root / tools / burnin @ 1817dca9
History | View | Annotate | Download (38.8 kB)
1 |
#!/usr/bin/python |
---|---|
2 |
# |
3 |
|
4 |
# Copyright (C) 2006, 2007, 2008, 2009, 2010, 2011 Google Inc. |
5 |
# |
6 |
# This program is free software; you can redistribute it and/or modify |
7 |
# it under the terms of the GNU General Public License as published by |
8 |
# the Free Software Foundation; either version 2 of the License, or |
9 |
# (at your option) any later version. |
10 |
# |
11 |
# This program is distributed in the hope that it will be useful, but |
12 |
# WITHOUT ANY WARRANTY; without even the implied warranty of |
13 |
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
14 |
# General Public License for more details. |
15 |
# |
16 |
# You should have received a copy of the GNU General Public License |
17 |
# along with this program; if not, write to the Free Software |
18 |
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA |
19 |
# 02110-1301, USA. |
20 |
|
21 |
|
22 |
"""Burnin program |
23 |
|
24 |
""" |
25 |
|
26 |
import sys |
27 |
import optparse |
28 |
import time |
29 |
import socket |
30 |
import urllib |
31 |
from itertools import izip, islice, cycle |
32 |
from cStringIO import StringIO |
33 |
|
34 |
from ganeti import opcodes |
35 |
from ganeti import constants |
36 |
from ganeti import cli |
37 |
from ganeti import errors |
38 |
from ganeti import utils |
39 |
from ganeti import hypervisor |
40 |
from ganeti import compat |
41 |
|
42 |
from ganeti.confd import client as confd_client |
43 |
|
44 |
|
45 |
USAGE = ("\tburnin -o OS_NAME [options...] instance_name ...") |
46 |
|
47 |
MAX_RETRIES = 3 |
48 |
LOG_HEADERS = { |
49 |
0: "- ", |
50 |
1: "* ", |
51 |
2: "" |
52 |
} |
53 |
|
54 |
|
55 |
class InstanceDown(Exception): |
56 |
"""The checked instance was not up""" |
57 |
|
58 |
|
59 |
class BurninFailure(Exception): |
60 |
"""Failure detected during burning""" |
61 |
|
62 |
|
63 |
def Usage(): |
64 |
"""Shows program usage information and exits the program.""" |
65 |
|
66 |
print >> sys.stderr, "Usage:" |
67 |
print >> sys.stderr, USAGE |
68 |
sys.exit(2) |
69 |
|
70 |
|
71 |
def Log(msg, *args, **kwargs): |
72 |
"""Simple function that prints out its argument. |
73 |
|
74 |
""" |
75 |
if args: |
76 |
msg = msg % args |
77 |
indent = kwargs.get("indent", 0) |
78 |
sys.stdout.write("%*s%s%s\n" % (2 * indent, "", |
79 |
LOG_HEADERS.get(indent, " "), msg)) |
80 |
sys.stdout.flush() |
81 |
|
82 |
|
83 |
def Err(msg, exit_code=1): |
84 |
"""Simple error logging that prints to stderr. |
85 |
|
86 |
""" |
87 |
sys.stderr.write(msg + "\n") |
88 |
sys.stderr.flush() |
89 |
sys.exit(exit_code) |
90 |
|
91 |
|
92 |
class SimpleOpener(urllib.FancyURLopener): |
93 |
"""A simple url opener""" |
94 |
# pylint: disable=W0221 |
95 |
|
96 |
def prompt_user_passwd(self, host, realm, clear_cache=0): |
97 |
"""No-interaction version of prompt_user_passwd.""" |
98 |
# we follow parent class' API |
99 |
# pylint: disable=W0613 |
100 |
return None, None |
101 |
|
102 |
def http_error_default(self, url, fp, errcode, errmsg, headers): |
103 |
"""Custom error handling""" |
104 |
# make sure sockets are not left in CLOSE_WAIT, this is similar |
105 |
# but with a different exception to the BasicURLOpener class |
106 |
_ = fp.read() # throw away data |
107 |
fp.close() |
108 |
raise InstanceDown("HTTP error returned: code %s, msg %s" % |
109 |
(errcode, errmsg)) |
110 |
|
111 |
|
112 |
OPTIONS = [ |
113 |
cli.cli_option("-o", "--os", dest="os", default=None, |
114 |
help="OS to use during burnin", |
115 |
metavar="<OS>", |
116 |
completion_suggest=cli.OPT_COMPL_ONE_OS), |
117 |
cli.HYPERVISOR_OPT, |
118 |
cli.OSPARAMS_OPT, |
119 |
cli.cli_option("--disk-size", dest="disk_size", |
120 |
help="Disk size (determines disk count)", |
121 |
default="128m", type="string", metavar="<size,size,...>", |
122 |
completion_suggest=("128M 512M 1G 4G 1G,256M" |
123 |
" 4G,1G,1G 10G").split()), |
124 |
cli.cli_option("--disk-growth", dest="disk_growth", help="Disk growth", |
125 |
default="128m", type="string", metavar="<size,size,...>"), |
126 |
cli.cli_option("--mem-size", dest="mem_size", help="Memory size", |
127 |
default=128, type="unit", metavar="<size>", |
128 |
completion_suggest=("128M 256M 512M 1G 4G 8G" |
129 |
" 12G 16G").split()), |
130 |
cli.cli_option("--vcpu-count", dest="vcpu_count", help="VCPU count", |
131 |
default=3, type="unit", metavar="<count>", |
132 |
completion_suggest=("1 2 3 4").split()), |
133 |
cli.DEBUG_OPT, |
134 |
cli.VERBOSE_OPT, |
135 |
cli.NOIPCHECK_OPT, |
136 |
cli.NONAMECHECK_OPT, |
137 |
cli.EARLY_RELEASE_OPT, |
138 |
cli.cli_option("--no-replace1", dest="do_replace1", |
139 |
help="Skip disk replacement with the same secondary", |
140 |
action="store_false", default=True), |
141 |
cli.cli_option("--no-replace2", dest="do_replace2", |
142 |
help="Skip disk replacement with a different secondary", |
143 |
action="store_false", default=True), |
144 |
cli.cli_option("--no-failover", dest="do_failover", |
