-
Notifications
You must be signed in to change notification settings - Fork 292
/
internal.py
726 lines (666 loc) · 24.8 KB
/
internal.py
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
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
"""
Internal tasks are tasks that are started from the teuthology infrastructure.
Note that there is no corresponding task defined for this module. All of
the calls are made from other modules, most notably teuthology/run.py
"""
from cStringIO import StringIO
import contextlib
import logging
import os
import time
import yaml
import subprocess
from teuthology import lockstatus
from teuthology import lock
from teuthology import misc
from teuthology import provision
from teuthology.job_status import get_status, set_status
from teuthology.config import config as teuth_config
from teuthology.parallel import parallel
from teuthology.suite import has_packages_for_distro
from ..orchestra import cluster, remote, run
from .. import report
log = logging.getLogger(__name__)
@contextlib.contextmanager
def base(ctx, config):
"""
Create the test directory that we will be using on the remote system
"""
log.info('Creating test directory...')
testdir = misc.get_testdir(ctx)
run.wait(
ctx.cluster.run(
args=[
'mkdir', '-m0755', '--',
testdir,
],
wait=False,
)
)
try:
yield
finally:
log.info('Tidying up after the test...')
# if this fails, one of the earlier cleanups is flawed; don't
# just cram an rm -rf here
run.wait(
ctx.cluster.run(
args=[
'rmdir',
'--',
testdir,
],
wait=False,
),
)
@contextlib.contextmanager
def lock_machines(ctx, config):
"""
Lock machines. Called when the teuthology run finds and locks
new machines. This is not called if the one has teuthology-locked
machines and placed those keys in the Targets section of a yaml file.
"""
# It's OK for os_type and os_version to be None here. If we're trying
# to lock a bare metal machine, we'll take whatever is available. If
# we want a vps, defaults will be provided by misc.get_distro and
# misc.get_distro_version in provision.create_if_vm
os_type = ctx.config.get("os_type")
os_version = ctx.config.get("os_version")
arch = ctx.config.get('arch')
log.info('Locking machines...')
assert isinstance(config[0], int), 'config[0] must be an integer'
machine_type = config[1]
how_many = config[0]
# We want to make sure there are always this many machines available
to_reserve = 5
# change the status during the locking process
report.try_push_job_info(ctx.config, dict(status='waiting'))
while True:
# get a candidate list of machines
machines = lock.list_locks(machine_type=machine_type, up=True,
locked=False, count=how_many + to_reserve)
if machines is None:
if ctx.block:
log.error('Error listing machines, trying again')
time.sleep(20)
continue
else:
raise RuntimeError('Error listing machines')
# make sure there are machines for non-automated jobs to run
if len(machines) < to_reserve + how_many and ctx.owner.startswith('scheduled'):
if ctx.block:
log.info(
'waiting for more machines to be free (need %s + %s, have %s)...',
to_reserve,
how_many,
len(machines),
)
time.sleep(10)
continue
else:
assert 0, ('not enough machines free; need %s + %s, have %s' %
(to_reserve, how_many, len(machines)))
newly_locked = lock.lock_many(ctx, how_many, machine_type, ctx.owner,
ctx.archive, os_type, os_version, arch)
if not newly_locked and not isinstance(newly_locked, list):
raise RuntimeError('Invalid parameters specified')
if len(newly_locked) == how_many:
vmlist = []
for lmach in newly_locked:
if misc.is_vm(lmach):
vmlist.append(lmach)
if vmlist:
log.info('Waiting for virtual machines to come up')
keys_dict = dict()
loopcount = 0
while len(keys_dict) != len(vmlist):
loopcount += 1
time.sleep(10)
keys_dict = lock.ssh_keyscan(vmlist)
log.info('virtual machine is still unavailable')
if loopcount == 40:
loopcount = 0
log.info('virtual machine(s) still not up, ' +
'recreating unresponsive ones.')
