root / tools / burnin @ 08db7c5c
History | View | Annotate | Download (18.2 kB)
1 |
#!/usr/bin/python |
---|---|
2 |
# |
3 |
|
4 |
# Copyright (C) 2006, 2007 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 os |
27 |
import sys |
28 |
import optparse |
29 |
import time |
30 |
from itertools import izip, islice, cycle |
31 |
from cStringIO import StringIO |
32 |
|
33 |
from ganeti import opcodes |
34 |
from ganeti import mcpu |
35 |
from ganeti import constants |
36 |
from ganeti import cli |
37 |
from ganeti import errors |
38 |
from ganeti import utils |
39 |
|
40 |
|
41 |
USAGE = ("\tburnin -o OS_NAME [options...] instance_name ...") |
42 |
|
43 |
|
44 |
def Usage(): |
45 |
"""Shows program usage information and exits the program.""" |
46 |
|
47 |
print >> sys.stderr, "Usage:" |
48 |
print >> sys.stderr, USAGE |
49 |
sys.exit(2) |
50 |
|
51 |
|
52 |
def Log(msg): |
53 |
"""Simple function that prints out its argument. |
54 |
|
55 |
""" |
56 |
print msg |
57 |
sys.stdout.flush() |
58 |
|
59 |
|
60 |
class Burner(object): |
61 |
"""Burner class.""" |
62 |
|
63 |
def __init__(self): |
64 |
"""Constructor.""" |
65 |
utils.SetupLogging(constants.LOG_BURNIN, debug=False, stderr_logging=True) |
66 |
self._feed_buf = StringIO() |
67 |
self.nodes = [] |
68 |
self.instances = [] |
69 |
self.to_rem = [] |
70 |
self.opts = None |
71 |
self.cl = cli.GetClient() |
72 |
self.ParseOptions() |
73 |
self.GetState() |
74 |
|
75 |
def ClearFeedbackBuf(self): |
76 |
"""Clear the feedback buffer.""" |
77 |
self._feed_buf.truncate(0) |
78 |
|
79 |
def GetFeedbackBuf(self): |
80 |
"""Return the contents of the buffer.""" |
81 |
return self._feed_buf.getvalue() |
82 |
|
83 |
def Feedback(self, msg): |
84 |
"""Acumulate feedback in our buffer.""" |
85 |
self._feed_buf.write("%s %s\n" % (time.ctime(utils.MergeTime(msg[0])), |
86 |
msg[2])) |
87 |
if self.opts.verbose: |
88 |
Log(msg) |
89 |
|
90 |
def ExecOp(self, op): |
91 |
"""Execute an opcode and manage the exec buffer.""" |
92 |
self.ClearFeedbackBuf() |
93 |
return cli.SubmitOpCode(op, feedback_fn=self.Feedback, cl=self.cl) |
94 |
|
95 |
def ExecJobSet(self, jobs): |
96 |
"""Execute a set of jobs and return once all are done. |
97 |
|
98 |
The method will return the list of results, if all jobs are |
99 |
successfull. Otherwise, OpExecError will be raised from within |
100 |
cli.py. |
101 |
|
102 |
""" |
103 |
self.ClearFeedbackBuf() |
104 |
job_ids = [cli.SendJob(job, cl=self.cl) for job in jobs] |
105 |
Log("- Submitted job IDs %s" % ", ".join(job_ids)) |
106 |
results = [] |
107 |
for jid in job_ids: |
108 |
Log("- Waiting for job %s" % jid) |
109 |
results.append(cli.PollJob(jid, cl=self.cl, feedback_fn=self.Feedback)) |
110 |
|
111 |
return results |
112 |
|
113 |
def ParseOptions(self): |
114 |
"""Parses the command line options. |
115 |
|
116 |
In case of command line errors, it will show the usage and exit the |
117 |
program. |
118 |
|
119 |
""" |
120 |
|
121 |
parser = optparse.OptionParser(usage="\n%s" % USAGE, |
122 |
version="%%prog (ganeti) %s" % |
123 |
constants.RELEASE_VERSION, |
124 |
option_class=cli.CliOption) |
125 |
|
126 |
parser.add_option("-o", "--os", dest="os", default=None, |
127 |
help="OS to use during burnin", |
128 |
metavar="<OS>") |
129 |
parser.add_option("--disk-size", dest="disk_size", |
130 |
help="Disk size (determines disk count)", |
131 |
default="128m", type="string", metavar="<size,size,...>") |
132 |