145 |
help="Skip instance failovers", action="store_false", |
146 |
default=True), |
147 |
cli.cli_option("--no-migrate", dest="do_migrate", |
148 |
help="Skip instance live migration", |
149 |
action="store_false", default=True), |
150 |
cli.cli_option("--no-move", dest="do_move", |
151 |
help="Skip instance moves", action="store_false", |
152 |
default=True), |
153 |
cli.cli_option("--no-importexport", dest="do_importexport", |
154 |
help="Skip instance export/import", action="store_false", |
155 |
default=True), |
156 |
cli.cli_option("--no-startstop", dest="do_startstop", |
157 |
help="Skip instance stop/start", action="store_false", |
158 |
default=True), |
159 |
cli.cli_option("--no-reinstall", dest="do_reinstall", |
160 |
help="Skip instance reinstall", action="store_false", |
161 |
default=True), |
162 |
cli.cli_option("--no-reboot", dest="do_reboot", |
163 |
help="Skip instance reboot", action="store_false", |
164 |
default=True), |
165 |
cli.cli_option("--reboot-types", dest="reboot_types", |
166 |
help="Specify the reboot types", default=None), |
167 |
cli.cli_option("--no-activate-disks", dest="do_activate_disks", |
168 |
help="Skip disk activation/deactivation", |
169 |
action="store_false", default=True), |
170 |
cli.cli_option("--no-add-disks", dest="do_addremove_disks", |
171 |
help="Skip disk addition/removal", |
172 |
action="store_false", default=True), |
173 |
cli.cli_option("--no-add-nics", dest="do_addremove_nics", |
174 |
help="Skip NIC addition/removal", |
175 |
action="store_false", default=True), |
176 |
cli.cli_option("--no-nics", dest="nics", |
177 |
help="No network interfaces", action="store_const", |
178 |
const=[], default=[{}]), |
179 |
cli.cli_option("--no-confd", dest="do_confd_tests", |
180 |
help="Skip confd queries", |
181 |
action="store_false", default=True), |
182 |
cli.cli_option("--rename", dest="rename", default=None, |
183 |
help=("Give one unused instance name which is taken" |
184 |
" to start the renaming sequence"), |
185 |
metavar="<instance_name>"), |
186 |
cli.cli_option("-t", "--disk-template", dest="disk_template", |
187 |
choices=list(constants.DISK_TEMPLATES), |
188 |
default=constants.DT_DRBD8, |
189 |
help="Disk template (diskless, file, plain, sharedfile" |
190 |
" or drbd) [drbd]"), |
191 |
cli.cli_option("-n", "--nodes", dest="nodes", default="", |
192 |
help=("Comma separated list of nodes to perform" |
193 |
" the burnin on (defaults to all nodes)"), |
194 |
completion_suggest=cli.OPT_COMPL_MANY_NODES), |
195 |
cli.cli_option("-I", "--iallocator", dest="iallocator", |
196 |
default=None, type="string", |
197 |
help=("Perform the allocation using an iallocator" |
198 |
" instead of fixed node spread (node restrictions no" |
199 |
" longer apply, therefore -n/--nodes must not be" |
200 |
" used"), |
201 |
completion_suggest=cli.OPT_COMPL_ONE_IALLOCATOR), |
202 |
cli.cli_option("-p", "--parallel", default=False, action="store_true", |
203 |
dest="parallel", |
204 |
help=("Enable parallelization of some operations in" |
205 |
" order to speed burnin or to test granular locking")), |
206 |
cli.cli_option("--net-timeout", default=15, type="int", |
207 |
dest="net_timeout", |
208 |
help=("The instance check network timeout in seconds" |
209 |
" (defaults to 15 seconds)"), |
210 |
completion_suggest="15 60 300 900".split()), |
211 |
cli.cli_option("-C", "--http-check", default=False, action="store_true", |
212 |
dest="http_check", |
213 |
help=("Enable checking of instance status via http," |
214 |
" looking for /hostname.txt that should contain the" |
215 |
" name of the instance")), |
216 |
cli.cli_option("-K", "--keep-instances", default=False, |
217 |
action="store_true", |
218 |
dest="keep_instances", |
219 |
help=("Leave instances on the cluster after burnin," |
220 |
" for investigation in case of errors or simply" |
221 |
" to use them")), |
222 |
] |
223 |
|
224 |
# Mainly used for bash completion |
225 |
ARGUMENTS = [cli.ArgInstance(min=1)] |
226 |
|
227 |
|
228 |
def _DoCheckInstances(fn): |
229 |
"""Decorator for checking instances. |
230 |
|
231 |
""" |
232 |
def wrapper(self, *args, **kwargs): |
233 |
val = fn(self, *args, **kwargs) |
234 |
for instance in self.instances: |
235 |
self._CheckInstanceAlive(instance) # pylint: disable=W0212 |
236 |
return val |
237 |
|
238 |
return wrapper |
239 |
|
240 |
|
241 |
def _DoBatch(retry): |
242 |
"""Decorator for possible batch operations. |
243 |
|
244 |
Must come after the _DoCheckInstances decorator (if any). |
245 |
|
246 |
@param retry: whether this is a retryable batch, will be |
247 |
passed to StartBatch |
248 |
|
249 |
""" |
250 |
def wrap(fn): |
251 |
def batched(self, *args, **kwargs): |
252 |
self.StartBatch(retry) |
253 |
val = fn(self, *args, **kwargs) |
254 |
self.CommitQueue() |
255 |
return val |
256 |
return batched |
257 |
|
258 |
return wrap |
259 |
|
260 |
|
261 |
class Burner(object): |
262 |
"""Burner class.""" |
263 |
|
264 |
def __init__(self): |