for guest in vmlist:
if guest not in keys_dict.keys():
log.info('recreating: ' + guest)
full_name = misc.canonicalize_hostname(guest)
provision.destroy_if_vm(ctx, full_name)
provision.create_if_vm(ctx, full_name)
if lock.do_update_keys(keys_dict):
log.info("Error in virtual machine keys")
newscandict = {}
for dkey in newly_locked.iterkeys():
stats = lockstatus.get_status(dkey)
newscandict[dkey] = stats['ssh_pub_key']
ctx.config['targets'] = newscandict
else:
ctx.config['targets'] = newly_locked
locked_targets = yaml.safe_dump(
ctx.config['targets'],
default_flow_style=False
).splitlines()
log.info('\n '.join(['Locked targets:', ] + locked_targets))
# successfully locked machines, change status back to running
report.try_push_job_info(ctx.config, dict(status='running'))
break
elif not ctx.block:
assert 0, 'not enough machines are available'
log.warn('Could not lock enough machines, waiting...')
time.sleep(10)
try:
yield
finally:
if ctx.config.get('unlock_on_failure', False) or \
get_status(ctx.summary) == 'pass':
log.info('Unlocking machines...')
for machine in ctx.config['targets'].iterkeys():
lock.unlock_one(ctx, machine, ctx.owner, ctx.archive)
def save_config(ctx, config):
"""
Store the config in a yaml file
"""
log.info('Saving configuration')
if ctx.archive is not None:
with file(os.path.join(ctx.archive, 'config.yaml'), 'w') as f:
yaml.safe_dump(ctx.config, f, default_flow_style=False)
def check_lock(ctx, config):
"""
Check lock status of remote machines.
"""
if not teuth_config.lock_server or ctx.config.get('check-locks') is False:
log.info('Lock checking disabled.')
return
log.info('Checking locks...')
for machine in ctx.config['targets'].iterkeys():
status = lockstatus.get_status(machine)
log.debug('machine status is %s', repr(status))
assert status is not None, \
'could not read lock status for {name}'.format(name=machine)
assert status['up'], 'machine {name} is marked down'.format(name=machine)
assert status['locked'], \
'machine {name} is not locked'.format(name=machine)
assert status['locked_by'] == ctx.owner, \
'machine {name} is locked by {user}, not {owner}'.format(
name=machine,
user=status['locked_by'],
owner=ctx.owner,
)
def check_packages(ctx, config):
"""
Checks gitbuilder to determine if there are missing packages for this job.
If there are missing packages, fail the job.
"""
log.info("Checking packages...")
os_type = ctx.config.get("os_type", None)
sha1 = ctx.config.get("sha1", None)
# We can only do this check if there are a defined sha1 and os_type
# in the job config.
if os_type and sha1:
log.info(
"Checking packages for os_type '{os}' and ceph hash '{ver}'".format(
os=os_type,
ver=sha1,
)
)
if not has_packages_for_distro(sha1, os_type):
msg = "Packages for os_type '{os}' and ceph hash '{ver}' not found"
msg = msg.format(
os=os_type,
ver=sha1,
)
log.error(msg)
# set the failure message and update paddles with the status
ctx.summary["failure_reason"] = msg
set_status(ctx.summary, "dead")
report.try_push_job_info(ctx.config, dict(status='dead'))
raise RuntimeError(msg)
else:
log.info(
"Checking packages skipped, missing os_type '{os}' or ceph hash '{ver}'".format(
os=os_type,
ver=sha1,
)
)
@contextlib.contextmanager
def timer(ctx, config):
"""
Start the timer used by teuthology
"""
log.info('Starting timer...')
start = time.time()
try:
yield
finally:
duration = time.time() - start
log.info('Duration was %f seconds', duration)
ctx.summary['duration'] = duration
def connect(ctx, config):
"""
Open a connection to a remote host.
"""
log.info('Opening connections...')
