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
|
# vim: tabstop=4 shiftwidth=4 softtabstop=4
# Copyright 2010 United States Government as represented by the
# Administrator of the National Aeronautics and Space Administration.
# Copyright 2011 Justin Santa Barbara
# All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License"); you may
# not use this file except in compliance with the License. You may obtain
# a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
# License for the specific language governing permissions and limitations
# under the License.
"""Generic Node base class for all workers that run on hosts."""
import errno
import os
import random
import signal
import sys
import time
import eventlet
import logging as std_logging
from openstack.common import cfg
from openstack.common import eventlet_backdoor
from openstack.common.gettextutils import _
from openstack.common import importutils
from openstack.common import log as logging
from openstack.common import threadgroup
rpc = importutils.try_import('openstack.common.rpc')
CONF = cfg.CONF
LOG = logging.getLogger(__name__)
class Launcher(object):
"""Launch one or more services and wait for them to complete."""
def __init__(self):
"""Initialize the service launcher.
:returns: None
"""
self._services = threadgroup.ThreadGroup()
eventlet_backdoor.initialize_if_enabled()
@staticmethod
def run_service(service):
"""Start and wait for a service to finish.
:param service: service to run and wait for.
:returns: None
"""
service.start()
service.wait()
def launch_service(self, service):
"""Load and start the given service.
:param service: The service you would like to start.
:returns: None
"""
self._services.add_thread(self.run_service, service)
def stop(self):
"""Stop all services which are currently running.
:returns: None
"""
self._services.stop()
def wait(self):
"""Waits until all services have been stopped, and then returns.
:returns: None
"""
self._services.wait()
class SignalExit(SystemExit):
def __init__(self, signo, exccode=1):
super(SignalExit, self).__init__(exccode)
self.signo = signo
class ServiceLauncher(Launcher):
def _handle_signal(self, signo, frame):
# Allow the process to be killed again and die from natural causes
signal.signal(signal.SIGTERM, signal.SIG_DFL)
signal.signal(signal.SIGINT, signal.SIG_DFL)
raise SignalExit(signo)
def wait(self):
signal.signal(signal.SIGTERM, self._handle_signal)
signal.signal(signal.SIGINT, self._handle_signal)
LOG.debug(_('Full set of CONF:'))
CONF.log_opt_values(LOG, std_logging.DEBUG)
status = None
try:
super(ServiceLauncher, self).wait()
except SignalExit as exc:
signame = {signal.SIGTERM: 'SIGTERM',
signal.SIGINT: 'SIGINT'}[exc.signo]
LOG.info(_('Caught %s, exiting'), signame)
status = exc.code
except SystemExit as exc:
status = exc.code
finally:
if rpc:
rpc.cleanup()
self.stop()
return status
class ServiceWrapper(object):
def __init__(self, service, workers):
self.service = service
self.workers = workers
self.children = set()
self.forktimes = []
class ProcessLauncher(object):
def __init__(self):
self.children = {}
self.sigcaught = None
self.running = True
rfd, self.writepipe = os.pipe()
self.readpipe = eventlet.greenio.GreenPipe(rfd, 'r')
signal.signal(signal.SIGTERM, self._handle_signal)
signal.signal(signal.SIGINT, self._handle_signal)
def _handle_signal(self, signo, frame):
self.sigcaught = signo
self.running = False
# Allow the process to be killed again and die from natural causes
signal.signal(signal.SIGTERM, signal.SIG_DFL)
signal.signal(signal.SIGINT, signal.SIG_DFL)
def _pipe_watcher(self):
# This will block until the write end is closed when the parent
# dies unexpectedly
self.readpipe.read()
LOG.info(_('Parent process has died unexpectedly, exiting'))
sys.exit(1)
def _child_process(self, service):
# Setup child signal handlers differently
def _sigterm(*args):
signal.signal(signal.SIGTERM, signal.SIG_DFL)
raise SignalExit(signal.SIGTERM)
signal.signal(signal.SIGTERM, _sigterm)
# Block SIGINT and let the parent send us a SIGTERM
signal.signal(signal.SIGINT, signal.SIG_IGN)
# Reopen the eventlet hub to make sure we don't share an epoll
# fd with parent and/or siblings, which would be bad
eventlet.hubs.use_hub()
# Close write to ensure only parent has it open
os.close(self.writepipe)
# Create greenthread to watch for parent to close pipe
eventlet.spawn_n(self._pipe_watcher)
# Reseed random number generator
random.seed()
launcher = Launcher()
launcher.run_service(service)
def _start_child(self, wrap):
if len(wrap.forktimes) > wrap.workers:
# Limit ourselves to one process a second (over the period of
# number of workers * 1 second). This will allow workers to
# start up quickly but ensure we don't fork off children that
# die instantly too quickly.
if time.time() - wrap.forktimes[0] < wrap.workers:
LOG.info(_('Forking too fast, sleeping'))
time.sleep(1)
wrap.forktimes.pop(0)
wrap.forktimes.append(time.time())
pid = os.fork()
if pid == 0:
# NOTE(johannes): All exceptions are caught to ensure this
# doesn't fallback into the loop spawning children. It would
# be bad for a child to spawn more children.
status = 0
try:
self._child_process(wrap.service)
except SignalExit as exc:
signame = {signal.SIGTERM: 'SIGTERM',
signal.SIGINT: 'SIGINT'}[exc.signo]
LOG.info(_('Caught %s, exiting'), signame)
status = exc.code
except SystemExit as exc:
status = exc.code
except BaseException:
LOG.exception(_('Unhandled exception'))
status = 2
finally:
wrap.service.stop()
os._exit(status)
LOG.info(_('Started child %d'), pid)
wrap.children.add(pid)
self.children[pid] = wrap
return pid
def launch_service(self, service, workers=1):
wrap = ServiceWrapper(service, workers)
LOG.info(_('Starting %d workers'), wrap.workers)
while self.running and len(wrap.children) < wrap.workers:
self._start_child(wrap)
def _wait_child(self):
try:
# Don't block if no child processes have exited
pid, status = os.waitpid(0, os.WNOHANG)
if not pid:
return None
except OSError as exc:
if exc.errno not in (errno.EINTR, errno.ECHILD):
raise
return None
if os.WIFSIGNALED(status):
sig = os.WTERMSIG(status)
LOG.info(_('Child %(pid)d killed by signal %(sig)d'),
dict(pid=pid, sig=sig))
else:
code = os.WEXITSTATUS(status)
LOG.info(_('Child %(pid)s exited with status %(code)d'),
dict(pid=pid, code=code))
if pid not in self.children:
LOG.warning(_('pid %d not in child list'), pid)
return None
wrap = self.children.pop(pid)
wrap.children.remove(pid)
return wrap
def wait(self):
"""Loop waiting on children to die and respawning as necessary"""
LOG.debug(_('Full set of CONF:'))
CONF.log_opt_values(LOG, std_logging.DEBUG)
while self.running:
wrap = self._wait_child()
if not wrap:
# Yield to other threads if no children have exited
# Sleep for a short time to avoid excessive CPU usage
# (see bug #1095346)
eventlet.greenthread.sleep(.01)
continue
while self.running and len(wrap.children) < wrap.workers:
self._start_child(wrap)
if self.sigcaught:
signame = {signal.SIGTERM: 'SIGTERM',
signal.SIGINT: 'SIGINT'}[self.sigcaught]
LOG.info(_('Caught %s, stopping children'), signame)
for pid in self.children:
try:
os.kill(pid, signal.SIGTERM)
except OSError as exc:
if exc.errno != errno.ESRCH:
raise
# Wait for children to die
if self.children:
LOG.info(_('Waiting on %d children to exit'), len(self.children))
while self.children:
self._wait_child()
class Service(object):
"""Service object for binaries running on hosts."""
def __init__(self, threads=1000):
self.tg = threadgroup.ThreadGroup(threads)
def start(self):
pass
def stop(self):
self.tg.stop()
def wait(self):
self.tg.wait()
def launch(service, workers=None):
if workers:
launcher = ProcessLauncher()
launcher.launch_service(service, workers=workers)
else:
launcher = ServiceLauncher()
launcher.launch_service(service)
return launcher
|