265 |
"""Constructor.""" |
266 |
self.url_opener = SimpleOpener() |
267 |
self._feed_buf = StringIO() |
268 |
self.nodes = [] |
269 |
self.instances = [] |
270 |
self.to_rem = [] |
271 |
self.queued_ops = [] |
272 |
self.opts = None |
273 |
self.queue_retry = False |
274 |
self.disk_count = self.disk_growth = self.disk_size = None |
275 |
self.hvp = self.bep = None |
276 |
self.ParseOptions() |
277 |
self.cl = cli.GetClient() |
278 |
self.GetState() |
279 |
|
280 |
def ClearFeedbackBuf(self): |
281 |
"""Clear the feedback buffer.""" |
282 |
self._feed_buf.truncate(0) |
283 |
|
284 |
def GetFeedbackBuf(self): |
285 |
"""Return the contents of the buffer.""" |
286 |
return self._feed_buf.getvalue() |
287 |
|
288 |
def Feedback(self, msg): |
289 |
"""Acumulate feedback in our buffer.""" |
290 |
formatted_msg = "%s %s" % (time.ctime(utils.MergeTime(msg[0])), msg[2]) |
291 |
self._feed_buf.write(formatted_msg + "\n") |
292 |
if self.opts.verbose: |
293 |
Log(formatted_msg, indent=3) |
294 |
|
295 |
def MaybeRetry(self, retry_count, msg, fn, *args): |
296 |
"""Possibly retry a given function execution. |
297 |
|
298 |
@type retry_count: int |
299 |
@param retry_count: retry counter: |
300 |
- 0: non-retryable action |
301 |
- 1: last retry for a retryable action |
302 |
- MAX_RETRIES: original try for a retryable action |
303 |
@type msg: str |
304 |
@param msg: the kind of the operation |
305 |
@type fn: callable |
306 |
@param fn: the function to be called |
307 |
|
308 |
""" |
309 |
try: |
310 |
val = fn(*args) |
311 |
if retry_count > 0 and retry_count < MAX_RETRIES: |
312 |
Log("Idempotent %s succeeded after %d retries", |
313 |
msg, MAX_RETRIES - retry_count) |
314 |
return val |
315 |
except Exception, err: # pylint: disable=W0703 |
316 |
if retry_count == 0: |
317 |
Log("Non-idempotent %s failed, aborting", msg) |
318 |
raise |
319 |
elif retry_count == 1: |
320 |
Log("Idempotent %s repeated failure, aborting", msg) |
321 |
raise |
322 |
else: |
323 |
Log("Idempotent %s failed, retry #%d/%d: %s", |
324 |
msg, MAX_RETRIES - retry_count + 1, MAX_RETRIES, err) |
325 |
self.MaybeRetry(retry_count - 1, msg, fn, *args) |
326 |
|
327 |
def _ExecOp(self, *ops): |
328 |
"""Execute one or more opcodes and manage the exec buffer. |
329 |
|
330 |
@return: if only opcode has been passed, we return its result; |
331 |
otherwise we return the list of results |
332 |
|
333 |
""" |
334 |
job_id = cli.SendJob(ops, cl=self.cl) |
335 |
results = cli.PollJob(job_id, cl=self.cl, feedback_fn=self.Feedback) |
336 |
if len(ops) == 1: |
337 |
return results[0] |
338 |
else: |
339 |
return results |
340 |
|
341 |
def ExecOp(self, retry, *ops): |
342 |
"""Execute one or more opcodes and manage the exec buffer. |
343 |
|
344 |
@return: if only opcode has been passed, we return its result; |
345 |
otherwise we return the list of results |
346 |
|
347 |
""" |
348 |
if retry: |
349 |
rval = MAX_RETRIES |
350 |
else: |
351 |
rval = 0 |
352 |
cli.SetGenericOpcodeOpts(ops, self.opts) |
353 |
return self.MaybeRetry(rval, "opcode", self._ExecOp, *ops) |
354 |
|
355 |
def ExecOrQueue(self, name, ops, post_process=None): |
356 |
"""Execute an opcode and manage the exec buffer.""" |
357 |
if self.opts.parallel: |
358 |
cli.SetGenericOpcodeOpts(ops, self.opts) |
359 |
self.queued_ops.append((ops, name, post_process)) |
360 |
else: |
361 |
val = self.ExecOp(self.queue_retry, *ops) # pylint: disable=W0142 |
362 |
if post_process is not None: |
363 |
post_process() |
364 |
return val |
365 |
|
366 |
def StartBatch(self, retry): |
367 |
"""Start a new batch of jobs. |
368 |
|
369 |
@param retry: whether this is a retryable batch |
370 |
|
371 |
""" |
372 |
self.queued_ops = [] |
373 |
self.queue_retry = retry |
374 |
|
375 |
def CommitQueue(self): |
376 |
"""Execute all submitted opcodes in case of parallel burnin""" |
377 |
if not self.opts.parallel or not self.queued_ops: |
378 |
return |
379 |
|
380 |
if self.queue_retry: |
381 |
rval = MAX_RETRIES |
382 |
else: |
383 |
rval = 0 |
384 |
|
385 |
try: |
386 |
results = self.MaybeRetry(rval, "jobset", self.ExecJobSet, |
387 |
self.queued_ops) |
388 |
finally: |
389 |
self.queued_ops = [] |
390 |
return results |
391 |
|
392 |
def ExecJobSet(self, jobs): |
393 |
"""Execute a set of jobs and return once all are done. |
394 |
|
395 |
The method will return the list of results, if all jobs are |
396 |
successful. Otherwise, OpExecError will be raised from within |
397 |
cli.py. |
398 |
|
399 |
""" |
400 |
self.ClearFeedbackBuf() |
401 |
jex = cli.JobExecutor(cl=self.cl, feedback_fn=self.Feedback) |
402 |
for ops, name, _ in jobs: |
403 |
jex.QueueJob(name, *ops) # pylint: disable=W0142 |
404 |
try: |
405 |
results = jex.GetResults() |
406 |
except Exception, err: # pylint: disable=W0703 |
407 |
Log("Jobs failed: %s", err) |
408 |
raise BurninFailure() |
409 |
|
410 |
fail = False |
411 |
val = [] |
412 |
for (_, name, post_process), (success, result) in zip(jobs, results): |
413 |
if success: |
414 |