remotes = []
machs = []
for name in ctx.config['targets'].iterkeys():
machs.append(name)
for t, key in ctx.config['targets'].iteritems():
t = misc.canonicalize_hostname(t)
log.debug('connecting to %s', t)
try:
if ctx.config['sshkeys'] == 'ignore':
key = None
except (AttributeError, KeyError):
pass
remotes.append(
remote.Remote(name=t, host_key=key, keep_alive=True, console=None))
ctx.cluster = cluster.Cluster()
if 'roles' in ctx.config:
for rem, roles in zip(remotes, ctx.config['roles']):
assert all(isinstance(role, str) for role in roles), \
"Roles in config must be strings: %r" % roles
ctx.cluster.add(rem, roles)
log.info('roles: %s - %s' % (rem, roles))
else:
for rem in remotes:
ctx.cluster.add(rem, rem.name)
def push_inventory(ctx, config):
if not teuth_config.lock_server:
return
def push():
for rem in ctx.cluster.remotes.keys():
info = rem.inventory_info
lock.update_inventory(info)
try:
push()
except Exception:
log.exception("Error pushing inventory")
def serialize_remote_roles(ctx, config):
"""
Provides an explicit mapping for which remotes have been assigned what roles
So that other software can be loosely coupled to teuthology
"""
if ctx.archive is not None:
with file(os.path.join(ctx.archive, 'info.yaml'), 'r+') as info_file:
info_yaml = yaml.safe_load(info_file)
info_file.seek(0)
info_yaml['cluster'] = dict([(rem.name, {'roles': roles}) for rem, roles in ctx.cluster.remotes.iteritems()])
yaml.safe_dump(info_yaml, info_file, default_flow_style=False)
def check_ceph_data(ctx, config):
"""
Check for old /var/lib/ceph directories and detect staleness.
"""
log.info('Checking for old /var/lib/ceph...')
processes = ctx.cluster.run(
args=[
'test', '!', '-e', '/var/lib/ceph',
],
wait=False,
)
failed = False
for proc in processes:
try:
proc.wait()
except run.CommandFailedError:
log.error('Host %s has stale /var/lib/ceph, check lock and nuke/cleanup.', proc.remote.shortname)
failed = True
if failed:
raise RuntimeError('Stale /var/lib/ceph detected, aborting.')
def check_conflict(ctx, config):
"""
Note directory use conflicts and stale directories.
"""
log.info('Checking for old test directory...')
testdir = misc.get_testdir(ctx)
processes = ctx.cluster.run(
args=[
'test', '!', '-e', testdir,
],
wait=False,
)
failed = False
for proc in processes:
try:
proc.wait()
except run.CommandFailedError:
log.error('Host %s has stale test directory %s, check lock and cleanup.', proc.remote.shortname, testdir)
failed = True
if failed:
raise RuntimeError('Stale jobs detected, aborting.')
def fetch_binaries_for_coredumps(path, remote):
"""
Pul ELFs (debug and stripped) for each coredump found
"""
# Check for Coredumps:
coredump_path = os.path.join(path, 'coredump')
if os.path.isdir(coredump_path):
log.info('Transferring binaries for coredumps...')
for dump in os.listdir(coredump_path):
# Pull program from core file
dump_path = os.path.join(coredump_path, dump)
dump_info = subprocess.Popen(['file', dump_path],
stdout=subprocess.PIPE)
dump_out = dump_info.communicate()
# Parse file output to get program, Example output:
# 1422917770.7450.core: ELF 64-bit LSB core file x86-64, version 1 (SYSV), SVR4-style, \
# from 'radosgw --rgw-socket-path /home/ubuntu/cephtest/apache/tmp.client.0/fastcgi_soc'
dump_program = dump_out.split("from '")[1].split(' ')[0]
# Find path on remote server:
r = remote.run(args=['which', dump_program], stdout=StringIO())
remote_path = r.stdout.getvalue()
# Pull remote program into coredump folder:
remote._sftp_get_file(remote_path, os.path.join(coredump_path,
dump_program))
# Pull Debug symbols:
debug_path = os.path.join('/usr/lib/debug', remote_path)
# RPM distro's append their non-stripped ELF's with .debug
# When deb based distro's do not.
if remote.system_type == 'rpm':
debug_path = '{debug_path}.debug'.format(debug_path=debug_path)
remote.get_file(debug_path, coredump_path)
@contextlib.contextmanager
def archive(ctx, config):
"""
Handle the creation and deletion of the archive directory.
