root / lib / backend.py @ b1206984
History | View | Annotate | Download (59 kB)
1 |
#
|
---|---|
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 |
"""Functions used by the node daemon"""
|
23 |
|
24 |
|
25 |
import os |
26 |
import os.path |
27 |
import shutil |
28 |
import time |
29 |
import stat |
30 |
import errno |
31 |
import re |
32 |
import subprocess |
33 |
import random |
34 |
import logging |
35 |
import tempfile |
36 |
|
37 |
from ganeti import errors |
38 |
from ganeti import utils |
39 |
from ganeti import ssh |
40 |
from ganeti import hypervisor |
41 |
from ganeti import constants |
42 |
from ganeti import bdev |
43 |
from ganeti import objects |
44 |
from ganeti import ssconf |
45 |
|
46 |
|
47 |
def _GetConfig(): |
48 |
return ssconf.SimpleConfigReader()
|
49 |
|
50 |
|
51 |
def _GetSshRunner(cluster_name): |
52 |
return ssh.SshRunner(cluster_name)
|
53 |
|
54 |
|
55 |
def _CleanDirectory(path, exclude=[]): |
56 |
"""Removes all regular files in a directory.
|
57 |
|
58 |
@param exclude: List of files to be excluded.
|
59 |
@type exclude: list
|
60 |
|
61 |
"""
|
62 |
if not os.path.isdir(path): |
63 |
return
|
64 |
|
65 |
# Normalize excluded paths
|
66 |
exclude = [os.path.normpath(i) for i in exclude] |
67 |
|
68 |
for rel_name in utils.ListVisibleFiles(path): |
69 |
full_name = os.path.normpath(os.path.join(path, rel_name)) |
70 |
if full_name in exclude: |
71 |
continue
|
72 |
if os.path.isfile(full_name) and not os.path.islink(full_name): |
73 |
utils.RemoveFile(full_name) |
74 |
|
75 |
|
76 |
def JobQueuePurge(): |
77 |
"""Removes job queue files and archived jobs
|
78 |
|
79 |
"""
|
80 |
_CleanDirectory(constants.QUEUE_DIR, exclude=[constants.JOB_QUEUE_LOCK_FILE]) |
81 |
_CleanDirectory(constants.JOB_QUEUE_ARCHIVE_DIR) |
82 |
|
83 |
|
84 |
def GetMasterInfo(): |
85 |
"""Returns master information.
|
86 |
|
87 |
This is an utility function to compute master information, either
|
88 |
for consumption here or from the node daemon.
|
89 |
|
90 |
@rtype: tuple
|
91 |
@return: (master_netdev, master_ip, master_name)
|
92 |
|
93 |
"""
|
94 |
try:
|
95 |
cfg = _GetConfig() |
96 |
master_netdev = cfg.GetMasterNetdev() |
97 |
master_ip = cfg.GetMasterIP() |
98 |
master_node = cfg.GetMasterNode() |
99 |
except errors.ConfigurationError, err:
|
100 |
logging.exception("Cluster configuration incomplete")
|
101 |
return (None, None) |
102 |
return (master_netdev, master_ip, master_node)
|
103 |
|
104 |
|
105 |
def StartMaster(start_daemons): |
106 |
"""Activate local node as master node.
|
107 |
|
108 |
The function will always try activate the IP address of the master
|
109 |
(if someone else has it, then it won't). Then, if the start_daemons
|
110 |
parameter is True, it will also start the master daemons
|
111 |
(ganet-masterd and ganeti-rapi).
|
112 |
|
113 |
"""
|
114 |
ok = True
|
115 |
master_netdev, master_ip, _ = GetMasterInfo() |
116 |
if not master_netdev: |
117 |
return False |
118 |
|
119 |
if utils.TcpPing(master_ip, constants.DEFAULT_NODED_PORT):
|
120 |
if utils.OwnIpAddress(master_ip):
|
121 |
# we already have the ip:
|
122 |
logging.debug("Already started")
|
123 |
else:
|
124 |
logging.error("Someone else has the master ip, not activating")
|
125 |
ok = False
|
126 |
else:
|
127 |
result = utils.RunCmd(["ip", "address", "add", "%s/32" % master_ip, |
128 |
"dev", master_netdev, "label", |
129 |
"%s:0" % master_netdev])
|
130 |
if result.failed:
|
131 |
logging.error("Can't activate master IP: %s", result.output)
|
132 |
ok = False
|
133 |
|
134 |
result = utils.RunCmd(["arping", "-q", "-U", "-c 3", "-I", master_netdev, |
135 |
"-s", master_ip, master_ip])
|
136 |
# we'll ignore the exit code of arping
|
137 |
|
138 |
# and now start the master and rapi daemons
|
139 |
if start_daemons:
|
140 |
for daemon in 'ganeti-masterd', 'ganeti-rapi': |
141 |
result = utils.RunCmd([daemon]) |
142 |
if result.failed:
|
143 |
logging.error("Can't start daemon %s: %s", daemon, result.output)
|
144 |
ok = False
|
145 |
return ok
|
146 |
|
147 |
|
148 |
def StopMaster(stop_daemons): |
149 |
"""Deactivate this node as master.
|
150 |
|
151 |
The function will always try to deactivate the IP address of the
|
152 |
master. Then, if the stop_daemons parameter is True, it will also
|
153 |
stop the master daemons (ganet-masterd and ganeti-rapi).
|
154 |
|
155 |
"""
|
156 |
master_netdev, master_ip, _ = GetMasterInfo() |
157 |
if not master_netdev: |
158 |
return False |
159 |
|
160 |
result = utils.RunCmd(["ip", "address", "del", "%s/32" % master_ip, |
161 |
"dev", master_netdev])
|
162 |
if result.failed:
|
163 |
logging.error("Can't remove the master IP, error: %s", result.output)
|
164 |
# but otherwise ignore the failure
|
165 |
|
166 |
if stop_daemons:
|
167 |
# stop/kill the rapi and the master daemon
|
168 |
for daemon in constants.RAPI_PID, constants.MASTERD_PID: |
169 |
utils.KillProcess(utils.ReadPidFile(utils.DaemonPidFileName(daemon))) |
170 |
|
171 |
return True |
172 |
|
173 |
|
174 |
def AddNode(dsa, dsapub, rsa, rsapub, sshkey, sshpub): |
175 |
"""Joins this node to the cluster.
|
176 |
|
177 |
This does the following:
|
178 |
- updates the hostkeys of the machine (rsa and dsa)
|
179 |
- adds the ssh private key to the user
|
180 |
- adds the ssh public key to the users' authorized_keys file
|
181 |
|
182 |
"""
|
183 |
sshd_keys = [(constants.SSH_HOST_RSA_PRIV, rsa, 0600),
|
184 |
(constants.SSH_HOST_RSA_PUB, rsapub, 0644),
|
185 |
(constants.SSH_HOST_DSA_PRIV, dsa, 0600),
|
186 |
(constants.SSH_HOST_DSA_PUB, dsapub, 0644)]
|
187 |
for name, content, mode in sshd_keys: |
188 |
utils.WriteFile(name, data=content, mode=mode) |
189 |
|
190 |
try:
|
191 |
priv_key, pub_key, auth_keys = ssh.GetUserFiles(constants.GANETI_RUNAS, |
192 |
mkdir=True)
|
193 |
except errors.OpExecError, err:
|
194 |
logging.exception("Error while processing user ssh files")
|
195 |
return False |
196 |
|
197 |
for name, content in [(priv_key, sshkey), (pub_key, sshpub)]: |
198 |
utils.WriteFile(name, data=content, mode=0600)
|
199 |
|
200 |
utils.AddAuthorizedKey(auth_keys, sshpub) |
201 |
|
202 |
utils.RunCmd([constants.SSH_INITD_SCRIPT, "restart"])
|
203 |
|
204 |
return True |
205 |
|
206 |
|
207 |
def LeaveCluster(): |
208 |
"""Cleans up the current node and prepares it to be removed from the cluster.
|
209 |
|
210 |
"""
|
211 |
_CleanDirectory(constants.DATA_DIR) |
212 |
JobQueuePurge() |
213 |
|
214 |
try:
|
215 |
priv_key, pub_key, auth_keys = ssh.GetUserFiles(constants.GANETI_RUNAS) |
216 |
except errors.OpExecError:
|
217 |
logging.exception("Error while processing ssh files")
|
218 |
return
|
219 |
|
220 |
f = open(pub_key, 'r') |
221 |
try:
|
222 |
utils.RemoveAuthorizedKey(auth_keys, f.read(8192))
|
223 |
finally:
|
224 |
f.close() |
225 |
|
226 |
utils.RemoveFile(priv_key) |
227 |
utils.RemoveFile(pub_key) |
228 |
|
229 |
# Return a reassuring string to the caller, and quit
|
230 |
raise errors.QuitGanetiException(False, 'Shutdown scheduled') |
231 |
|
232 |
|
233 |
def GetNodeInfo(vgname, hypervisor_type): |
234 |
"""Gives back a hash with different informations about the node.
|
235 |
|
236 |
@type vgname: C{string}
|
237 |
@param vgname: the name of the volume group to ask for disk space information
|
238 |
@type hypervisor_type: C{str}
|
239 |
@param hypervisor_type: the name of the hypervisor to ask for
|
240 |
memory information
|
241 |
@rtype: C{dict}
|
242 |
@return: dictionary with the following keys:
|
243 |
- vg_size is the size of the configured volume group in MiB
|
244 |
- vg_free is the free size of the volume group in MiB
|
245 |
- memory_dom0 is the memory allocated for domain0 in MiB
|
246 |
- memory_free is the currently available (free) ram in MiB
|
247 |
- memory_total is the total number of ram in MiB
|
248 |
|
249 |
"""
|
250 |
outputarray = {} |
251 |
vginfo = _GetVGInfo(vgname) |
252 |
outputarray['vg_size'] = vginfo['vg_size'] |
253 |
outputarray['vg_free'] = vginfo['vg_free'] |
254 |
|
255 |
hyper = hypervisor.GetHypervisor(hypervisor_type) |
256 |
hyp_info = hyper.GetNodeInfo() |
257 |
if hyp_info is not None: |
258 |
outputarray.update(hyp_info) |
259 |
|
260 |
f = open("/proc/sys/kernel/random/boot_id", 'r') |
261 |
try:
|
262 |
outputarray["bootid"] = f.read(128).rstrip("\n") |
263 |
finally:
|
264 |
f.close() |
265 |
|
266 |
return outputarray
|
267 |
|
268 |
|
269 |
def VerifyNode(what, cluster_name): |
270 |
"""Verify the status of the local node.
|
271 |
|
272 |
Based on the input L{what} parameter, various checks are done on the
|
273 |
local node.