if post_process: |
415 |
try: |
416 |
post_process() |
417 |
except Exception, err: # pylint: disable=W0703 |
418 |
Log("Post process call for job %s failed: %s", name, err) |
419 |
fail = True |
420 |
val.append(result) |
421 |
else: |
422 |
fail = True |
423 |
|
424 |
if fail: |
425 |
raise BurninFailure() |
426 |
|
427 |
return val |
428 |
|
429 |
def ParseOptions(self): |
430 |
"""Parses the command line options. |
431 |
|
432 |
In case of command line errors, it will show the usage and exit the |
433 |
program. |
434 |
|
435 |
""" |
436 |
parser = optparse.OptionParser(usage="\n%s" % USAGE, |
437 |
version=("%%prog (ganeti) %s" % |
438 |
constants.RELEASE_VERSION), |
439 |
option_list=OPTIONS) |
440 |
|
441 |
options, args = parser.parse_args() |
442 |
if len(args) < 1 or options.os is None: |
443 |
Usage() |
444 |
|
445 |
supported_disk_templates = (constants.DT_DISKLESS, |
446 |
constants.DT_FILE, |
447 |
constants.DT_SHARED_FILE, |
448 |
constants.DT_PLAIN, |
449 |
constants.DT_DRBD8) |
450 |
if options.disk_template not in supported_disk_templates: |
451 |
Err("Unknown disk template '%s'" % options.disk_template) |
452 |
|
453 |
if options.disk_template == constants.DT_DISKLESS: |
454 |
disk_size = disk_growth = [] |
455 |
options.do_addremove_disks = False |
456 |
else: |
457 |
disk_size = [utils.ParseUnit(v) for v in options.disk_size.split(",")] |
458 |
disk_growth = [utils.ParseUnit(v) |
459 |
for v in options.disk_growth.split(",")] |
460 |
if len(disk_growth) != len(disk_size): |
461 |
Err("Wrong disk sizes/growth combination") |
462 |
if ((disk_size and options.disk_template == constants.DT_DISKLESS) or |
463 |
(not disk_size and options.disk_template != constants.DT_DISKLESS)): |
464 |
Err("Wrong disk count/disk template combination") |
465 |
|
466 |
self.disk_size = disk_size |
467 |
self.disk_growth = disk_growth |
468 |
self.disk_count = len(disk_size) |
469 |
|
470 |
if options.nodes and options.iallocator: |
471 |
Err("Give either the nodes option or the iallocator option, not both") |
472 |
|
473 |
if options.http_check and not options.name_check: |
474 |
Err("Can't enable HTTP checks without name checks") |
475 |
|
476 |
self.opts = options |
477 |
self.instances = args |
478 |
self.bep = { |
479 |
constants.BE_MEMORY: options.mem_size, |
480 |
constants.BE_VCPUS: options.vcpu_count, |
481 |
} |
482 |
|
483 |
self.hypervisor = None |
484 |
self.hvp = {} |
485 |
if options.hypervisor: |
486 |
self.hypervisor, self.hvp = options.hypervisor |
487 |
|
488 |
if options.reboot_types is None: |
489 |
options.reboot_types = constants.REBOOT_TYPES |
490 |
else: |
491 |
options.reboot_types = options.reboot_types.split(",") |
492 |
rt_diff = set(options.reboot_types).difference(constants.REBOOT_TYPES) |
493 |
if rt_diff: |
494 |
Err("Invalid reboot types specified: %s" % utils.CommaJoin(rt_diff)) |
495 |
|
496 |
socket.setdefaulttimeout(options.net_timeout) |
497 |
|
498 |
def GetState(self): |
499 |
"""Read the cluster state from the master daemon.""" |
500 |
if self.opts.nodes: |
501 |
names = self.opts.nodes.split(",") |
502 |
else: |
503 |
names = [] |
504 |
try: |
505 |
op = opcodes.OpNodeQuery(output_fields=["name", "offline", "drained"], |
506 |
names=names, use_locking=True) |
507 |
result = self.ExecOp(True, op) |
508 |
except errors.GenericError, err: |
509 |
err_code, msg = cli.FormatError(err) |
510 |
Err(msg, exit_code=err_code) |
511 |
self.nodes = [data[0] for data in result if not (data[1] or data[2])] |
512 |
|
513 |
op_diagnose = opcodes.OpOsDiagnose(output_fields=["name", |
514 |
"variants", |
515 |
"hidden"], |
516 |
names=[]) |
517 |
result = self.ExecOp(True, op_diagnose) |
518 |
|
519 |
if not result: |
520 |
Err("Can't get the OS list") |
521 |
|
522 |
found = False |
523 |
for (name, variants, _) in result: |
524 |
if self.opts.os in cli.CalculateOSNames(name, variants): |
525 |
found = True |
526 |
break |
527 |
|
528 |
if not found: |
529 |
Err("OS '%s' not found" % self.opts.os) |
530 |
|
531 |
cluster_info = self.cl.QueryClusterInfo() |
532 |
self.cluster_info = cluster_info |
533 |
if not self.cluster_info: |
534 |
Err("Can't get cluster info") |
535 |
|
536 |
default_nic_params = self.cluster_info["nicparams"][constants.PP_DEFAULT] |
537 |
self.cluster_default_nicparams = default_nic_params |
538 |
if self.hypervisor is None: |
539 |
self.hypervisor = self.cluster_info["default_hypervisor"] |
540 |
self.hv_class = hypervisor.GetHypervisorClass(self.hypervisor) |
541 |
|
542 |
@_DoCheckInstances |
543 |
@_DoBatch(False) |
544 |
def BurnCreateInstances(self): |
545 |
"""Create the given instances. |
546 |
|
547 |
""" |
548 |
self.to_rem = [] |
549 |
mytor = izip(cycle(self.nodes), |
550 |
islice(cycle(self.nodes), 1, None), |
551 |
self.instances) |
552 |
|
553 |
Log("Creating instances") |
554 |
for pnode, snode, instance in mytor: |
555 |
Log("instance %s", instance, indent=1) |
556 |
if self.opts.iallocator: |
557 |
pnode = snode = None |
558 |