parser.add_option("--disk-growth", dest="disk_growth", help="Disk growth", |
133 |
default=128, type="string", metavar="<size,size,...>") |
134 |
parser.add_option("--mem-size", dest="mem_size", help="Memory size", |
135 |
default=128, type="unit", metavar="<size>") |
136 |
parser.add_option("-v", "--verbose", |
137 |
action="store_true", dest="verbose", default=False, |
138 |
help="print command execution messages to stdout") |
139 |
parser.add_option("--no-replace1", dest="do_replace1", |
140 |
help="Skip disk replacement with the same secondary", |
141 |
action="store_false", default=True) |
142 |
parser.add_option("--no-replace2", dest="do_replace2", |
143 |
help="Skip disk replacement with a different secondary", |
144 |
action="store_false", default=True) |
145 |
parser.add_option("--no-failover", dest="do_failover", |
146 |
help="Skip instance failovers", action="store_false", |
147 |
default=True) |
148 |
parser.add_option("--no-importexport", dest="do_importexport", |
149 |
help="Skip instance export/import", action="store_false", |
150 |
default=True) |
151 |
parser.add_option("--no-startstop", dest="do_startstop", |
152 |
help="Skip instance stop/start", action="store_false", |
153 |
default=True) |
154 |
parser.add_option("--rename", dest="rename", default=None, |
155 |
help="Give one unused instance name which is taken" |
156 |
" to start the renaming sequence", |
157 |
metavar="<instance_name>") |
158 |
parser.add_option("-t", "--disk-template", dest="disk_template", |
159 |
choices=("diskless", "file", "plain", "drbd"), |
160 |
default="drbd", |
161 |
help="Disk template (diskless, file, plain or drbd)" |
162 |
" [drbd]") |
163 |
parser.add_option("-n", "--nodes", dest="nodes", default="", |
164 |
help="Comma separated list of nodes to perform" |
165 |
" the burnin on (defaults to all nodes)") |
166 |
parser.add_option("--iallocator", dest="iallocator", |
167 |
default=None, type="string", |
168 |
help="Perform the allocation using an iallocator" |
169 |
" instead of fixed node spread (node restrictions no" |
170 |
" longer apply, therefore -n/--nodes must not be used") |
171 |
parser.add_option("-p", "--parallel", default=False, action="store_true", |
172 |
dest="parallel", |
173 |
help="Enable parallelization of some operations in" |
174 |
" order to speed burnin or to test granular locking") |
175 |
|
176 |
options, args = parser.parse_args() |
177 |
if len(args) < 1 or options.os is None: |
178 |
Usage() |
179 |
|
180 |
supported_disk_templates = (constants.DT_DISKLESS, |
181 |
constants.DT_FILE, |
182 |
constants.DT_PLAIN, |
183 |
constants.DT_DRBD8) |
184 |
if options.disk_template not in supported_disk_templates: |
185 |
Log("Unknown disk template '%s'" % options.disk_template) |
186 |
sys.exit(1) |
187 |
|
188 |
disk_size = [utils.ParseUnit(v) for v in options.disk_size.split(",")] |
189 |
disk_growth = [utils.ParseUnit(v) for v in options.disk_growth.split(",")] |
190 |
if len(disk_growth) != len(disk_size): |
191 |
Log("Wrong disk sizes/growth combination") |
192 |
sys.exit(1) |
193 |
if ((disk_size and options.disk_template == constants.DT_DISKLESS) or |
194 |
(not disk_size and options.disk_template != constants.DT_DISKLESS)): |
195 |
Log("Wrong disk count/disk template combination") |
196 |
sys.exit(1) |
197 |
|
198 |
self.disk_size = disk_size |
199 |
self.disk_growth = disk_growth |
200 |
self.disk_count = len(disk_size) |
201 |
|
202 |