|
274 |
|
275 |
If the I{filelist} key is present, this list of
|
276 |
files is checksummed and the file/checksum pairs are returned.
|
277 |
|
278 |
If the I{nodelist} key is present, we check that we have
|
279 |
connectivity via ssh with the target nodes (and check the hostname
|
280 |
report).
|
281 |
|
282 |
If the I{node-net-test} key is present, we check that we have
|
283 |
connectivity to the given nodes via both primary IP and, if
|
284 |
applicable, secondary IPs.
|
285 |
|
286 |
@type what: C{dict}
|
287 |
@param what: a dictionary of things to check:
|
288 |
- filelist: list of files for which to compute checksums
|
289 |
- nodelist: list of nodes we should check ssh communication with
|
290 |
- node-net-test: list of nodes we should check node daemon port
|
291 |
connectivity with
|
292 |
- hypervisor: list with hypervisors to run the verify for
|
293 |
|
294 |
"""
|
295 |
result = {} |
296 |
|
297 |
if 'hypervisor' in what: |
298 |
result['hypervisor'] = my_dict = {}
|
299 |
for hv_name in what['hypervisor']: |
300 |
my_dict[hv_name] = hypervisor.GetHypervisor(hv_name).Verify() |
301 |
|
302 |
if 'filelist' in what: |
303 |
result['filelist'] = utils.FingerprintFiles(what['filelist']) |
304 |
|
305 |
if 'nodelist' in what: |
306 |
result['nodelist'] = {}
|
307 |
random.shuffle(what['nodelist'])
|
308 |
for node in what['nodelist']: |
309 |
success, message = _GetSshRunner(cluster_name).VerifyNodeHostname(node) |
310 |
if not success: |
311 |
result['nodelist'][node] = message
|
312 |
if 'node-net-test' in what: |
313 |
result['node-net-test'] = {}
|
314 |
my_name = utils.HostInfo().name |
315 |
my_pip = my_sip = None
|
316 |
for name, pip, sip in what['node-net-test']: |
317 |
if name == my_name:
|
318 |
my_pip = pip |
319 |
my_sip = sip |
320 |
break
|
321 |
if not my_pip: |
322 |
result['node-net-test'][my_name] = ("Can't find my own" |
323 |
" primary/secondary IP"
|
324 |
" in the node list")
|
325 |
else:
|
326 |
port = utils.GetNodeDaemonPort() |
327 |
for name, pip, sip in what['node-net-test']: |
328 |
fail = [] |
329 |
if not utils.TcpPing(pip, port, source=my_pip): |
330 |
fail.append("primary")
|
331 |
if sip != pip:
|
332 |
if not utils.TcpPing(sip, port, source=my_sip): |
333 |
fail.append("secondary")
|
334 |
if fail:
|
335 |
result['node-net-test'][name] = ("failure using the %s" |
336 |
" interface(s)" %
|
337 |
" and ".join(fail))
|
338 |
|
339 |
return result
|
340 |
|
341 |
|
342 |
def GetVolumeList(vg_name): |
343 |
"""Compute list of logical volumes and their size.
|
344 |
|
345 |
Returns:
|
346 |
dictionary of all partions (key) with their size (in MiB), inactive
|
347 |
and online status:
|
348 |
{'test1': ('20.06', True, True)}
|
349 |
|
350 |
"""
|
351 |
lvs = {} |
352 |
sep = '|'
|
353 |
result = utils.RunCmd(["lvs", "--noheadings", "--units=m", "--nosuffix", |
354 |
"--separator=%s" % sep,
|
355 |
"-olv_name,lv_size,lv_attr", vg_name])
|
356 |
if result.failed:
|
357 |
logging.error("Failed to list logical volumes, lvs output: %s",
|
358 |
result.output) |
359 |
return result.output
|
360 |
|
361 |
valid_line_re = re.compile("^ *([^|]+)\|([0-9.]+)\|([^|]{6})\|?$")
|
362 |
for line in result.stdout.splitlines(): |
363 |
line = line.strip() |
364 |
match = valid_line_re.match(line) |
365 |
if not match: |
366 |
logging.error("Invalid line returned from lvs output: '%s'", line)
|
367 |
continue
|
368 |
name, size, attr = match.groups() |
369 |
inactive = attr[4] == '-' |
370 |
online = attr[5] == 'o' |
371 |
lvs[name] = (size, inactive, online) |
372 |
|
373 |
return lvs
|
374 |
|
375 |
|
376 |
def ListVolumeGroups(): |
377 |
"""List the volume groups and their size.
|
378 |
|
379 |
Returns:
|
380 |
Dictionary with keys volume name and values the size of the volume
|
381 |
|
382 |
"""
|
383 |
return utils.ListVolumeGroups()
|
384 |
|
385 |
|
386 |
def NodeVolumes(): |
387 |
"""List all volumes on this node.
|
388 |
|
389 |
"""
|
390 |
result = utils.RunCmd(["lvs", "--noheadings", "--units=m", "--nosuffix", |
391 |
"--separator=|",
|
392 |
"--options=lv_name,lv_size,devices,vg_name"])
|
393 |
if result.failed:
|
394 |
logging.error("Failed to list logical volumes, lvs output: %s",
|
395 |
result.output) |
396 |
return {}
|
397 |
|
398 |
def parse_dev(dev): |
399 |
if '(' in dev: |
400 |
return dev.split('(')[0] |
401 |
else:
|
402 |
return dev
|
403 |
|
404 |
def map_line(line): |
405 |
return {
|
406 |
'name': line[0].strip(), |
407 |
'size': line[1].strip(), |
408 |
'dev': parse_dev(line[2].strip()), |
409 |
'vg': line[3].strip(), |
410 |
} |
411 |
|
412 |
return [map_line(line.split('|')) for line in result.stdout.splitlines() |
413 |
if line.count('|') >= 3] |
414 |
|
415 |
|
416 |
def BridgesExist(bridges_list): |
417 |
"""Check if a list of bridges exist on the current node.
|
418 |
|
419 |
@rtype: boolean
|
420 |
@return: C{True} if all of them exist, C{False} otherwise
|
421 |
|
422 |
"""
|
423 |
for bridge in bridges_list: |
424 |
if not utils.BridgeExists(bridge): |
425 |
return False |
426 |
|
427 |
return True |
428 |
|
429 |
|
430 |
def GetInstanceList(hypervisor_list): |
431 |
"""Provides a list of instances.
|
432 |
|
433 |
@type hypervisor_list: list
|
434 |
@param hypervisor_list: the list of hypervisors to query information
|
435 |
|
436 |
@rtype: list
|
437 |
@return: a list of all running instances on the current node
|
438 |
- instance1.example.com
|
439 |
- instance2.example.com
|
440 |
|
441 |
"""
|
442 |
results = [] |
443 |
for hname in hypervisor_list: |
444 |
try:
|
445 |
names = hypervisor.GetHypervisor(hname).ListInstances() |
446 |
results.extend(names) |
447 |
except errors.HypervisorError, err:
|
448 |
logging.exception("Error enumerating instances for hypevisor %s", hname)
|
449 |
# FIXME: should we somehow not propagate this to the master?
|
450 |
raise
|
451 |
|
452 |
return results
|
453 |
|
454 |
|
455 |
def GetInstanceInfo(instance, hname): |
456 |
"""Gives back the informations about an instance as a dictionary.
|
457 |
|
458 |
@type instance: string
|
459 |
@param instance: the instance name
|
460 |
@type hname: string
|
461 |
@param hname: the hypervisor type of the instance
|
462 |
|
463 |
@rtype: dict
|
464 |
@return: dictionary with the following keys:
|
465 |
- memory: memory size of instance (int)
|
466 |
- state: xen state of instance (string)
|
467 |
- time: cpu time of instance (float)
|
468 |
|
469 |
"""
|
470 |
output = {} |
471 |
|
472 |
iinfo = hypervisor.GetHypervisor(hname).GetInstanceInfo(instance) |
473 |
if iinfo is not None: |
474 |
output['memory'] = iinfo[2] |
475 |
output['state'] = iinfo[4] |
476 |
output['time'] = iinfo[5] |
477 |
|
478 |
return output
|
479 |
|
480 |
|
481 |
def GetAllInstancesInfo(hypervisor_list): |
482 |
"""Gather data about all instances.
|
483 |
|
484 |
This is the equivalent of `GetInstanceInfo()`, except that it
|
485 |
computes data for all instances at once, thus being faster if one
|
486 |
needs data about more than one instance.
|
487 |
|
488 |
@type hypervisor_list: list
|
489 |
@param hypervisor_list: list of hypervisors to query for instance data
|
490 |
|
491 |
@rtype: dict of dicts
|
492 |
@return: dictionary of instance: data, with data having the following keys:
|
493 |
- memory: memory size of instance (int)
|
494 |
- state: xen state of instance (string)
|
495 |
- time: cpu time of instance (float)
|
496 |
- vcpuus: the number of vcpus
|
497 |
|
498 |
"""
|
499 |
output = {} |
500 |
|
501 |
for hname in hypervisor_list: |
502 |
iinfo = hypervisor.GetHypervisor(hname).GetAllInstancesInfo() |
503 |
if iinfo:
|
504 |
for name, inst_id, memory, vcpus, state, times in iinfo: |
505 |
value = { |
506 |
'memory': memory,
|
507 |
'vcpus': vcpus,
|
508 |
'state': state,
|
509 |
'time': times,
|
510 |
} |
511 |
if name in output and output[name] != value: |
512 |
raise errors.HypervisorError("Instance %s running duplicate" |
513 |
" with different parameters" % name)
|
514 |
output[name] = value |
515 |
|
516 |
return output
|
517 |
|
518 |
|
519 |
def AddOSToInstance(instance): |
520 |
"""Add an OS to an instance.
|
521 |
|
522 |
@type instance: L{objects.Instance}
|
523 |
@param instance: Instance whose OS is to be installed
|
524 |
|
525 |
"""
|
526 |
inst_os = OSFromDisk(instance.os) |
527 |
|
528 |
create_script = inst_os.create_script |
529 |
create_env = OSEnvironment(instance) |
530 |
|
531 |
logfile = "%s/add-%s-%s-%d.log" % (constants.LOG_OS_DIR, instance.os,
|
532 |
instance.name, int(time.time()))
|
533 |
if not os.path.exists(constants.LOG_OS_DIR): |
534 |
os.mkdir(constants.LOG_OS_DIR, 0750)
|
535 |
|
536 |
command = utils.BuildShellCmd("cd %s && %s &>%s",
|
537 |
inst_os.path, create_script, logfile) |
538 |
|
539 |
result = utils.RunCmd(command, env=create_env) |
540 |
if result.failed:
|
541 |
logging.error("os create command '%s' returned error: %s, logfile: %s,"
|
542 |
" output: %s", command, result.fail_reason, logfile,
|
543 |
result.output) |
544 |
return False |
545 |
|
546 |
return True |
547 |
|
548 |
|
549 |
def RunRenameInstance(instance, old_name): |
550 |
"""Run the OS rename script for an instance.