msg = "with iallocator %s" % self.opts.iallocator |
559 |
elif self.opts.disk_template not in constants.DTS_INT_MIRROR: |
560 |
snode = None |
561 |
msg = "on %s" % pnode |
562 |
else: |
563 |
msg = "on %s, %s" % (pnode, snode) |
564 |
|
565 |
Log(msg, indent=2) |
566 |
|
567 |
op = opcodes.OpInstanceCreate(instance_name=instance, |
568 |
disks=[{"size": size} |
569 |
for size in self.disk_size], |
570 |
disk_template=self.opts.disk_template, |
571 |
nics=self.opts.nics, |
572 |
mode=constants.INSTANCE_CREATE, |
573 |
os_type=self.opts.os, |
574 |
pnode=pnode, |
575 |
snode=snode, |
576 |
start=True, |
577 |
ip_check=self.opts.ip_check, |
578 |
name_check=self.opts.name_check, |
579 |
wait_for_sync=True, |
580 |
file_driver="loop", |
581 |
file_storage_dir=None, |
582 |
iallocator=self.opts.iallocator, |
583 |
beparams=self.bep, |
584 |
hvparams=self.hvp, |
585 |
hypervisor=self.hypervisor, |
586 |
osparams=self.opts.osparams, |
587 |
) |
588 |
remove_instance = lambda name: lambda: self.to_rem.append(name) |
589 |
self.ExecOrQueue(instance, [op], post_process=remove_instance(instance)) |
590 |
|
591 |
@_DoBatch(False) |
592 |
def BurnGrowDisks(self): |
593 |
"""Grow both the os and the swap disks by the requested amount, if any.""" |
594 |
Log("Growing disks") |
595 |
for instance in self.instances: |
596 |
Log("instance %s", instance, indent=1) |
597 |
for idx, growth in enumerate(self.disk_growth): |
598 |
if growth > 0: |
599 |
op = opcodes.OpInstanceGrowDisk(instance_name=instance, disk=idx, |
600 |
amount=growth, wait_for_sync=True) |
601 |
Log("increase disk/%s by %s MB", idx, growth, indent=2) |
602 |
self.ExecOrQueue(instance, [op]) |
603 |
|
604 |
@_DoBatch(True) |
605 |
def BurnReplaceDisks1D8(self): |
606 |
"""Replace disks on primary and secondary for drbd8.""" |
607 |
Log("Replacing disks on the same nodes") |
608 |
early_release = self.opts.early_release |
609 |
for instance in self.instances: |
610 |
Log("instance %s", instance, indent=1) |
611 |
ops = [] |
612 |
for mode in constants.REPLACE_DISK_SEC, constants.REPLACE_DISK_PRI: |
613 |
op = opcodes.OpInstanceReplaceDisks(instance_name=instance, |
614 |
mode=mode, |
615 |
disks=list(range(self.disk_count)), |
616 |
early_release=early_release) |
617 |
Log("run %s", mode, indent=2) |
618 |
ops.append(op) |
619 |
self.ExecOrQueue(instance, ops) |
620 |
|
621 |
@_DoBatch(True) |
622 |
def BurnReplaceDisks2(self): |
623 |
"""Replace secondary node.""" |
624 |
Log("Changing the secondary node") |
625 |
mode = constants.REPLACE_DISK_CHG |
626 |
|
627 |
mytor = izip(islice(cycle(self.nodes), 2, None), |
628 |
self.instances) |
629 |
for tnode, instance in mytor: |
630 |
Log("instance %s", instance, indent=1) |
631 |
if self.opts.iallocator: |
632 |
tnode = None |
633 |
msg = "with iallocator %s" % self.opts.iallocator |
634 |
else: |
635 |
msg = tnode |
636 |
op = opcodes.OpInstanceReplaceDisks(instance_name=instance, |
637 |
mode=mode, |
638 |
remote_node=tnode, |
639 |
iallocator=self.opts.iallocator, |
640 |
disks=[], |
641 |
early_release=self.opts.early_release) |
642 |
Log("run %s %s", mode, msg, indent=2) |
643 |
self.ExecOrQueue(instance, [op]) |
644 |
|
645 |
@_DoCheckInstances |
646 |
@_DoBatch(False) |
647 |
def BurnFailover(self): |
648 |
"""Failover the instances.""" |
649 |
Log("Failing over instances") |
650 |
for instance in self.instances: |
651 |
Log("instance %s", instance, indent=1) |
652 |
op = opcodes.OpInstanceFailover(instance_name=instance, |
653 |
ignore_consistency=False) |
654 |
self.ExecOrQueue(instance, [op]) |
655 |
|
656 |
@_DoCheckInstances |
657 |
@_DoBatch(False) |
658 |
def BurnMove(self): |
659 |
"""Move the instances.""" |
660 |
Log("Moving instances") |
661 |
mytor = izip(islice(cycle(self.nodes), 1, None), |
662 |
self.instances) |
663 |
for tnode, instance in mytor: |
664 |
Log("instance %s", instance, indent=1) |
665 |
op = opcodes.OpInstanceMove(instance_name=instance, |
666 |
target_node=tnode) |
667 |
self.ExecOrQueue(instance, [op]) |
668 |
|
669 |
@_DoBatch(False) |
670 |
def BurnMigrate(self): |
671 |
"""Migrate the instances.""" |
672 |
Log("Migrating instances") |
673 |
for instance in self.instances: |
674 |
Log("instance %s", instance, indent=1) |
675 |
op1 = opcodes.OpInstanceMigrate(instance_name=instance, mode=None, |
676 |
cleanup=False) |
677 |
|
678 |
op2 = opcodes.OpInstanceMigrate(instance_name=instance, mode=None, |
679 |
cleanup=True) |
680 |
Log("migration and migration cleanup", indent=2) |
681 |
self.ExecOrQueue(instance, [op1, op2]) |
682 |
|
683 |
@_DoCheckInstances |
684 |
@_DoBatch(False) |
685 |
def BurnImportExport(self): |
686 |
"""Export the instance, delete it, and import it back. |
687 |
|
688 |
""" |
689 |
Log("Exporting and re-importing instances") |
690 |
mytor = izip(cycle(self.nodes), |
691 |
islice(cycle(self.nodes), 1, None), |
692 |
islice(cycle(self.nodes), 2, None), |
693 |
self.instances) |
694 |
|
695 |
for pnode, snode, enode, instance in mytor: |
696 |