if options.nodes and options.iallocator: |
203 |
Log("Give either the nodes option or the iallocator option, not both") |
204 |
sys.exit(1) |
205 |
|
206 |
self.opts = options |
207 |
self.instances = args |
208 |
self.bep = { |
209 |
constants.BE_MEMORY: options.mem_size, |
210 |
constants.BE_VCPUS: 1, |
211 |
} |
212 |
self.hvp = {} |
213 |
|
214 |
def GetState(self): |
215 |
"""Read the cluster state from the config.""" |
216 |
if self.opts.nodes: |
217 |
names = self.opts.nodes.split(",") |
218 |
else: |
219 |
names = [] |
220 |
try: |
221 |
op = opcodes.OpQueryNodes(output_fields=["name"], names=names) |
222 |
result = self.ExecOp(op) |
223 |
except errors.GenericError, err: |
224 |
err_code, msg = cli.FormatError(err) |
225 |
Log(msg) |
226 |
sys.exit(err_code) |
227 |
self.nodes = [data[0] for data in result] |
228 |
|
229 |
result = self.ExecOp(opcodes.OpDiagnoseOS(output_fields=["name", "valid"], |
230 |
names=[])) |
231 |
|
232 |
if not result: |
233 |
Log("Can't get the OS list") |
234 |
sys.exit(1) |
235 |
|
236 |
# filter non-valid OS-es |
237 |
os_set = [val[0] for val in result if val[1]] |
238 |
|
239 |
if self.opts.os not in os_set: |
240 |
Log("OS '%s' not found" % self.opts.os) |
241 |
sys.exit(1) |
242 |
|
243 |
def CreateInstances(self): |
244 |
"""Create the given instances. |
245 |
|
246 |
""" |
247 |
self.to_rem = [] |
248 |
mytor = izip(cycle(self.nodes), |
249 |
islice(cycle(self.nodes), 1, None), |
250 |
self.instances) |
251 |
jobset = [] |
252 |
|
253 |
for pnode, snode, instance in mytor: |
254 |
if self.opts.iallocator: |
255 |
pnode = snode = None |
256 |
Log("- Add instance %s (iallocator: %s)" % |
257 |
(instance, self.opts.iallocator)) |
258 |
elif self.opts.disk_template not in constants.DTS_NET_MIRROR: |
259 |
snode = None |
260 |
Log("- Add instance %s on node %s" % (instance, pnode)) |
261 |
else: |
262 |
Log("- Add instance %s on nodes %s/%s" % (instance, pnode, snode)) |
263 |
|
264 |
op = opcodes.OpCreateInstance(instance_name=instance, |
265 |
disks = [ {"size": size} |
266 |
for size in self.disk_size], |
267 |
disk_template=self.opts.disk_template, |
268 |
nics=[{}], |
269 |
mode=constants.INSTANCE_CREATE, |
270 |
os_type=self.opts.os, |
271 |
pnode=pnode, |
272 |
snode=snode, |
273 |
start=True, |
274 |
ip_check=True, |
275 |
wait_for_sync=True, |
276 |
file_driver="loop", |
277 |
file_storage_dir=None, |
278 |
iallocator=self.opts.iallocator, |
279 |
beparams=self.bep, |
280 |
hvparams=self.hvp, |
281 |
) |
282 |
|
283 |
if self.opts.parallel: |
284 |
jobset.append([op]) |
285 |
# FIXME: here we should not append to to_rem uncoditionally, |
286 |
# but only when the job is successful |
287 |
self.to_rem.append(instance) |
288 |
else: |
289 |
self.ExecOp(op) |
290 |
self.to_rem.append(instance) |
291 |
if self.opts.parallel: |
292 |
self.ExecJobSet(jobset) |
293 |
|
294 |
def GrowDisks(self): |
295 |
"""Grow both the os and the swap disks by the requested amount, if any.""" |
296 |
for instance in self.instances: |
297 |
for idx, growth in enumerate(self.disk_growth): |
298 |
if growth > 0: |
299 |
op = opcodes.OpGrowDisk(instance_name=instance, disk=idx, |
300 |
amount=growth, wait_for_sync=True) |
301 |
Log("- Increase %s's %s disk by %s MB" % (instance, idx, growth)) |
302 |
self.ExecOp(op) |
303 |
|
304 |
def ReplaceDisks1D8(self): |
305 |
"""Replace disks on primary and secondary for drbd8.""" |
306 |
for instance in self.instances: |
307 |