|
551 |
|
552 |
@type instance: L{objects.Instance}
|
553 |
@param instance: Instance whose OS is to be installed
|
554 |
@type old_name: string
|
555 |
@param old_name: previous instance name
|
556 |
|
557 |
"""
|
558 |
inst_os = OSFromDisk(instance.os) |
559 |
|
560 |
script = inst_os.rename_script |
561 |
rename_env = OSEnvironment(instance) |
562 |
rename_env['OLD_INSTANCE_NAME'] = old_name
|
563 |
|
564 |
logfile = "%s/rename-%s-%s-%s-%d.log" % (constants.LOG_OS_DIR, instance.os,
|
565 |
old_name, |
566 |
instance.name, int(time.time()))
|
567 |
if not os.path.exists(constants.LOG_OS_DIR): |
568 |
os.mkdir(constants.LOG_OS_DIR, 0750)
|
569 |
|
570 |
command = utils.BuildShellCmd("cd %s && %s &>%s",
|
571 |
inst_os.path, script, logfile) |
572 |
|
573 |
result = utils.RunCmd(command, env=rename_env) |
574 |
|
575 |
if result.failed:
|
576 |
logging.error("os create command '%s' returned error: %s output: %s",
|
577 |
command, result.fail_reason, result.output) |
578 |
return False |
579 |
|
580 |
return True |
581 |
|
582 |
|
583 |
def _GetVGInfo(vg_name): |
584 |
"""Get informations about the volume group.
|
585 |
|
586 |
Args:
|
587 |
vg_name: the volume group
|
588 |
|
589 |
Returns:
|
590 |
{ 'vg_size' : xxx, 'vg_free' : xxx, 'pv_count' : xxx }
|
591 |
where
|
592 |
vg_size is the total size of the volume group in MiB
|
593 |
vg_free is the free size of the volume group in MiB
|
594 |
pv_count are the number of physical disks in that vg
|
595 |
|
596 |
If an error occurs during gathering of data, we return the same dict
|
597 |
with keys all set to None.
|
598 |
|
599 |
"""
|
600 |
retdic = dict.fromkeys(["vg_size", "vg_free", "pv_count"]) |
601 |
|
602 |
retval = utils.RunCmd(["vgs", "-ovg_size,vg_free,pv_count", "--noheadings", |
603 |
"--nosuffix", "--units=m", "--separator=:", vg_name]) |
604 |
|
605 |
if retval.failed:
|
606 |
logging.error("volume group %s not present", vg_name)
|
607 |
return retdic
|
608 |
valarr = retval.stdout.strip().rstrip(':').split(':') |
609 |
if len(valarr) == 3: |
610 |
try:
|
611 |
retdic = { |
612 |
"vg_size": int(round(float(valarr[0]), 0)), |
613 |
"vg_free": int(round(float(valarr[1]), 0)), |
614 |
"pv_count": int(valarr[2]), |
615 |
} |
616 |
except ValueError, err: |
617 |
logging.exception("Fail to parse vgs output")
|
618 |
else:
|
619 |
logging.error("vgs output has the wrong number of fields (expected"
|
620 |
" three): %s", str(valarr)) |
621 |
return retdic
|
622 |
|
623 |
|
624 |
def _GatherBlockDevs(instance): |
625 |
"""Set up an instance's block device(s).
|
626 |
|
627 |
This is run on the primary node at instance startup. The block
|
628 |
devices must be already assembled.
|
629 |
|
630 |
"""
|
631 |
block_devices = [] |
632 |
for disk in instance.disks: |
633 |
device = _RecursiveFindBD(disk) |
634 |
if device is None: |
635 |
raise errors.BlockDeviceError("Block device '%s' is not set up." % |
636 |
str(disk))
|
637 |
device.Open() |
638 |
block_devices.append((disk, device)) |
639 |
return block_devices
|
640 |
|
641 |
|
642 |
def StartInstance(instance, extra_args): |
643 |
"""Start an instance.
|
644 |
|
645 |
@type instance: instance object
|
646 |
@param instance: the instance object
|
647 |
@rtype: boolean
|
648 |
@return: whether the startup was successful or not
|
649 |
|
650 |
"""
|
651 |
running_instances = GetInstanceList([instance.hypervisor]) |
652 |
|
653 |
if instance.name in running_instances: |
654 |
return True |
655 |
|
656 |
block_devices = _GatherBlockDevs(instance) |
657 |
hyper = hypervisor.GetHypervisor(instance.hypervisor) |
658 |
|
659 |
try:
|
660 |
hyper.StartInstance(instance, block_devices, extra_args) |
661 |
except errors.HypervisorError, err:
|
662 |
logging.exception("Failed to start instance")
|
663 |
return False |
664 |
|
665 |
return True |
666 |
|
667 |
|
668 |
def ShutdownInstance(instance): |
669 |
"""Shut an instance down.
|
670 |
|
671 |
@type instance: instance object
|
672 |
@param instance: the instance object
|
673 |
@rtype: boolean
|
674 |
@return: whether the startup was successful or not
|
675 |
|
676 |
"""
|
677 |
hv_name = instance.hypervisor |
678 |
running_instances = GetInstanceList([hv_name]) |
679 |
|
680 |
if instance.name not in running_instances: |
681 |
return True |
682 |
|
683 |
hyper = hypervisor.GetHypervisor(hv_name) |
684 |
try:
|
685 |
hyper.StopInstance(instance) |
686 |
except errors.HypervisorError, err:
|
687 |
logging.error("Failed to stop instance")
|
688 |
return False |
689 |
|
690 |
# test every 10secs for 2min
|
691 |
shutdown_ok = False
|
692 |
|
693 |
time.sleep(1)
|
694 |
for dummy in range(11): |
695 |
if instance.name not in GetInstanceList([hv_name]): |
696 |
break
|
697 |
time.sleep(10)
|
698 |
else:
|
699 |
# the shutdown did not succeed
|
700 |
logging.error("shutdown of '%s' unsuccessful, using destroy", instance)
|
701 |
|
702 |
try:
|
703 |
hyper.StopInstance(instance, force=True)
|
704 |
except errors.HypervisorError, err:
|
705 |
logging.exception("Failed to stop instance")
|
706 |
return False |
707 |
|
708 |
time.sleep(1)
|
709 |
if instance.name in GetInstanceList([hv_name]): |
710 |
logging.error("could not shutdown instance '%s' even by destroy",
|
711 |
instance.name) |
712 |
return False |
713 |
|
714 |
return True |
715 |
|
716 |
|
717 |
def RebootInstance(instance, reboot_type, extra_args): |
718 |
"""Reboot an instance.
|
719 |
|
720 |
Args:
|
721 |
instance - name of instance to reboot
|
722 |
reboot_type - how to reboot [soft,hard,full]
|
723 |
|
724 |
"""
|
725 |
running_instances = GetInstanceList([instance.hypervisor]) |
726 |
|
727 |
if instance.name not in running_instances: |
728 |
logging.error("Cannot reboot instance that is not running")
|
729 |
return False |
730 |
|
731 |
hyper = hypervisor.GetHypervisor(instance.hypervisor) |
732 |
if reboot_type == constants.INSTANCE_REBOOT_SOFT:
|
733 |
try:
|
734 |
hyper.RebootInstance(instance) |
735 |
except errors.HypervisorError, err:
|
736 |
logging.exception("Failed to soft reboot instance")
|
737 |
return False |
738 |
elif reboot_type == constants.INSTANCE_REBOOT_HARD:
|
739 |
try:
|
740 |
ShutdownInstance(instance) |
741 |
StartInstance(instance, extra_args) |
742 |
except errors.HypervisorError, err:
|
743 |
logging.exception("Failed to hard reboot instance")
|
744 |
return False |
745 |
else:
|
746 |
raise errors.ParameterError("reboot_type invalid") |
747 |
|
748 |
return True |
749 |
|
750 |
|
751 |
def MigrateInstance(instance, target, live): |
752 |
"""Migrates an instance to another node.
|
753 |
|
754 |
@type instance: L{objects.Instance}
|
755 |
@param instance: the instance definition
|
756 |
@type target: string
|
757 |
@param target: the target node name
|
758 |
@type live: boolean
|
759 |
@param live: whether the migration should be done live or not (the
|
760 |
interpretation of this parameter is left to the hypervisor)
|
761 |
@rtype: tuple
|
762 |
@return: a tuple of (success, msg) where:
|
763 |
- succes is a boolean denoting the success/failure of the operation
|
764 |
- msg is a string with details in case of failure
|
765 |
|
766 |
"""
|
767 |
hyper = hypervisor.GetHypervisor(instance.hypervisor_name) |
768 |
|
769 |
try:
|
770 |
hyper.MigrateInstance(instance.name, target, live) |
771 |
except errors.HypervisorError, err:
|
772 |
msg = "Failed to migrate instance: %s" % str(err) |
773 |
logging.error(msg) |
774 |
return (False, msg) |
775 |
return (True, "Migration successfull") |
776 |
|
777 |
|
778 |
def CreateBlockDevice(disk, size, owner, on_primary, info): |
779 |
"""Creates a block device for an instance.
|
780 |
|
781 |
@type disk: L{objects.Disk}
|
782 |
@param disk: the object describing the disk we should create
|
783 |
@type size: int
|
784 |
@param size: the size of the physical underlying device, in MiB
|
785 |
@type owner: str
|
786 |
@param owner: the name of the instance for which disk is created,
|
787 |
used for device cache data
|
788 |
@type on_primary: boolean
|
789 |
@param on_primary: indicates if it is the primary node or not
|
790 |
@type info: string
|
791 |
@param info: string that will be sent to the physical device
|
792 |
creation, used for example to set (LVM) tags on LVs
|
793 |
|
794 |
@return: the new unique_id of the device (this can sometime be
|
795 |
computed only after creation), or None. On secondary nodes,
|
796 |
it's not required to return anything.