"""
log.info('Creating archive directory...')
archive_dir = misc.get_archive_dir(ctx)
run.wait(
ctx.cluster.run(
args=[
'install', '-d', '-m0755', '--', archive_dir,
],
wait=False,
)
)
try:
yield
except Exception:
# we need to know this below
set_status(ctx.summary, 'fail')
raise
finally:
passed = get_status(ctx.summary) == 'pass'
if ctx.archive is not None and \
not (ctx.config.get('archive-on-error') and passed):
log.info('Transferring archived files...')
logdir = os.path.join(ctx.archive, 'remote')
if (not os.path.exists(logdir)):
os.mkdir(logdir)
for rem in ctx.cluster.remotes.iterkeys():
path = os.path.join(logdir, rem.shortname)
misc.pull_directory(rem, archive_dir, path)
# Check for coredumps and pull binaries
fetch_binaries_for_coredumps(path, rem)
log.info('Removing archive directory...')
run.wait(
ctx.cluster.run(
args=[
'rm',
'-rf',
'--',
archive_dir,
],
wait=False,
),
)
@contextlib.contextmanager
def sudo(ctx, config):
"""
Enable use of sudo
"""
log.info('Configuring sudo...')
sudoers_file = '/etc/sudoers'
backup_ext = '.orig.teuthology'
tty_expr = r's/^\([^#]*\) \(requiretty\)/\1 !\2/g'
pw_expr = r's/^\([^#]*\) !\(visiblepw\)/\1 \2/g'
run.wait(
ctx.cluster.run(
args="sudo sed -i{ext} -e '{tty}' -e '{pw}' {path}".format(
ext=backup_ext, tty=tty_expr, pw=pw_expr,
path=sudoers_file
),
wait=False,
)
)
try:
yield
finally:
log.info('Restoring {0}...'.format(sudoers_file))
ctx.cluster.run(
args="sudo mv -f {path}{ext} {path}".format(
path=sudoers_file, ext=backup_ext
)
)
@contextlib.contextmanager
def coredump(ctx, config):
"""
Stash a coredump of this system if an error occurs.
"""
log.info('Enabling coredump saving...')
archive_dir = misc.get_archive_dir(ctx)
run.wait(
ctx.cluster.run(
args=[
'install', '-d', '-m0755', '--',
'{adir}/coredump'.format(adir=archive_dir),
run.Raw('&&'),
'sudo', 'sysctl', '-w', 'kernel.core_pattern={adir}/coredump/%t.%p.core'.format(adir=archive_dir),
],
wait=False,
)
)
try:
yield
finally:
run.wait(
ctx.cluster.run(
args=[
'sudo', 'sysctl', '-w', 'kernel.core_pattern=core',
run.Raw('&&'),
# don't litter the archive dir if there were no cores dumped
'rmdir',
'--ignore-fail-on-non-empty',
'--',
'{adir}/coredump'.format(adir=archive_dir),
],
wait=False,
)
)
# set status = 'fail' if the dir is still there = coredumps were
# seen
for rem in ctx.cluster.remotes.iterkeys():
r = rem.run(
args=[
'if', 'test', '!', '-e', '{adir}/coredump'.format(adir=archive_dir), run.Raw(';'), 'then',
'echo', 'OK', run.Raw(';'),
'fi',
],
stdout=StringIO(),
)
if r.stdout.getvalue() != 'OK\n':
log.warning('Found coredumps on %s, flagging run as failed', rem)
set_status(ctx.summary, 'fail')
if 'failure_reason' not in ctx.summary:
ctx.summary['failure_reason'] = \
'Found coredumps on {rem}'.format(rem=rem)
@contextlib.contextmanager
def syslog(ctx, config):
"""
start syslog / stop syslog on exit.
"""
if ctx.archive is None:
# disable this whole feature if we're not going to archive the data anyway
yield
return
log.info('Starting syslog monitoring...')
archive_dir = misc.get_archive_dir(ctx)
run.wait(
ctx.cluster.run(
args=[
'mkdir', '-m0755', '--',
'{adir}/syslog'.format(adir=archive_dir),
],
wait=False,
)
)
CONF = '/etc/rsyslog.d/80-cephtest.conf'
conf_fp = StringIO('''
kern.* -{adir}/syslog/kern.log;RSYSLOG_FileFormat
*.*;kern.none -{adir}/syslog/misc.log;RSYSLOG_FileFormat
'''.format(adir=archive_dir))
try:
for rem in ctx.cluster.remotes.iterkeys():
misc.sudo_write_file(
remote=rem,
path=CONF,
data=conf_fp,
)
conf_fp.seek(0)
run.wait(
ctx.cluster.run(
args=[
'sudo',
'service',
# a mere reload (SIGHUP) doesn't seem to make
# rsyslog open the files
'rsyslog',
'restart',
],
wait=False,
),
)
yield
finally:
log.info('Shutting down syslog monitoring...')
