'cloud-1234', 'state': 'RUNNING', ...}, ...]
try:
id = int(i['name'].split('-')[1])
except:
pass # name format doesn't match
else:
domains[id] = i['state']
instances = [{'id': i.id, 'state': i.state}
for i in self.instance_set.order_by('id').all()]
for i in instances:
try:
d = domains[i['id']]
except KeyError:
logger.info('Node %s update: instance %s missing from '
'libvirt', self, i['id'])
# Set state to STOPPED when instance is missing
self.instance_set.get(id=i['id']).vm_state_changed(
'STOPPED', None)
else:
if d != i['state']:
logger.info('Node %s update: instance %s state changed '
'(libvirt: %s, db: %s)',
self, i['id'], d, i['state'])
self.instance_set.get(id=i['id']).vm_state_changed(d)
del domains[i['id']]
for id, state in domains.iteritems():
from .instance import Instance
logger.error('Node %s update: domain %s in libvirt but not in db.',
self, id)
Instance.objects.get(id=id).vm_state_changed(state, self)
@classmethod
def get_state_count(cls, online, enabled):
return len([1 for i in
cls.objects.filter(enabled=enabled).select_related('host')
if i.online == online])
@permalink
def get_absolute_url(self):
return ('dashboard.views.node-detail', None, {'pk': self.id})
def save(self, *args, **kwargs):
if not self.enabled:
self.schedule_enabled = False
super(Node, self).save(*args, **kwargs)
@property
def metric_prefix(self):
return 'circle.%s' % self.host.hostname