|
797 |
|
798 |
"""
|
799 |
clist = [] |
800 |
if disk.children:
|
801 |
for child in disk.children: |
802 |
crdev = _RecursiveAssembleBD(child, owner, on_primary) |
803 |
if on_primary or disk.AssembleOnSecondary(): |
804 |
# we need the children open in case the device itself has to
|
805 |
# be assembled
|
806 |
crdev.Open() |
807 |
clist.append(crdev) |
808 |
try:
|
809 |
device = bdev.FindDevice(disk.dev_type, disk.physical_id, clist) |
810 |
if device is not None: |
811 |
logging.info("removing existing device %s", disk)
|
812 |
device.Remove() |
813 |
except errors.BlockDeviceError, err:
|
814 |
pass
|
815 |
|
816 |
device = bdev.Create(disk.dev_type, disk.physical_id, |
817 |
clist, size) |
818 |
if device is None: |
819 |
raise ValueError("Can't create child device for %s, %s" % |
820 |
(disk, size)) |
821 |
if on_primary or disk.AssembleOnSecondary(): |
822 |
if not device.Assemble(): |
823 |
errorstring = "Can't assemble device after creation"
|
824 |
logging.error(errorstring) |
825 |
raise errors.BlockDeviceError("%s, very unusual event - check the node" |
826 |
" daemon logs" % errorstring)
|
827 |
device.SetSyncSpeed(constants.SYNC_SPEED) |
828 |
if on_primary or disk.OpenOnSecondary(): |
829 |
device.Open(force=True)
|
830 |
DevCacheManager.UpdateCache(device.dev_path, owner, |
831 |
on_primary, disk.iv_name) |
832 |
|
833 |
device.SetInfo(info) |
834 |
|
835 |
physical_id = device.unique_id |
836 |
return physical_id
|
837 |
|
838 |
|
839 |
def RemoveBlockDevice(disk): |
840 |
"""Remove a block device.
|
841 |
|
842 |
This is intended to be called recursively.
|
843 |
|
844 |
"""
|
845 |
try:
|
846 |
# since we are removing the device, allow a partial match
|
847 |
# this allows removal of broken mirrors
|
848 |
rdev = _RecursiveFindBD(disk, allow_partial=True)
|
849 |
except errors.BlockDeviceError, err:
|
850 |
# probably can't attach
|
851 |
logging.info("Can't attach to device %s in remove", disk)
|
852 |
rdev = None
|
853 |
if rdev is not None: |
854 |
r_path = rdev.dev_path |
855 |
result = rdev.Remove() |
856 |
if result:
|
857 |
DevCacheManager.RemoveCache(r_path) |
858 |
else:
|
859 |
result = True
|
860 |
if disk.children:
|
861 |
for child in disk.children: |
862 |
result = result and RemoveBlockDevice(child)
|
863 |
return result
|
864 |
|
865 |
|
866 |
def _RecursiveAssembleBD(disk, owner, as_primary): |
867 |
"""Activate a block device for an instance.
|
868 |
|
869 |
This is run on the primary and secondary nodes for an instance.
|
870 |
|
871 |
This function is called recursively.
|
872 |
|
873 |
Args:
|
874 |
disk: a objects.Disk object
|
875 |
as_primary: if we should make the block device read/write
|
876 |
|
877 |
Returns:
|
878 |
the assembled device or None (in case no device was assembled)
|
879 |
|
880 |
If the assembly is not successful, an exception is raised.
|
881 |
|
882 |
"""
|
883 |
children = [] |
884 |
if disk.children:
|
885 |
mcn = disk.ChildrenNeeded() |
886 |
if mcn == -1: |
887 |
mcn = 0 # max number of Nones allowed |
888 |
else:
|
889 |
mcn = len(disk.children) - mcn # max number of Nones |
890 |
for chld_disk in disk.children: |
891 |
try:
|
892 |
cdev = _RecursiveAssembleBD(chld_disk, owner, as_primary) |
893 |
except errors.BlockDeviceError, err:
|
894 |
if children.count(None) >= mcn: |
895 |
raise
|
896 |
cdev = None
|
897 |
logging.debug("Error in child activation: %s", str(err)) |
898 |
children.append(cdev) |
899 |
|
900 |
if as_primary or disk.AssembleOnSecondary(): |
901 |
r_dev = bdev.AttachOrAssemble(disk.dev_type, disk.physical_id, children) |
902 |
r_dev.SetSyncSpeed(constants.SYNC_SPEED) |
903 |
result = r_dev |
904 |
if as_primary or disk.OpenOnSecondary(): |
905 |
r_dev.Open() |
906 |
DevCacheManager.UpdateCache(r_dev.dev_path, owner, |
907 |
as_primary, disk.iv_name) |
908 |
|
909 |
else:
|
910 |
result = True
|
911 |
return result
|
912 |
|
913 |
|
914 |
def AssembleBlockDevice(disk, owner, as_primary): |
915 |
"""Activate a block device for an instance.
|
916 |
|
917 |
This is a wrapper over _RecursiveAssembleBD.
|
918 |
|
919 |
@rtype: str or boolean
|
920 |
@return: a C{/dev/...} path for primary nodes, and
|
921 |
C{True} for secondary nodes
|
922 |
|
923 |
"""
|
924 |
result = _RecursiveAssembleBD(disk, owner, as_primary) |
925 |
if isinstance(result, bdev.BlockDev): |
926 |
result = result.dev_path |
927 |
return result
|
928 |
|
929 |
|
930 |
def ShutdownBlockDevice(disk): |
931 |
"""Shut down a block device.
|
932 |
|
933 |
First, if the device is assembled (can `Attach()`), then the device
|
934 |
is shutdown. Then the children of the device are shutdown.
|
935 |
|
936 |
This function is called recursively. Note that we don't cache the
|
937 |
children or such, as oppossed to assemble, shutdown of different
|
938 |
devices doesn't require that the upper device was active.
|
939 |
|
940 |
"""
|
941 |
r_dev = _RecursiveFindBD(disk) |
942 |
if r_dev is not None: |
943 |
r_path = r_dev.dev_path |
944 |
result = r_dev.Shutdown() |
945 |
if result:
|
946 |
DevCacheManager.RemoveCache(r_path) |
947 |
else:
|
948 |
result = True
|
949 |
if disk.children:
|
950 |
for child in disk.children: |
951 |
result = result and ShutdownBlockDevice(child)
|
952 |
return result
|
953 |
|
954 |
|
955 |
def MirrorAddChildren(parent_cdev, new_cdevs): |
956 |
"""Extend a mirrored block device.
|
957 |
|
958 |
"""
|
959 |
parent_bdev = _RecursiveFindBD(parent_cdev, allow_partial=True)
|
960 |
if parent_bdev is None: |
961 |
logging.error("Can't find parent device")
|
962 |
return False |
963 |
new_bdevs = [_RecursiveFindBD(disk) for disk in new_cdevs] |
964 |
if new_bdevs.count(None) > 0: |
965 |
logging.error("Can't find new device(s) to add: %s:%s",
|
966 |
new_bdevs, new_cdevs) |
967 |
return False |
968 |
parent_bdev.AddChildren(new_bdevs) |
969 |
return True |
970 |
|
971 |
|
972 |
def MirrorRemoveChildren(parent_cdev, new_cdevs): |
973 |
"""Shrink a mirrored block device.
|
974 |
|
975 |
"""
|
976 |
parent_bdev = _RecursiveFindBD(parent_cdev) |
977 |
if parent_bdev is None: |
978 |
logging.error("Can't find parent in remove children: %s", parent_cdev)
|
979 |
return False |
980 |
devs = [] |
981 |
for disk in new_cdevs: |
982 |
rpath = disk.StaticDevPath() |
983 |
if rpath is None: |
984 |
bd = _RecursiveFindBD(disk) |
985 |
if bd is None: |
986 |
logging.error("Can't find dynamic device %s while removing children",
|
987 |
disk) |
988 |
return False |
989 |
else:
|
990 |
devs.append(bd.dev_path) |
991 |
else:
|
992 |
devs.append(rpath) |
993 |
parent_bdev.RemoveChildren(devs) |
994 |
return True |
995 |
|
996 |
|
997 |
def GetMirrorStatus(disks): |
998 |
"""Get the mirroring status of a list of devices.
|
999 |
|
1000 |
Args:
|
1001 |
disks: list of `objects.Disk`
|
1002 |
|
1003 |
Returns:
|
1004 |
list of (mirror_done, estimated_time) tuples, which
|
1005 |
are the result of bdev.BlockDevice.CombinedSyncStatus()
|
1006 |
|
1007 |
"""
|
1008 |
stats = [] |
1009 |
for dsk in disks: |
1010 |
rbd = _RecursiveFindBD(dsk) |
1011 |
if rbd is None: |
1012 |
raise errors.BlockDeviceError("Can't find device %s" % str(dsk)) |
1013 |
stats.append(rbd.CombinedSyncStatus()) |
1014 |
return stats
|
1015 |
|
1016 |
|
1017 |
def _RecursiveFindBD(disk, allow_partial=False): |
1018 |
"""Check if a device is activated.
|
1019 |
|
1020 |
If so, return informations about the real device.
|
1021 |
|
1022 |
Args:
|
1023 |
disk: the objects.Disk instance
|
1024 |
allow_partial: don't abort the find if a child of the
|
1025 |
device can't be found; this is intended to be
|
1026 |
used when repairing mirrors
|
1027 |
|
1028 |
Returns:
|
1029 |
None if the device can't be found
|
1030 |
otherwise the device instance
|
1031 |
|
1032 |
"""
|
1033 |
children = [] |
1034 |
if disk.children:
|
1035 |
for chdisk in disk.children: |
1036 |
children.append(_RecursiveFindBD(chdisk)) |
1037 |
|
1038 |
return bdev.FindDevice(disk.dev_type, disk.physical_id, children)
|
1039 |
|
1040 |
|
1041 |
def FindBlockDevice(disk): |
1042 |
"""Check if a device is activated.
|
1043 |
|
1044 |
If so, return informations about the real device.
|
1045 |
|
1046 |
Args:
|
1047 |
disk: the objects.Disk instance
|
1048 |
Returns:
|
1049 |
None if the device can't be found
|
1050 |
(device_path, major, minor, sync_percent, estimated_time, is_degraded)
|
1051 |
|
1052 |
"""
|
1053 |
rbd = _RecursiveFindBD(disk) |
1054 |
if rbd is None: |
1055 |
return rbd
|
1056 |
return (rbd.dev_path, rbd.major, rbd.minor) + rbd.GetSyncStatus()
|
1057 |
|
1058 |
|
1059 |
def UploadFile(file_name, data, mode, uid, gid, atime, mtime): |
1060 |
"""Write a file to the filesystem.
|
1061 |
|
1062 |
This allows the master to overwrite(!) a file. It will only perform
|
1063 |
the operation if the file belongs to a list of configuration files.
|
1064 |
|
1065 |
"""
|
1066 |
if not os.path.isabs(file_name): |
1067 |
logging.error("Filename passed to UploadFile is not absolute: '%s'",
|
1068 |
file_name) |
1069 |
return False |
1070 |
|
1071 |
allowed_files = [ |
1072 |
constants.CLUSTER_CONF_FILE, |
1073 |
constants.ETC_HOSTS, |
1074 |
constants.SSH_KNOWN_HOSTS_FILE, |
1075 |
constants.VNC_PASSWORD_FILE, |
1076 |
] |
1077 |
|
1078 |
if file_name not in allowed_files: |
1079 |
logging.error("Filename passed to UploadFile not in allowed"
|
1080 |
" upload targets: '%s'", file_name)
|
1081 |
return False |
1082 |
|
1083 |
utils.WriteFile(file_name, data=data, mode=mode, uid=uid, gid=gid, |
1084 |
atime=atime, mtime=mtime) |
1085 |
return True |
1086 |
|
1087 |
|
1088 |
def _ErrnoOrStr(err): |
1089 |
"""Format an EnvironmentError exception.