for mode in constants.REPLACE_DISK_SEC, constants.REPLACE_DISK_PRI: |
308 |
op = opcodes.OpReplaceDisks(instance_name=instance, |
309 |
mode=mode, |
310 |
disks=[i for i in range(self.disk_count)]) |
311 |
Log("- Replace disks (%s) for instance %s" % (mode, instance)) |
312 |
self.ExecOp(op) |
313 |
|
314 |
def ReplaceDisks2(self): |
315 |
"""Replace secondary node.""" |
316 |
mode = constants.REPLACE_DISK_SEC |
317 |
|
318 |
mytor = izip(islice(cycle(self.nodes), 2, None), |
319 |
self.instances) |
320 |
for tnode, instance in mytor: |
321 |
if self.opts.iallocator: |
322 |
tnode = None |
323 |
op = opcodes.OpReplaceDisks(instance_name=instance, |
324 |
mode=mode, |
325 |
remote_node=tnode, |
326 |
iallocator=self.opts.iallocator, |
327 |
disks=[i for i in range(self.disk_count)]) |
328 |
Log("- Replace secondary (%s) for instance %s" % (mode, instance)) |
329 |
self.ExecOp(op) |
330 |
|
331 |
def Failover(self): |
332 |
"""Failover the instances.""" |
333 |
|
334 |
for instance in self.instances: |
335 |
op = opcodes.OpFailoverInstance(instance_name=instance, |
336 |
ignore_consistency=False) |
337 |
|
338 |
Log("- Failover instance %s" % (instance)) |
339 |
self.ExecOp(op) |
340 |
|
341 |
def ImportExport(self): |
342 |
"""Export the instance, delete it, and import it back. |
343 |
|
344 |
""" |
345 |
|
346 |
mytor = izip(cycle(self.nodes), |
347 |
islice(cycle(self.nodes), 1, None), |
348 |
islice(cycle(self.nodes), 2, None), |
349 |
self.instances) |
350 |
|
351 |
for pnode, snode, enode, instance in mytor: |
352 |
|
353 |
if self.opts.iallocator: |
354 |
pnode = snode = None |
355 |
import_log_msg = ("- Import instance %s from node %s (iallocator: %s)" % |
356 |
(instance, enode, self.opts.iallocator)) |
357 |
elif self.opts.disk_template not in constants.DTS_NET_MIRROR: |
358 |
snode = None |
359 |
import_log_msg = ("- Import instance %s from node %s to node %s" % |
360 |
(instance, enode, pnode)) |
361 |
else: |
362 |
import_log_msg = ("- Import instance %s from node %s to nodes %s/%s" % |
363 |
(instance, enode, pnode, snode)) |
364 |
|
365 |
exp_op = opcodes.OpExportInstance(instance_name=instance, |
366 |
target_node=enode, |
367 |
shutdown=True) |
368 |
rem_op = opcodes.OpRemoveInstance(instance_name=instance, |
369 |
ignore_failures=True) |
370 |
nam_op = opcodes.OpQueryInstances(output_fields=["name"], |
371 |
names=[instance]) |
372 |
full_name = self.ExecOp(nam_op)[0][0] |
373 |
imp_dir = os.path.join(constants.EXPORT_DIR, full_name) |
374 |
imp_op = opcodes.OpCreateInstance(instance_name=instance, |
375 |
disk_size=self.opts.os_size, |
376 |
swap_size=self.opts.swap_size, |
377 |
disk_template=self.opts.disk_template, |
378 |
mode=constants.INSTANCE_IMPORT, |
379 |
src_node=enode, |
380 |
src_path=imp_dir, |
381 |
pnode=pnode, |
382 |
snode=snode, |
383 |
start=True, |
384 |
ip_check=True, |
385 |
wait_for_sync=True, |
386 |
mac="auto", |
387 |
file_storage_dir=None, |
388 |
file_driver=None, |
389 |
iallocator=self.opts.iallocator, |
390 |
beparams=self.bep, |
391 |
hvparams=self.hvp, |
392 |
) |
393 |
|
394 |
erem_op = opcodes.OpRemoveExport(instance_name=instance) |
395 |
|
396 |
Log("- Export instance %s to node %s" % (instance, enode)) |
397 |
self.ExecOp(exp_op) |
398 |
Log("- Remove instance %s" % (instance)) |
399 |
self.ExecOp(rem_op) |
400 |
self.to_rem.remove(instance) |
401 |
Log(import_log_msg) |
402 |
self.ExecOp(imp_op) |
403 |
Log("- Remove export of instance %s" % (instance)) |