Log("instance %s", instance, indent=1) |
697 |
# read the full name of the instance |
698 |
nam_op = opcodes.OpInstanceQuery(output_fields=["name"], |
699 |
names=[instance], use_locking=True) |
700 |
full_name = self.ExecOp(False, nam_op)[0][0] |
701 |
|
702 |
if self.opts.iallocator: |
703 |
pnode = snode = None |
704 |
import_log_msg = ("import from %s" |
705 |
" with iallocator %s" % |
706 |
(enode, self.opts.iallocator)) |
707 |
elif self.opts.disk_template not in constants.DTS_INT_MIRROR: |
708 |
snode = None |
709 |
import_log_msg = ("import from %s to %s" % |
710 |
(enode, pnode)) |
711 |
else: |
712 |
import_log_msg = ("import from %s to %s, %s" % |
713 |
(enode, pnode, snode)) |
714 |
|
715 |
exp_op = opcodes.OpBackupExport(instance_name=instance, |
716 |
target_node=enode, |
717 |
mode=constants.EXPORT_MODE_LOCAL, |
718 |
shutdown=True) |
719 |
rem_op = opcodes.OpInstanceRemove(instance_name=instance, |
720 |
ignore_failures=True) |
721 |
imp_dir = utils.PathJoin(constants.EXPORT_DIR, full_name) |
722 |
imp_op = opcodes.OpInstanceCreate(instance_name=instance, |
723 |
disks=[{"size": size} |
724 |
for size in self.disk_size], |
725 |
disk_template=self.opts.disk_template, |
726 |
nics=self.opts.nics, |
727 |
mode=constants.INSTANCE_IMPORT, |
728 |
src_node=enode, |
729 |
src_path=imp_dir, |
730 |
pnode=pnode, |
731 |
snode=snode, |
732 |
start=True, |
733 |
ip_check=self.opts.ip_check, |
734 |
name_check=self.opts.name_check, |
735 |
wait_for_sync=True, |
736 |
file_storage_dir=None, |
737 |
file_driver="loop", |
738 |
iallocator=self.opts.iallocator, |
739 |
beparams=self.bep, |
740 |
hvparams=self.hvp, |
741 |
osparams=self.opts.osparams, |
742 |
) |
743 |
|
744 |
erem_op = opcodes.OpBackupRemove(instance_name=instance) |
745 |
|
746 |
Log("export to node %s", enode, indent=2) |
747 |
Log("remove instance", indent=2) |
748 |
Log(import_log_msg, indent=2) |
749 |
Log("remove export", indent=2) |
750 |
self.ExecOrQueue(instance, [exp_op, rem_op, imp_op, erem_op]) |
751 |
|
752 |
@staticmethod |
753 |
def StopInstanceOp(instance): |
754 |
"""Stop given instance.""" |
755 |
return opcodes.OpInstanceShutdown(instance_name=instance) |
756 |
|
757 |
@staticmethod |
758 |
def StartInstanceOp(instance): |
759 |
"""Start given instance.""" |
760 |
return opcodes.OpInstanceStartup(instance_name=instance, force=False) |
761 |
|
762 |
@staticmethod |
763 |
def RenameInstanceOp(instance, instance_new): |
764 |
"""Rename instance.""" |
765 |
return opcodes.OpInstanceRename(instance_name=instance, |
766 |
new_name=instance_new) |
767 |
|
768 |
@_DoCheckInstances |
769 |
@_DoBatch(True) |
770 |
def BurnStopStart(self): |
771 |
"""Stop/start the instances.""" |
772 |
Log("Stopping and starting instances") |
773 |
for instance in self.instances: |
774 |
Log("instance %s", instance, indent=1) |
775 |
op1 = self.StopInstanceOp(instance) |
776 |
op2 = self.StartInstanceOp(instance) |
777 |
self.ExecOrQueue(instance, [op1, op2]) |
778 |
|
779 |
@_DoBatch(False) |
780 |
def BurnRemove(self): |
781 |
"""Remove the instances.""" |
782 |
Log("Removing instances") |
783 |
for instance in self.to_rem: |
784 |
Log("instance %s", instance, indent=1) |
785 |
op = opcodes.OpInstanceRemove(instance_name=instance, |
786 |
ignore_failures=True) |
787 |
self.ExecOrQueue(instance, [op]) |
788 |
|
789 |
def BurnRename(self): |
790 |
"""Rename the instances. |
791 |
|
792 |
Note that this function will not execute in parallel, since we |
793 |
only have one target for rename. |
794 |
|
795 |
""" |
796 |
Log("Renaming instances") |
797 |
rename = self.opts.rename |
798 |
for instance in self.instances: |
799 |
Log("instance %s", instance, indent=1) |
800 |
op_stop1 = self.StopInstanceOp(instance) |
801 |
op_stop2 = self.StopInstanceOp(rename) |
802 |
op_rename1 = self.RenameInstanceOp(instance, rename) |
803 |
op_rename2 = self.RenameInstanceOp(rename, instance) |
804 |
op_start1 = self.StartInstanceOp(rename) |
805 |
op_start2 = self.StartInstanceOp(instance) |
806 |
self.ExecOp(False, op_stop1, op_rename1, op_start1) |
807 |
self._CheckInstanceAlive(rename) |
808 |
self.ExecOp(False, op_stop2, op_rename2, op_start2) |
809 |
self._CheckInstanceAlive(instance) |
810 |
|
811 |
@_DoCheckInstances |
812 |
@_DoBatch(True) |
813 |
def BurnReinstall(self): |
814 |
"""Reinstall the instances.""" |
815 |
Log("Reinstalling instances") |
816 |
for instance in self.instances: |
817 |
Log("instance %s", instance, indent=1) |
818 |
op1 = self.StopInstanceOp(instance) |
819 |
op2 = opcodes.OpInstanceReinstall(instance_name=instance) |
820 |
Log("reinstall without passing the OS", indent=2) |
821 |
op3 = opcodes.OpInstanceReinstall(instance_name=instance, |
822 |
os_type=self.opts.os) |
823 |
Log("reinstall specifying the OS", indent=2) |
824 |
op4 = self.StartInstanceOp(instance) |
825 |
self.ExecOrQueue(instance, [op1, op2, op3, op4]) |
826 |
|
827 |
@_DoCheckInstances |
828 |
@_DoBatch(True) |
829 |
def BurnReboot(self): |
830 |
"""Reboot the instances.""" |