|
1090 |
|
1091 |
If the `err` argument has an errno attribute, it will be looked up
|
1092 |
and converted into a textual EXXXX description. Otherwise the string
|
1093 |
representation of the error will be returned.
|
1094 |
|
1095 |
"""
|
1096 |
if hasattr(err, 'errno'): |
1097 |
detail = errno.errorcode[err.errno] |
1098 |
else:
|
1099 |
detail = str(err)
|
1100 |
return detail
|
1101 |
|
1102 |
|
1103 |
def _OSOndiskVersion(name, os_dir): |
1104 |
"""Compute and return the API version of a given OS.
|
1105 |
|
1106 |
This function will try to read the API version of the os given by
|
1107 |
the 'name' parameter and residing in the 'os_dir' directory.
|
1108 |
|
1109 |
Return value will be either an integer denoting the version or None in the
|
1110 |
case when this is not a valid OS name.
|
1111 |
|
1112 |
"""
|
1113 |
api_file = os.path.sep.join([os_dir, "ganeti_api_version"])
|
1114 |
|
1115 |
try:
|
1116 |
st = os.stat(api_file) |
1117 |
except EnvironmentError, err: |
1118 |
raise errors.InvalidOS(name, os_dir, "'ganeti_api_version' file not" |
1119 |
" found (%s)" % _ErrnoOrStr(err))
|
1120 |
|
1121 |
if not stat.S_ISREG(stat.S_IFMT(st.st_mode)): |
1122 |
raise errors.InvalidOS(name, os_dir, "'ganeti_api_version' file is not" |
1123 |
" a regular file")
|
1124 |
|
1125 |
try:
|
1126 |
f = open(api_file)
|
1127 |
try:
|
1128 |
api_versions = f.readlines() |
1129 |
finally:
|
1130 |
f.close() |
1131 |
except EnvironmentError, err: |
1132 |
raise errors.InvalidOS(name, os_dir, "error while reading the" |
1133 |
" API version (%s)" % _ErrnoOrStr(err))
|
1134 |
|
1135 |
api_versions = [version.strip() for version in api_versions] |
1136 |
try:
|
1137 |
api_versions = [int(version) for version in api_versions] |
1138 |
except (TypeError, ValueError), err: |
1139 |
raise errors.InvalidOS(name, os_dir,
|
1140 |
"API version is not integer (%s)" % str(err)) |
1141 |
|
1142 |
return api_versions
|
1143 |
|
1144 |
|
1145 |
def DiagnoseOS(top_dirs=None): |
1146 |
"""Compute the validity for all OSes.
|
1147 |
|
1148 |
Returns an OS object for each name in all the given top directories
|
1149 |
(if not given defaults to constants.OS_SEARCH_PATH)
|
1150 |
|
1151 |
Returns:
|
1152 |
list of OS objects
|
1153 |
|
1154 |
"""
|
1155 |
if top_dirs is None: |
1156 |
top_dirs = constants.OS_SEARCH_PATH |
1157 |
|
1158 |
result = [] |
1159 |
for dir_name in top_dirs: |
1160 |
if os.path.isdir(dir_name):
|
1161 |
try:
|
1162 |
f_names = utils.ListVisibleFiles(dir_name) |
1163 |
except EnvironmentError, err: |
1164 |
logging.exception("Can't list the OS directory %s", dir_name)
|
1165 |
break
|
1166 |
for name in f_names: |
1167 |
try:
|
1168 |
os_inst = OSFromDisk(name, base_dir=dir_name) |
1169 |
result.append(os_inst) |
1170 |
except errors.InvalidOS, err:
|
1171 |
result.append(objects.OS.FromInvalidOS(err)) |
1172 |
|
1173 |
return result
|
1174 |
|
1175 |
|
1176 |
def OSFromDisk(name, base_dir=None): |
1177 |
"""Create an OS instance from disk.
|
1178 |
|
1179 |
This function will return an OS instance if the given name is a
|
1180 |
valid OS name. Otherwise, it will raise an appropriate
|
1181 |
`errors.InvalidOS` exception, detailing why this is not a valid
|
1182 |
OS.
|
1183 |
|
1184 |
@type base_dir: string
|
1185 |
@keyword base_dir: Base directory containing OS installations.
|
1186 |
Defaults to a search in all the OS_SEARCH_PATH dirs.
|
1187 |
|
1188 |
"""
|
1189 |
|
1190 |
if base_dir is None: |
1191 |
os_dir = utils.FindFile(name, constants.OS_SEARCH_PATH, os.path.isdir) |
1192 |
if os_dir is None: |
1193 |
raise errors.InvalidOS(name, None, "OS dir not found in search path") |
1194 |
else:
|
1195 |
os_dir = os.path.sep.join([base_dir, name]) |
1196 |
|
1197 |
api_versions = _OSOndiskVersion(name, os_dir) |
1198 |
|
1199 |
if constants.OS_API_VERSION not in api_versions: |
1200 |
raise errors.InvalidOS(name, os_dir, "API version mismatch" |
1201 |
" (found %s want %s)"
|
1202 |
% (api_versions, constants.OS_API_VERSION)) |
1203 |
|
1204 |
# OS Scripts dictionary, we will populate it with the actual script names
|
1205 |
os_scripts = dict.fromkeys(constants.OS_SCRIPTS)
|
1206 |
|
1207 |
for script in os_scripts: |
1208 |
os_scripts[script] = os.path.sep.join([os_dir, script]) |
1209 |
|
1210 |
try:
|
1211 |
st = os.stat(os_scripts[script]) |
1212 |
except EnvironmentError, err: |
1213 |
raise errors.InvalidOS(name, os_dir, "'%s' script missing (%s)" % |
1214 |
(script, _ErrnoOrStr(err))) |
1215 |
|
1216 |
if stat.S_IMODE(st.st_mode) & stat.S_IXUSR != stat.S_IXUSR:
|
1217 |
raise errors.InvalidOS(name, os_dir, "'%s' script not executable" % |
1218 |
script) |
1219 |
|
1220 |
if not stat.S_ISREG(stat.S_IFMT(st.st_mode)): |
1221 |
raise errors.InvalidOS(name, os_dir, "'%s' is not a regular file" % |
1222 |
script) |
1223 |
|
1224 |
|
1225 |
return objects.OS(name=name, path=os_dir, status=constants.OS_VALID_STATUS,
|
1226 |
create_script=os_scripts[constants.OS_SCRIPT_CREATE], |
1227 |
export_script=os_scripts[constants.OS_SCRIPT_EXPORT], |
1228 |
import_script=os_scripts[constants.OS_SCRIPT_IMPORT], |
1229 |
rename_script=os_scripts[constants.OS_SCRIPT_RENAME], |
1230 |
api_versions=api_versions) |
1231 |
|
1232 |
def OSEnvironment(instance, debug=0): |
1233 |
"""Calculate the environment for an os script.
|
1234 |
|
1235 |
@type instance: instance object
|
1236 |
@param instance: target instance for the os script run
|
1237 |
@type debug: integer
|
1238 |
@param debug: debug level (0 or 1, for os api 10)
|
1239 |
@rtype: dict
|
1240 |
@return: dict of environment variables
|
1241 |
|
1242 |
"""
|
1243 |
result = {} |
1244 |
result['OS_API_VERSION'] = '%d' % constants.OS_API_VERSION |
1245 |
result['INSTANCE_NAME'] = instance.name
|
1246 |
result['HYPERVISOR'] = instance.hypervisor
|
1247 |
result['DISK_COUNT'] = '%d' % len(instance.disks) |
1248 |
result['NIC_COUNT'] = '%d' % len(instance.nics) |
1249 |
result['DEBUG_LEVEL'] = '%d' % debug |
1250 |
for idx, disk in enumerate(instance.disks): |
1251 |
real_disk = _RecursiveFindBD(disk) |
1252 |
if real_disk is None: |
1253 |
raise errors.BlockDeviceError("Block device '%s' is not set up" % |
1254 |
str(disk))
|
1255 |
real_disk.Open() |
1256 |
result['DISK_%d_PATH' % idx] = real_disk.dev_path
|
1257 |
# FIXME: When disks will have read-only mode, populate this
|
1258 |
result['DISK_%d_ACCESS' % idx] = 'W' |
1259 |
if constants.HV_DISK_TYPE in instance.hvparams: |
1260 |
result['DISK_%d_FRONTEND_TYPE' % idx] = \
|
1261 |
instance.hvparams[constants.HV_DISK_TYPE] |
1262 |
if disk.dev_type in constants.LDS_BLOCK: |
1263 |
result['DISK_%d_BACKEND_TYPE' % idx] = 'block' |
1264 |
elif disk.dev_type == constants.LD_FILE:
|
1265 |
result['DISK_%d_BACKEND_TYPE' % idx] = \
|
1266 |
'file:%s' % disk.physical_id[0] |
1267 |
for idx, nic in enumerate(instance.nics): |
1268 |
result['NIC_%d_MAC' % idx] = nic.mac
|
1269 |
if nic.ip:
|
1270 |
result['NIC_%d_IP' % idx] = nic.ip
|
1271 |
result['NIC_%d_BRIDGE' % idx] = nic.bridge
|
1272 |
if constants.HV_NIC_TYPE in instance.hvparams: |
1273 |
result['NIC_%d_FRONTEND_TYPE' % idx] = \
|
1274 |
instance.hvparams[constants.HV_NIC_TYPE] |
1275 |
|
1276 |
return result
|
1277 |
|
1278 |
def GrowBlockDevice(disk, amount): |
1279 |
"""Grow a stack of block devices.
|
1280 |
|
1281 |
This function is called recursively, with the childrens being the
|
1282 |
first one resize.
|
1283 |
|
1284 |
Args:
|
1285 |
disk: the disk to be grown
|
1286 |
|
1287 |
Returns: a tuple of (status, result), with:
|
1288 |
status: the result (true/false) of the operation
|
1289 |
result: the error message if the operation failed, otherwise not used
|
1290 |
|
1291 |
"""
|
1292 |
r_dev = _RecursiveFindBD(disk) |
1293 |
if r_dev is None: |
1294 |
return False, "Cannot find block device %s" % (disk,) |
1295 |
|
1296 |
try:
|
1297 |
r_dev.Grow(amount) |
1298 |
except errors.BlockDeviceError, err:
|
1299 |
return False, str(err) |
1300 |
|
1301 |
return True, None |
1302 |
|
1303 |
|
1304 |
def SnapshotBlockDevice(disk): |
1305 |
"""Create a snapshot copy of a block device.