404 |
self.ExecOp(erem_op) |
405 |
|
406 |
self.to_rem.append(instance) |
407 |
|
408 |
def StopInstance(self, instance): |
409 |
"""Stop given instance.""" |
410 |
op = opcodes.OpShutdownInstance(instance_name=instance) |
411 |
Log("- Shutdown instance %s" % instance) |
412 |
self.ExecOp(op) |
413 |
|
414 |
def StartInstance(self, instance): |
415 |
"""Start given instance.""" |
416 |
op = opcodes.OpStartupInstance(instance_name=instance, force=False) |
417 |
Log("- Start instance %s" % instance) |
418 |
self.ExecOp(op) |
419 |
|
420 |
def RenameInstance(self, instance, instance_new): |
421 |
"""Rename instance.""" |
422 |
op = opcodes.OpRenameInstance(instance_name=instance, |
423 |
new_name=instance_new) |
424 |
Log("- Rename instance %s to %s" % (instance, instance_new)) |
425 |
self.ExecOp(op) |
426 |
|
427 |
def StopStart(self): |
428 |
"""Stop/start the instances.""" |
429 |
for instance in self.instances: |
430 |
self.StopInstance(instance) |
431 |
self.StartInstance(instance) |
432 |
|
433 |
def Remove(self): |
434 |
"""Remove the instances.""" |
435 |
for instance in self.to_rem: |
436 |
op = opcodes.OpRemoveInstance(instance_name=instance, |
437 |
ignore_failures=True) |
438 |
Log("- Remove instance %s" % instance) |
439 |
self.ExecOp(op) |
440 |
|
441 |
|
442 |
def Rename(self): |
443 |
"""Rename the instances.""" |
444 |
rename = self.opts.rename |
445 |
for instance in self.instances: |
446 |
self.StopInstance(instance) |
447 |
self.RenameInstance(instance, rename) |
448 |
self.StartInstance(rename) |
449 |
self.StopInstance(rename) |
450 |
self.RenameInstance(rename, instance) |
451 |
self.StartInstance(instance) |
452 |
|
453 |
def BurninCluster(self): |
454 |
"""Test a cluster intensively. |
455 |
|
456 |
This will create instances and then start/stop/failover them. |
457 |
It is safe for existing instances but could impact performance. |
458 |
|
459 |
""" |
460 |
|
461 |
opts = self.opts |
462 |
|
463 |
Log("- Testing global parameters") |
464 |
|
465 |
if (len(self.nodes) == 1 and |
466 |
opts.disk_template not in (constants.DT_DISKLESS, constants.DT_PLAIN, |
467 |
constants.DT_FILE)): |
468 |
Log("When one node is available/selected the disk template must" |
469 |
" be 'diskless', 'file' or 'plain'") |
470 |
sys.exit(1) |
471 |
|
472 |
has_err = True |
473 |
try: |
474 |
self.CreateInstances() |
475 |
if opts.do_replace1 and opts.disk_template in constants.DTS_NET_MIRROR: |
476 |
self.ReplaceDisks1D8() |
477 |
if (opts.do_replace2 and len(self.nodes) > 2 and |
478 |
opts.disk_template in constants.DTS_NET_MIRROR) : |
479 |
self.ReplaceDisks2() |
480 |
|
481 |
if opts.disk_template != constants.DT_DISKLESS: |
482 |
self.GrowDisks() |
483 |
|
484 |
if opts.do_failover and opts.disk_template in constants.DTS_NET_MIRROR: |
485 |
self.Failover() |
486 |
|
487 |
if opts.do_importexport: |
488 |
self.ImportExport() |
489 |
|
490 |
if opts.do_startstop: |
491 |
self.StopStart() |
492 |
|
493 |
if opts.rename: |
494 |
self.Rename() |
495 |
|
496 |
has_err = False |
497 |
finally: |
498 |
if has_err: |
499 |
Log("Error detected: opcode buffer follows:\n\n") |
500 |
Log(self.GetFeedbackBuf()) |
501 |
Log("\n\n") |
502 |
self.Remove() |
503 |
|
504 |
return 0 |
505 |
|
506 |
|
507 |
def main(): |
508 |
"""Main function""" |
509 |
|
510 |
burner = Burner() |
511 |
return burner.BurninCluster() |
512 |
|
513 |
|
514 |
if __name__ == "__main__": |
515 |
main() |