831 |
Log("Rebooting instances") |
832 |
for instance in self.instances: |
833 |
Log("instance %s", instance, indent=1) |
834 |
ops = [] |
835 |
for reboot_type in self.opts.reboot_types: |
836 |
op = opcodes.OpInstanceReboot(instance_name=instance, |
837 |
reboot_type=reboot_type, |
838 |
ignore_secondaries=False) |
839 |
Log("reboot with type '%s'", reboot_type, indent=2) |
840 |
ops.append(op) |
841 |
self.ExecOrQueue(instance, ops) |
842 |
|
843 |
@_DoCheckInstances |
844 |
@_DoBatch(True) |
845 |
def BurnActivateDisks(self): |
846 |
"""Activate and deactivate disks of the instances.""" |
847 |
Log("Activating/deactivating disks") |
848 |
for instance in self.instances: |
849 |
Log("instance %s", instance, indent=1) |
850 |
op_start = self.StartInstanceOp(instance) |
851 |
op_act = opcodes.OpInstanceActivateDisks(instance_name=instance) |
852 |
op_deact = opcodes.OpInstanceDeactivateDisks(instance_name=instance) |
853 |
op_stop = self.StopInstanceOp(instance) |
854 |
Log("activate disks when online", indent=2) |
855 |
Log("activate disks when offline", indent=2) |
856 |
Log("deactivate disks (when offline)", indent=2) |
857 |
self.ExecOrQueue(instance, [op_act, op_stop, op_act, op_deact, op_start]) |
858 |
|
859 |
@_DoCheckInstances |
860 |
@_DoBatch(False) |
861 |
def BurnAddRemoveDisks(self): |
862 |
"""Add and remove an extra disk for the instances.""" |
863 |
Log("Adding and removing disks") |
864 |
for instance in self.instances: |
865 |
Log("instance %s", instance, indent=1) |
866 |
op_add = opcodes.OpInstanceSetParams(\ |
867 |
instance_name=instance, |
868 |
disks=[(constants.DDM_ADD, {"size": self.disk_size[0]})]) |
869 |
op_rem = opcodes.OpInstanceSetParams(\ |
870 |
instance_name=instance, disks=[(constants.DDM_REMOVE, {})]) |
871 |
op_stop = self.StopInstanceOp(instance) |
872 |
op_start = self.StartInstanceOp(instance) |
873 |
Log("adding a disk", indent=2) |
874 |
Log("removing last disk", indent=2) |
875 |
self.ExecOrQueue(instance, [op_add, op_stop, op_rem, op_start]) |
876 |
|
877 |
@_DoBatch(False) |
878 |
def BurnAddRemoveNICs(self): |
879 |
"""Add and remove an extra NIC for the instances.""" |
880 |
Log("Adding and removing NICs") |
881 |
for instance in self.instances: |
882 |
Log("instance %s", instance, indent=1) |
883 |
op_add = opcodes.OpInstanceSetParams(\ |
884 |
instance_name=instance, nics=[(constants.DDM_ADD, {})]) |
885 |
op_rem = opcodes.OpInstanceSetParams(\ |
886 |
instance_name=instance, nics=[(constants.DDM_REMOVE, {})]) |
887 |
Log("adding a NIC", indent=2) |
888 |
Log("removing last NIC", indent=2) |
889 |
self.ExecOrQueue(instance, [op_add, op_rem]) |
890 |
|
891 |
def ConfdCallback(self, reply): |
892 |
"""Callback for confd queries""" |
893 |
if reply.type == confd_client.UPCALL_REPLY: |
894 |
if reply.server_reply.status != constants.CONFD_REPL_STATUS_OK: |
895 |
Err("Query %s gave non-ok status %s: %s" % (reply.orig_request, |
896 |
reply.server_reply.status, |
897 |
reply.server_reply)) |
898 |
if reply.orig_request.type == constants.CONFD_REQ_PING: |
899 |
Log("Ping: OK", indent=1) |
900 |
elif reply.orig_request.type == constants.CONFD_REQ_CLUSTER_MASTER: |
901 |
if reply.server_reply.answer == self.cluster_info["master"]: |
902 |
Log("Master: OK", indent=1) |
903 |
else: |
904 |
Err("Master: wrong: %s" % reply.server_reply.answer) |
905 |
elif reply.orig_request.type == constants.CONFD_REQ_NODE_ROLE_BYNAME: |
906 |
if reply.server_reply.answer == constants.CONFD_NODE_ROLE_MASTER: |
907 |
Log("Node role for master: OK", indent=1) |
908 |
else: |
909 |
Err("Node role for master: wrong: %s" % reply.server_reply.answer) |
910 |
|
911 |
def DoConfdRequestReply(self, req): |
912 |
self.confd_counting_callback.RegisterQuery(req.rsalt) |
913 |
self.confd_client.SendRequest(req, async=False) |
914 |
while not self.confd_counting_callback.AllAnswered(): |
915 |
if not self.confd_client.ReceiveReply(): |
916 |
Err("Did not receive all expected confd replies") |
917 |
break |
918 |
|
919 |
def BurnConfd(self): |
920 |
"""Run confd queries for our instances. |
921 |
|
922 |
The following confd queries are tested: |
923 |
- CONFD_REQ_PING: simple ping |
924 |
- CONFD_REQ_CLUSTER_MASTER: cluster master |
925 |
- CONFD_REQ_NODE_ROLE_BYNAME: node role, for the master |
926 |
|
927 |
""" |
928 |
Log("Checking confd results") |
929 |
|
930 |
filter_callback = confd_client.ConfdFilterCallback(self.ConfdCallback) |
931 |
counting_callback = confd_client.ConfdCountingCallback(filter_callback) |
932 |
self.confd_counting_callback = counting_callback |
933 |
|
934 |
self.confd_client = confd_client.GetConfdClient(counting_callback) |
935 |
|
936 |
req = confd_client.ConfdClientRequest(type=constants.CONFD_REQ_PING) |
937 |
self.DoConfdRequestReply(req) |
938 |
|
939 |
req = confd_client.ConfdClientRequest( |
940 |
type=constants.CONFD_REQ_CLUSTER_MASTER) |
941 |
self.DoConfdRequestReply(req) |
942 |
|
943 |
req = confd_client.ConfdClientRequest( |
944 |