|
1306 |
|
1307 |
This function is called recursively, and the snapshot is actually created
|
1308 |
just for the leaf lvm backend device.
|
1309 |
|
1310 |
@type disk: L{objects.Disk}
|
1311 |
@param disk: the disk to be snapshotted
|
1312 |
@rtype: string
|
1313 |
@return: snapshot disk path
|
1314 |
|
1315 |
"""
|
1316 |
if disk.children:
|
1317 |
if len(disk.children) == 1: |
1318 |
# only one child, let's recurse on it
|
1319 |
return SnapshotBlockDevice(disk.children[0]) |
1320 |
else:
|
1321 |
# more than one child, choose one that matches
|
1322 |
for child in disk.children: |
1323 |
if child.size == disk.size:
|
1324 |
# return implies breaking the loop
|
1325 |
return SnapshotBlockDevice(child)
|
1326 |
elif disk.dev_type == constants.LD_LV:
|
1327 |
r_dev = _RecursiveFindBD(disk) |
1328 |
if r_dev is not None: |
1329 |
# let's stay on the safe side and ask for the full size, for now
|
1330 |
return r_dev.Snapshot(disk.size)
|
1331 |
else:
|
1332 |
return None |
1333 |
else:
|
1334 |
raise errors.ProgrammerError("Cannot snapshot non-lvm block device" |
1335 |
" '%s' of type '%s'" %
|
1336 |
(disk.unique_id, disk.dev_type)) |
1337 |
|
1338 |
|
1339 |
def ExportSnapshot(disk, dest_node, instance, cluster_name, idx): |
1340 |
"""Export a block device snapshot to a remote node.
|
1341 |
|
1342 |
@type disk: L{objects.Disk}
|
1343 |
@param disk: the description of the disk to export
|
1344 |
@type dest_node: str
|
1345 |
@param dest_node: the destination node to export to
|
1346 |
@type instance: L{objects.Instance}
|
1347 |
@param instance: the instance object to whom the disk belongs
|
1348 |
@type cluster_name: str
|
1349 |
@param cluster_name: the cluster name, needed for SSH hostalias
|
1350 |
@type idx: int
|
1351 |
@param idx: the index of the disk in the instance's disk list,
|
1352 |
used to export to the OS scripts environment
|
1353 |
@rtype: bool
|
1354 |
@return: the success of the operation
|
1355 |
|
1356 |
"""
|
1357 |
export_env = OSEnvironment(instance) |
1358 |
|
1359 |
inst_os = OSFromDisk(instance.os) |
1360 |
export_script = inst_os.export_script |
1361 |
|
1362 |
logfile = "%s/exp-%s-%s-%s.log" % (constants.LOG_OS_DIR, inst_os.name,
|
1363 |
instance.name, int(time.time()))
|
1364 |
if not os.path.exists(constants.LOG_OS_DIR): |
1365 |
os.mkdir(constants.LOG_OS_DIR, 0750)
|
1366 |
real_disk = _RecursiveFindBD(disk) |
1367 |
if real_disk is None: |
1368 |
raise errors.BlockDeviceError("Block device '%s' is not set up" % |
1369 |
str(disk))
|
1370 |
real_disk.Open() |
1371 |
|
1372 |
export_env['EXPORT_DEVICE'] = real_disk.dev_path
|
1373 |
export_env['EXPORT_INDEX'] = str(idx) |
1374 |
|
1375 |
destdir = os.path.join(constants.EXPORT_DIR, instance.name + ".new")
|
1376 |
destfile = disk.physical_id[1]
|
1377 |
|
1378 |
# the target command is built out of three individual commands,
|
1379 |
# which are joined by pipes; we check each individual command for
|
1380 |
# valid parameters
|
1381 |
expcmd = utils.BuildShellCmd("cd %s; %s 2>%s", inst_os.path,
|
1382 |
export_script, logfile) |
1383 |
|
1384 |
comprcmd = "gzip"
|
1385 |
|
1386 |
destcmd = utils.BuildShellCmd("mkdir -p %s && cat > %s/%s",
|
1387 |
destdir, destdir, destfile) |
1388 |
remotecmd = _GetSshRunner(cluster_name).BuildCmd(dest_node, |
1389 |
constants.GANETI_RUNAS, |
1390 |
destcmd) |
1391 |
|
1392 |
# all commands have been checked, so we're safe to combine them
|
1393 |
command = '|'.join([expcmd, comprcmd, utils.ShellQuoteArgs(remotecmd)])
|
1394 |
|
1395 |
result = utils.RunCmd(command, env=export_env) |
1396 |
|
1397 |
if result.failed:
|
1398 |
logging.error("os snapshot export command '%s' returned error: %s"
|
1399 |
" output: %s", command, result.fail_reason, result.output)
|
1400 |
return False |
1401 |
|
1402 |
return True |
1403 |
|
1404 |
|
1405 |
def FinalizeExport(instance, snap_disks): |
1406 |
"""Write out the export configuration information.
|
1407 |
|
1408 |
Args:
|
1409 |
instance: instance configuration
|
1410 |
snap_disks: snapshot block devices
|
1411 |
|
1412 |
Returns:
|
1413 |
False in case of error, True otherwise.
|
1414 |
|
1415 |
"""
|
1416 |
destdir = os.path.join(constants.EXPORT_DIR, instance.name + ".new")
|
1417 |
finaldestdir = os.path.join(constants.EXPORT_DIR, instance.name) |
1418 |
|
1419 |
config = objects.SerializableConfigParser() |
1420 |
|
1421 |
config.add_section(constants.INISECT_EXP) |
1422 |
config.set(constants.INISECT_EXP, 'version', '0') |
1423 |
config.set(constants.INISECT_EXP, 'timestamp', '%d' % int(time.time())) |
1424 |
config.set(constants.INISECT_EXP, 'source', instance.primary_node)
|
1425 |
config.set(constants.INISECT_EXP, 'os', instance.os)
|
1426 |
config.set(constants.INISECT_EXP, 'compression', 'gzip') |
1427 |
|
1428 |
config.add_section(constants.INISECT_INS) |
1429 |
config.set(constants.INISECT_INS, 'name', instance.name)
|
1430 |
config.set(constants.INISECT_INS, 'memory', '%d' % |
1431 |
instance.beparams[constants.BE_MEMORY]) |
1432 |
config.set(constants.INISECT_INS, 'vcpus', '%d' % |
1433 |
instance.beparams[constants.BE_VCPUS]) |
1434 |
config.set(constants.INISECT_INS, 'disk_template', instance.disk_template)
|
1435 |
|
1436 |
nic_count = 0
|
1437 |
for nic_count, nic in enumerate(instance.nics): |
1438 |
config.set(constants.INISECT_INS, 'nic%d_mac' %
|
1439 |
nic_count, '%s' % nic.mac)
|
1440 |
config.set(constants.INISECT_INS, 'nic%d_ip' % nic_count, '%s' % nic.ip) |
1441 |
config.set(constants.INISECT_INS, 'nic%d_bridge' % nic_count,
|
1442 |
'%s' % nic.bridge)
|
1443 |
# TODO: redundant: on load can read nics until it doesn't exist
|
1444 |
config.set(constants.INISECT_INS, 'nic_count' , '%d' % nic_count) |
1445 |
|
1446 |
disk_count = 0
|
1447 |
for disk_count, disk in enumerate(snap_disks): |
1448 |
if disk:
|
1449 |
config.set(constants.INISECT_INS, 'disk%d_ivname' % disk_count,
|
1450 |
('%s' % disk.iv_name))
|
1451 |
config.set(constants.INISECT_INS, 'disk%d_dump' % disk_count,
|
1452 |
('%s' % disk.physical_id[1])) |
1453 |
config.set(constants.INISECT_INS, 'disk%d_size' % disk_count,
|
1454 |
('%d' % disk.size))
|
1455 |
config.set(constants.INISECT_INS, 'disk_count' , '%d' % disk_count) |
1456 |
|
1457 |
cff = os.path.join(destdir, constants.EXPORT_CONF_FILE) |
1458 |
cfo = open(cff, 'w') |
1459 |
try:
|
1460 |
config.write(cfo) |
1461 |
finally:
|
1462 |
cfo.close() |
1463 |
|
1464 |
shutil.rmtree(finaldestdir, True)
|
1465 |
shutil.move(destdir, finaldestdir) |
1466 |
|
1467 |
return True |
1468 |
|
1469 |
|
1470 |
def ExportInfo(dest): |
1471 |
"""Get export configuration information.
|
1472 |
|
1473 |
Args:
|
1474 |
dest: directory containing the export
|
1475 |
|
1476 |
Returns:
|
1477 |
A serializable config file containing the export info.
|
1478 |
|
1479 |
"""
|
1480 |
cff = os.path.join(dest, constants.EXPORT_CONF_FILE) |
1481 |
|
1482 |
config = objects.SerializableConfigParser() |
1483 |
config.read(cff) |
1484 |
|
1485 |
if (not config.has_section(constants.INISECT_EXP) or |
1486 |
not config.has_section(constants.INISECT_INS)):
|
1487 |
return None |
1488 |
|
1489 |
return config
|
1490 |
|
1491 |
|
1492 |
def ImportOSIntoInstance(instance, src_node, src_images, cluster_name): |
1493 |
"""Import an os image into an instance.
|
1494 |
|
1495 |
@type instance: L{objects.Instance}
|
1496 |
@param instance: instance to import the disks into
|
1497 |
@type src_node: string
|
1498 |
@param src_node: source node for the disk images
|
1499 |
@type src_images: list of string
|
1500 |
@param src_images: absolute paths of the disk images
|
1501 |
@rtype: list of boolean
|
1502 |
@return: each boolean represent the success of importing the n-th disk
|
1503 |
|
1504 |
"""
|
1505 |
import_env = OSEnvironment(instance) |
1506 |
inst_os = OSFromDisk(instance.os) |
1507 |
import_script = inst_os.import_script |
1508 |
|
1509 |
logfile = "%s/import-%s-%s-%s.log" % (constants.LOG_OS_DIR, instance.os,
|
1510 |
instance.name, int(time.time()))
|
1511 |
if not os.path.exists(constants.LOG_OS_DIR): |
1512 |
os.mkdir(constants.LOG_OS_DIR, 0750)
|
1513 |
|
1514 |
comprcmd = "gunzip"
|
1515 |
impcmd = utils.BuildShellCmd("(cd %s; %s &>%s)", inst_os.path, import_script,
|
1516 |
logfile) |
1517 |
|
1518 |
final_result = [] |
1519 |
for idx, image in enumerate(src_images): |
1520 |
if image:
|
1521 |
destcmd = utils.BuildShellCmd('cat %s', image)
|
1522 |
remotecmd = _GetSshRunner(cluster_name).BuildCmd(src_node, |
1523 |
constants.GANETI_RUNAS, |
1524 |
destcmd) |
1525 |
command = '|'.join([utils.ShellQuoteArgs(remotecmd), comprcmd, impcmd])
|
1526 |
import_env['IMPORT_DEVICE'] = import_env['DISK_%d_PATH' % idx] |
1527 |
import_env['IMPORT_INDEX'] = str(idx) |
1528 |
result = utils.RunCmd(command, env=import_env) |
1529 |
if result.failed:
|
1530 |
logging.error("disk import command '%s' returned error: %s"
|
1531 |
" output: %s", command, result.fail_reason, result.output)
|
1532 |
final_result.append(False)
|
1533 |
else:
|
1534 |
final_result.append(True)
|
1535 |
else:
|
1536 |
final_result.append(True)
|
1537 |
|
1538 |
return final_result
|
1539 |
|
1540 |
|
1541 |
def ListExports(): |
1542 |
"""Return a list of exports currently available on this machine.