run.wait(
ctx.cluster.run(
args=[
'sudo',
'rm',
'-f',
'--',
CONF,
run.Raw('&&'),
'sudo',
'service',
'rsyslog',
'restart',
],
wait=False,
),
)
# race condition: nothing actually says rsyslog had time to
# flush the file fully. oh well.
log.info('Checking logs for errors...')
for rem in ctx.cluster.remotes.iterkeys():
log.debug('Checking %s', rem.name)
r = rem.run(
args=[
'egrep', '--binary-files=text',
'\\bBUG\\b|\\bINFO\\b|\\bDEADLOCK\\b',
run.Raw('{adir}/syslog/*.log'.format(adir=archive_dir)),
run.Raw('|'),
'grep', '-v', 'task .* blocked for more than .* seconds',
run.Raw('|'),
'grep', '-v', 'lockdep is turned off',
run.Raw('|'),
'grep', '-v', 'trying to register non-static key',
run.Raw('|'),
'grep', '-v', 'DEBUG: fsize', # xfs_fsr
run.Raw('|'),
'grep', '-v', 'CRON', # ignore cron noise
run.Raw('|'),
'grep', '-v', 'BUG: bad unlock balance detected', # #6097
run.Raw('|'),
'grep', '-v', 'inconsistent lock state', # FIXME see #2523
run.Raw('|'),
'grep', '-v', '*** DEADLOCK ***', # part of lockdep output
run.Raw('|'),
'grep', '-v', 'INFO: possible irq lock inversion dependency detected', # FIXME see #2590 and #147
run.Raw('|'),
'grep', '-v', 'INFO: NMI handler (perf_event_nmi_handler) took too long to run',
run.Raw('|'),
'grep', '-v', 'INFO: recovery required on readonly',
run.Raw('|'),
'head', '-n', '1',
],
stdout=StringIO(),
)
stdout = r.stdout.getvalue()
if stdout != '':
log.error('Error in syslog on %s: %s', rem.name, stdout)
set_status(ctx.summary, 'fail')
if 'failure_reason' not in ctx.summary:
ctx.summary['failure_reason'] = \
"'{error}' in syslog".format(error=stdout)
log.info('Compressing syslogs...')
run.wait(
ctx.cluster.run(
args=[
'find',
'{adir}/syslog'.format(adir=archive_dir),
'-name',
'*.log',
'-print0',
run.Raw('|'),
'sudo',
'xargs',
'-0',
'--no-run-if-empty',
'--',
'gzip',
'--',
],
wait=False,
),
)
def vm_setup(ctx, config):
"""
Look for virtual machines and handle their initialization
"""
all_tasks = [x.keys()[0] for x in ctx.config['tasks']]
need_chef = False
if 'chef' in all_tasks or 'kernel' in all_tasks:
need_chef = True
with parallel() as p:
editinfo = os.path.join(os.path.dirname(__file__),'edit_sudoers.sh')
for rem in ctx.cluster.remotes.iterkeys():
mname = rem.shortname
if misc.is_vm(mname):
r = rem.run(args=['test', '-e', '/ceph-qa-ready',],
stdout=StringIO(),
check_status=False,)
if r.returncode != 0:
p1 = subprocess.Popen(['cat', editinfo], stdout=subprocess.PIPE)
p2 = subprocess.Popen(
[
'ssh',
'-o', 'StrictHostKeyChecking=no',
'-t', '-t',
str(rem),
'sudo',
'sh'
],
stdin=p1.stdout, stdout=subprocess.PIPE
)
_, err = p2.communicate()
if err:
log.info("Edit of /etc/sudoers failed: %s", err)
if need_chef:
p.spawn(_download_and_run_chef, rem)
def _download_and_run_chef(remote_):
"""
Run ceph_qa_chef.
"""
log.info('Running ceph_qa_chef on %s', remote_)
remote_.run(
args=[
'wget', '-q', '-O-',
'http://ceph.com/git/?p=ceph-qa-chef.git;a=blob_plain;f=solo/solo-from-scratch;hb=HEAD',
run.Raw('|'),
'sh',
],
label="run chef solo-from-scratch"
)