type=constants.CONFD_REQ_NODE_ROLE_BYNAME, |
945 |
query=self.cluster_info["master"]) |
946 |
self.DoConfdRequestReply(req) |
947 |
|
948 |
def _CheckInstanceAlive(self, instance): |
949 |
"""Check if an instance is alive by doing http checks. |
950 |
|
951 |
This will try to retrieve the url on the instance /hostname.txt |
952 |
and check that it contains the hostname of the instance. In case |
953 |
we get ECONNREFUSED, we retry up to the net timeout seconds, for |
954 |
any other error we abort. |
955 |
|
956 |
""" |
957 |
if not self.opts.http_check: |
958 |
return |
959 |
end_time = time.time() + self.opts.net_timeout |
960 |
url = None |
961 |
while time.time() < end_time and url is None: |
962 |
try: |
963 |
url = self.url_opener.open("http://%s/hostname.txt" % instance) |
964 |
except IOError: |
965 |
# here we can have connection refused, no route to host, etc. |
966 |
time.sleep(1) |
967 |
if url is None: |
968 |
raise InstanceDown(instance, "Cannot contact instance") |
969 |
hostname = url.read().strip() |
970 |
url.close() |
971 |
if hostname != instance: |
972 |
raise InstanceDown(instance, ("Hostname mismatch, expected %s, got %s" % |
973 |
(instance, hostname))) |
974 |
|
975 |
def BurninCluster(self): |
976 |
"""Test a cluster intensively. |
977 |
|
978 |
This will create instances and then start/stop/failover them. |
979 |
It is safe for existing instances but could impact performance. |
980 |
|
981 |
""" |
982 |
|
983 |
opts = self.opts |
984 |
|
985 |
Log("Testing global parameters") |
986 |
|
987 |
if (len(self.nodes) == 1 and |
988 |
opts.disk_template not in (constants.DT_DISKLESS, constants.DT_PLAIN, |
989 |
constants.DT_FILE, |
990 |
constants.DT_SHARED_FILE)): |
991 |
Err("When one node is available/selected the disk template must" |
992 |
" be 'diskless', 'file' or 'plain'") |
993 |
|
994 |
has_err = True |
995 |
try: |
996 |
self.BurnCreateInstances() |
997 |
if opts.do_replace1 and opts.disk_template in constants.DTS_INT_MIRROR: |
998 |
self.BurnReplaceDisks1D8() |
999 |
if (opts.do_replace2 and len(self.nodes) > 2 and |
1000 |
opts.disk_template in constants.DTS_INT_MIRROR): |
1001 |
self.BurnReplaceDisks2() |
1002 |
|
1003 |
if (opts.disk_template in constants.DTS_GROWABLE and |
1004 |
compat.any(n > 0 for n in self.disk_growth)): |
1005 |
self.BurnGrowDisks() |
1006 |
|
1007 |
if opts.do_failover and opts.disk_template in constants.DTS_MIRRORED: |
1008 |
self.BurnFailover() |
1009 |
|
1010 |
if opts.do_migrate: |
1011 |
if opts.disk_template not in constants.DTS_MIRRORED: |
1012 |
Log("Skipping migration (disk template %s does not support it)", |
1013 |
opts.disk_template) |
1014 |
elif not self.hv_class.CAN_MIGRATE: |
1015 |
Log("Skipping migration (hypervisor %s does not support it)", |
1016 |
self.hypervisor) |
1017 |
else: |
1018 |
self.BurnMigrate() |
1019 |
|
1020 |
if (opts.do_move and len(self.nodes) > 1 and |
1021 |
opts.disk_template in [constants.DT_PLAIN, constants.DT_FILE]): |
1022 |
self.BurnMove() |
1023 |
|
1024 |
if (opts.do_importexport and |
1025 |
opts.disk_template not in (constants.DT_DISKLESS, |
1026 |
constants.DT_SHARED_FILE, |
1027 |
constants.DT_FILE)): |
1028 |
self.BurnImportExport() |
1029 |
|
1030 |
if opts.do_reinstall: |
1031 |
self.BurnReinstall() |
1032 |
|
1033 |
if opts.do_reboot: |
1034 |
self.BurnReboot() |
1035 |
|
1036 |
if opts.do_addremove_disks: |
1037 |
self.BurnAddRemoveDisks() |
1038 |
|
1039 |
default_nic_mode = self.cluster_default_nicparams[constants.NIC_MODE] |
1040 |
# Don't add/remove nics in routed mode, as we would need an ip to add |
1041 |
# them with |
1042 |
if opts.do_addremove_nics: |
1043 |
if default_nic_mode == constants.NIC_MODE_BRIDGED: |
1044 |
self.BurnAddRemoveNICs() |
1045 |
else: |
1046 |
Log("Skipping nic add/remove as the cluster is not in bridged mode") |
1047 |
|
1048 |
if opts.do_activate_disks: |
1049 |
self.BurnActivateDisks() |
1050 |
|
1051 |
if opts.rename: |
1052 |
self.BurnRename() |
1053 |
|
1054 |
if opts.do_confd_tests: |
1055 |
self.BurnConfd() |
1056 |
|
1057 |
if opts.do_startstop: |
1058 |
self.BurnStopStart() |
1059 |
|
1060 |
has_err = False |
1061 |
finally: |
1062 |
if has_err: |
1063 |
Log("Error detected: opcode buffer follows:\n\n") |
1064 |
Log(self.GetFeedbackBuf()) |
1065 |
Log("\n\n") |
1066 |
if not self.opts.keep_instances: |
1067 |
try: |
1068 |
self.BurnRemove() |
1069 |
except Exception, err: # pylint: disable=W0703 |
1070 |
if has_err: # already detected errors, so errors in removal |
1071 |
# are quite expected |
1072 |
Log("Note: error detected during instance remove: %s", err) |
1073 |
else: # non-expected error |
1074 |
raise |
1075 |
|
1076 |
return constants.EXIT_SUCCESS |
1077 |
|
1078 |
|
1079 |
def main(): |
1080 |
"""Main function. |
1081 |
|
1082 |
""" |
1083 |
utils.SetupLogging(constants.LOG_BURNIN, sys.argv[0], |
1084 |
debug=False, stderr_logging=True) |
1085 |
|
1086 |
return Burner().BurninCluster() |
1087 |
|
1088 |
|
1089 |
if __name__ == "__main__": |
1090 |
main() |