|
1543 |
|
1544 |
"""
|
1545 |
if os.path.isdir(constants.EXPORT_DIR):
|
1546 |
return utils.ListVisibleFiles(constants.EXPORT_DIR)
|
1547 |
else:
|
1548 |
return []
|
1549 |
|
1550 |
|
1551 |
def RemoveExport(export): |
1552 |
"""Remove an existing export from the node.
|
1553 |
|
1554 |
Args:
|
1555 |
export: the name of the export to remove
|
1556 |
|
1557 |
Returns:
|
1558 |
False in case of error, True otherwise.
|
1559 |
|
1560 |
"""
|
1561 |
target = os.path.join(constants.EXPORT_DIR, export) |
1562 |
|
1563 |
shutil.rmtree(target) |
1564 |
# TODO: catch some of the relevant exceptions and provide a pretty
|
1565 |
# error message if rmtree fails.
|
1566 |
|
1567 |
return True |
1568 |
|
1569 |
|
1570 |
def RenameBlockDevices(devlist): |
1571 |
"""Rename a list of block devices.
|
1572 |
|
1573 |
The devlist argument is a list of tuples (disk, new_logical,
|
1574 |
new_physical). The return value will be a combined boolean result
|
1575 |
(True only if all renames succeeded).
|
1576 |
|
1577 |
"""
|
1578 |
result = True
|
1579 |
for disk, unique_id in devlist: |
1580 |
dev = _RecursiveFindBD(disk) |
1581 |
if dev is None: |
1582 |
result = False
|
1583 |
continue
|
1584 |
try:
|
1585 |
old_rpath = dev.dev_path |
1586 |
dev.Rename(unique_id) |
1587 |
new_rpath = dev.dev_path |
1588 |
if old_rpath != new_rpath:
|
1589 |
DevCacheManager.RemoveCache(old_rpath) |
1590 |
# FIXME: we should add the new cache information here, like:
|
1591 |
# DevCacheManager.UpdateCache(new_rpath, owner, ...)
|
1592 |
# but we don't have the owner here - maybe parse from existing
|
1593 |
# cache? for now, we only lose lvm data when we rename, which
|
1594 |
# is less critical than DRBD or MD
|
1595 |
except errors.BlockDeviceError, err:
|
1596 |
logging.exception("Can't rename device '%s' to '%s'", dev, unique_id)
|
1597 |
result = False
|
1598 |
return result
|
1599 |
|
1600 |
|
1601 |
def _TransformFileStorageDir(file_storage_dir): |
1602 |
"""Checks whether given file_storage_dir is valid.
|
1603 |
|
1604 |
Checks wheter the given file_storage_dir is within the cluster-wide
|
1605 |
default file_storage_dir stored in SimpleStore. Only paths under that
|
1606 |
directory are allowed.
|
1607 |
|
1608 |
@type file_storage_dir: str
|
1609 |
@param file_storage_dir: the path to check
|
1610 |
|
1611 |
@return: the normalized path if valid, None otherwise
|
1612 |
|
1613 |
"""
|
1614 |
cfg = _GetConfig() |
1615 |
file_storage_dir = os.path.normpath(file_storage_dir) |
1616 |
base_file_storage_dir = cfg.GetFileStorageDir() |
1617 |
if (not os.path.commonprefix([file_storage_dir, base_file_storage_dir]) == |
1618 |
base_file_storage_dir): |
1619 |
logging.error("file storage directory '%s' is not under base file"
|
1620 |
" storage directory '%s'",
|
1621 |
file_storage_dir, base_file_storage_dir) |
1622 |
return None |
1623 |
return file_storage_dir
|
1624 |
|
1625 |
|
1626 |
def CreateFileStorageDir(file_storage_dir): |
1627 |
"""Create file storage directory.
|
1628 |
|
1629 |
@type file_storage_dir: str
|
1630 |
@param file_storage_dir: directory to create
|
1631 |
|
1632 |
@rtype: tuple
|
1633 |
@return: tuple with first element a boolean indicating wheter dir
|
1634 |
creation was successful or not
|
1635 |
|
1636 |
"""
|
1637 |
file_storage_dir = _TransformFileStorageDir(file_storage_dir) |
1638 |
result = True,
|
1639 |
if not file_storage_dir: |
1640 |
result = False,
|
1641 |
else:
|
1642 |
if os.path.exists(file_storage_dir):
|
1643 |
if not os.path.isdir(file_storage_dir): |
1644 |
logging.error("'%s' is not a directory", file_storage_dir)
|
1645 |
result = False,
|
1646 |
else:
|
1647 |
try:
|
1648 |
os.makedirs(file_storage_dir, 0750)
|
1649 |
except OSError, err: |
1650 |
logging.error("Cannot create file storage directory '%s': %s",
|
1651 |
file_storage_dir, err) |
1652 |
result = False,
|
1653 |
return result
|
1654 |
|
1655 |
|
1656 |
def RemoveFileStorageDir(file_storage_dir): |
1657 |
"""Remove file storage directory.
|
1658 |
|
1659 |
Remove it only if it's empty. If not log an error and return.
|
1660 |
|
1661 |
Args:
|
1662 |
file_storage_dir: string containing the path
|
1663 |
|
1664 |
Returns:
|
1665 |
tuple with first element a boolean indicating wheter dir
|
1666 |
removal was successful or not
|
1667 |
|
1668 |
"""
|
1669 |
file_storage_dir = _TransformFileStorageDir(file_storage_dir) |
1670 |
result = True,
|
1671 |
if not file_storage_dir: |
1672 |
result = False,
|
1673 |
else:
|
1674 |
if os.path.exists(file_storage_dir):
|
1675 |
if not os.path.isdir(file_storage_dir): |
1676 |
logging.error("'%s' is not a directory", file_storage_dir)
|
1677 |
result = False,
|
1678 |
# deletes dir only if empty, otherwise we want to return False
|
1679 |
try:
|
1680 |
os.rmdir(file_storage_dir) |
1681 |
except OSError, err: |
1682 |
logging.exception("Cannot remove file storage directory '%s'",
|
1683 |
file_storage_dir) |
1684 |
result = False,
|
1685 |
return result
|
1686 |
|
1687 |
|
1688 |
def RenameFileStorageDir(old_file_storage_dir, new_file_storage_dir): |
1689 |
"""Rename the file storage directory.
|
1690 |
|
1691 |
Args:
|
1692 |
old_file_storage_dir: string containing the old path
|
1693 |
new_file_storage_dir: string containing the new path
|
1694 |
|
1695 |
Returns:
|
1696 |
tuple with first element a boolean indicating wheter dir
|
1697 |
rename was successful or not
|
1698 |
|
1699 |
"""
|
1700 |
old_file_storage_dir = _TransformFileStorageDir(old_file_storage_dir) |
1701 |
new_file_storage_dir = _TransformFileStorageDir(new_file_storage_dir) |
1702 |
result = True,
|
1703 |
if not old_file_storage_dir or not new_file_storage_dir: |
1704 |
result = False,
|
1705 |
else:
|
1706 |
if not os.path.exists(new_file_storage_dir): |
1707 |
if os.path.isdir(old_file_storage_dir):
|
1708 |
try:
|
1709 |
os.rename(old_file_storage_dir, new_file_storage_dir) |
1710 |
except OSError, err: |
1711 |
logging.exception("Cannot rename '%s' to '%s'",
|
1712 |
old_file_storage_dir, new_file_storage_dir) |
1713 |
result = False,
|
1714 |
else:
|
1715 |
logging.error("'%s' is not a directory", old_file_storage_dir)
|
1716 |
result = False,
|
1717 |
else:
|
1718 |
if os.path.exists(old_file_storage_dir):
|
1719 |
logging.error("Cannot rename '%s' to '%s'. Both locations exist.",
|
1720 |
old_file_storage_dir, new_file_storage_dir) |
1721 |
result = False,
|
1722 |
return result
|
1723 |
|
1724 |
|
1725 |
def _IsJobQueueFile(file_name): |
1726 |
"""Checks whether the given filename is in the queue directory.
|
1727 |
|
1728 |
"""
|
1729 |
queue_dir = os.path.normpath(constants.QUEUE_DIR) |
1730 |
result = (os.path.commonprefix([queue_dir, file_name]) == queue_dir) |
1731 |
|
1732 |
if not result: |
1733 |
logging.error("'%s' is not a file in the queue directory",
|
1734 |
file_name) |
1735 |
|
1736 |
return result
|
1737 |
|
1738 |
|
1739 |
def JobQueueUpdate(file_name, content): |
1740 |
"""Updates a file in the queue directory.
|
1741 |
|
1742 |
"""
|
1743 |
if not _IsJobQueueFile(file_name): |
1744 |
return False |
1745 |
|
1746 |
# Write and replace the file atomically
|
1747 |
utils.WriteFile(file_name, data=content) |
1748 |
|
1749 |
return True |
1750 |
|
1751 |
|
1752 |
def JobQueueRename(old, new): |
1753 |
"""Renames a job queue file.
|
1754 |
|
1755 |
"""
|
1756 |
if not (_IsJobQueueFile(old) and _IsJobQueueFile(new)): |
1757 |
return False |
1758 |
|
1759 |
os.rename(old, new) |
1760 |
|
1761 |
return True |
1762 |
|
1763 |
|
1764 |
def JobQueueSetDrainFlag(drain_flag): |
1765 |
"""Set the drain flag for the queue.
|
1766 |
|
1767 |
This will set or unset the queue drain flag.
|
1768 |
|
1769 |
@type drain_flag: bool
|
1770 |
@param drain_flag: if True, will set the drain flag, otherwise reset it.
|
1771 |
|
1772 |
"""
|
1773 |
if drain_flag:
|
1774 |
utils.WriteFile(constants.JOB_QUEUE_DRAIN_FILE, data="", close=True) |
1775 |
else:
|
1776 |
utils.RemoveFile(constants.JOB_QUEUE_DRAIN_FILE) |
1777 |
|
1778 |
return True |
1779 |
|
1780 |
|
1781 |
def CloseBlockDevices(disks): |
1782 |
"""Closes the given block devices.
|
1783 |
|
1784 |
This means they will be switched to secondary mode (in case of DRBD).
|
1785 |
|
1786 |
"""
|
1787 |
bdevs = [] |
1788 |
for cf in disks: |
1789 |
rd = _RecursiveFindBD(cf) |
1790 |
if rd is None: |
1791 |
return (False, "Can't find device %s" % cf) |
1792 |
bdevs.append(rd) |
1793 |
|
1794 |
msg = [] |
1795 |
for rd in bdevs: |
1796 |
try:
|
1797 |
rd.Close() |
1798 |
except errors.BlockDeviceError, err:
|
1799 |
msg.append(str(err))
|
1800 |
if msg:
|
1801 |
return (False, "Can't make devices secondary: %s" % ",".join(msg)) |
1802 |
else:
|
1803 |
return (True, "All devices secondary") |
1804 |
|
1805 |
|
1806 |
def ValidateHVParams(hvname, hvparams): |
1807 |
"""Validates the given hypervisor parameters.
|
1808 |
|
1809 |
@type hvname: string
|
1810 |
@param hvname: the hypervisor name
|
1811 |
@type hvparams: dict
|
1812 |
@param hvparams: the hypervisor parameters to be validated
|
1813 |
@rtype: tuple (bool, str)
|
1814 |
@return: tuple of (success, message)
|
1815 |
|
1816 |
"""
|
1817 |
try:
|
1818 |
hv_type = hypervisor.GetHypervisor(hvname) |
1819 |
hv_type.ValidateParameters(hvparams) |
1820 |
return (True, "Validation passed") |
1821 |
except errors.HypervisorError, err:
|
1822 |
return (False, str(err)) |
1823 |
|
1824 |
|
1825 |
class HooksRunner(object): |
1826 |
"""Hook runner.
|
1827 |
|
1828 |
This class is instantiated on the node side (ganeti-noded) and not on
|
1829 |
the master side.
|
1830 |
|
1831 |
"""
|
1832 |
RE_MASK = re.compile("^[a-zA-Z0-9_-]+$")
|
1833 |
|
1834 |
def __init__(self, hooks_base_dir=None): |
1835 |
"""Constructor for hooks runner.
|
1836 |
|
1837 |
Args:
|
1838 |
- hooks_base_dir: if not None, this overrides the
|
1839 |
constants.HOOKS_BASE_DIR (useful for unittests)
|
1840 |
|
1841 |
"""
|
1842 |
if hooks_base_dir is None: |
1843 |
hooks_base_dir = constants.HOOKS_BASE_DIR |
1844 |
self._BASE_DIR = hooks_base_dir
|
1845 |
|
1846 |
@staticmethod
|
1847 |
def ExecHook(script, env): |
1848 |
"""Exec one hook script.
|
1849 |
|
1850 |
Args:
|
1851 |
- script: the full path to the script
|
1852 |
- env: the environment with which to exec the script
|
1853 |
|
1854 |
"""
|
1855 |
# exec the process using subprocess and log the output
|
1856 |
fdstdin = None
|
1857 |
try:
|
1858 |
fdstdin = open("/dev/null", "r") |
1859 |
child = subprocess.Popen([script], stdin=fdstdin, stdout=subprocess.PIPE, |
1860 |
stderr=subprocess.STDOUT, close_fds=True,
|
1861 |
shell=False, cwd="/", env=env) |
1862 |
output = ""
|
1863 |
try:
|
1864 |
output = child.stdout.read(4096)
|
1865 |
child.stdout.close() |
1866 |
except EnvironmentError, err: |
1867 |
output += "Hook script error: %s" % str(err) |
1868 |
|
1869 |
while True: |
1870 |
try:
|
1871 |
result = child.wait() |
1872 |
break
|
1873 |
except EnvironmentError, err: |
1874 |
if err.errno == errno.EINTR:
|
1875 |
continue
|
1876 |
raise
|
1877 |
finally:
|
1878 |
# try not to leak fds
|
1879 |
for fd in (fdstdin, ): |
1880 |
if fd is not None: |
1881 |
try:
|
1882 |
fd.close() |
1883 |
except EnvironmentError, err: |
1884 |
# just log the error
|
1885 |
#logging.exception("Error while closing fd %s", fd)
|
1886 |
pass
|
1887 |
|
1888 |
return result == 0, output |
1889 |
|
1890 |
def RunHooks(self, hpath, phase, env): |
1891 |
"""Run the scripts in the hooks directory.
|
1892 |
|
1893 |
This method will not be usually overriden by child opcodes.
|
1894 |
|
1895 |
"""
|
1896 |
if phase == constants.HOOKS_PHASE_PRE:
|
1897 |
suffix = "pre"
|
1898 |
elif phase == constants.HOOKS_PHASE_POST:
|
1899 |
suffix = "post"
|
1900 |
else:
|
1901 |
raise errors.ProgrammerError("Unknown hooks phase: '%s'" % phase) |
1902 |
rr = [] |
1903 |
|
1904 |
subdir = "%s-%s.d" % (hpath, suffix)
|
1905 |
dir_name = "%s/%s" % (self._BASE_DIR, subdir) |
1906 |
try:
|
1907 |
dir_contents = utils.ListVisibleFiles(dir_name) |
1908 |
except OSError, err: |
1909 |
# must log
|
1910 |
return rr
|
1911 |
|
1912 |
# we use the standard python sort order,
|
1913 |
# so 00name is the recommended naming scheme
|
1914 |
dir_contents.sort() |
1915 |
for relname in dir_contents: |
1916 |
fname = os.path.join(dir_name, relname) |
1917 |
if not (os.path.isfile(fname) and os.access(fname, os.X_OK) and |
1918 |
self.RE_MASK.match(relname) is not None): |
1919 |
rrval = constants.HKR_SKIP |
1920 |
output = ""
|
1921 |
else:
|
1922 |
result, output = self.ExecHook(fname, env)
|
1923 |
if not result: |
1924 |
rrval = constants.HKR_FAIL |
1925 |
else:
|
1926 |
rrval = constants.HKR_SUCCESS |
1927 |
rr.append(("%s/%s" % (subdir, relname), rrval, output))
|
1928 |
|
1929 |
return rr
|
1930 |
|
1931 |
|
1932 |
class IAllocatorRunner(object): |
1933 |
"""IAllocator runner.
|
1934 |
|
1935 |
This class is instantiated on the node side (ganeti-noded) and not on
|
1936 |
the master side.
|
1937 |
|
1938 |
"""
|
1939 |
def Run(self, name, idata): |
1940 |
"""Run an iallocator script.
|
1941 |
|
1942 |
Return value: tuple of:
|
1943 |
- run status (one of the IARUN_ constants)
|
1944 |
- stdout
|
1945 |
- stderr
|
1946 |
- fail reason (as from utils.RunResult)
|
1947 |
|
1948 |
"""
|
1949 |
alloc_script = utils.FindFile(name, constants.IALLOCATOR_SEARCH_PATH, |
1950 |
os.path.isfile) |
1951 |
if alloc_script is None: |
1952 |
return (constants.IARUN_NOTFOUND, None, None, None) |
1953 |
|
1954 |
fd, fin_name = tempfile.mkstemp(prefix="ganeti-iallocator.")
|
1955 |
try:
|
1956 |
os.write(fd, idata) |
1957 |
os.close(fd) |
1958 |
result = utils.RunCmd([alloc_script, fin_name]) |
1959 |
if result.failed:
|
1960 |
return (constants.IARUN_FAILURE, result.stdout, result.stderr,
|
1961 |
result.fail_reason) |
1962 |
finally:
|
1963 |
os.unlink(fin_name) |
1964 |
|
1965 |
return (constants.IARUN_SUCCESS, result.stdout, result.stderr, None) |
1966 |
|
1967 |
|
1968 |
class DevCacheManager(object): |
1969 |
"""Simple class for managing a cache of block device information.
|
1970 |
|
1971 |
"""
|
1972 |
_DEV_PREFIX = "/dev/"
|
1973 |
_ROOT_DIR = constants.BDEV_CACHE_DIR |
1974 |
|
1975 |
@classmethod
|
1976 |
def _ConvertPath(cls, dev_path): |
1977 |
"""Converts a /dev/name path to the cache file name.
|
1978 |
|
1979 |
This replaces slashes with underscores and strips the /dev
|
1980 |
prefix. It then returns the full path to the cache file
|
1981 |
|
1982 |
"""
|
1983 |
if dev_path.startswith(cls._DEV_PREFIX):
|
1984 |
dev_path = dev_path[len(cls._DEV_PREFIX):]
|
1985 |
dev_path = dev_path.replace("/", "_") |
1986 |
fpath = "%s/bdev_%s" % (cls._ROOT_DIR, dev_path)
|
1987 |
return fpath
|
1988 |
|
1989 |
@classmethod
|
1990 |
def UpdateCache(cls, dev_path, owner, on_primary, iv_name): |
1991 |
"""Updates the cache information for a given device.
|
1992 |
|
1993 |
"""
|
1994 |
if dev_path is None: |
1995 |
logging.error("DevCacheManager.UpdateCache got a None dev_path")
|
1996 |
return
|
1997 |
fpath = cls._ConvertPath(dev_path) |
1998 |
if on_primary:
|
1999 |
state = "primary"
|
2000 |
else:
|
2001 |
state = "secondary"
|
2002 |
if iv_name is None: |
2003 |
iv_name = "not_visible"
|
2004 |
fdata = "%s %s %s\n" % (str(owner), state, iv_name) |
2005 |
try:
|
2006 |
utils.WriteFile(fpath, data=fdata) |
2007 |
except EnvironmentError, err: |
2008 |
logging.exception("Can't update bdev cache for %s", dev_path)
|
2009 |
|
2010 |
@classmethod
|
2011 |
def RemoveCache(cls, dev_path): |
2012 |
"""Remove data for a dev_path.
|
2013 |
|
2014 |
"""
|
2015 |
if dev_path is None: |
2016 |
logging.error("DevCacheManager.RemoveCache got a None dev_path")
|
2017 |
return
|
2018 |
fpath = cls._ConvertPath(dev_path) |
2019 |
try:
|
2020 |
utils.RemoveFile(fpath) |
2021 |
except EnvironmentError, err: |
2022 |
logging.exception("Can't update bdev cache for %s", dev_path)
|