root / lib / cmdlib / instance_operation.py @ 8ac806e6
History | View | Annotate | Download (15.8 kB)
1 |
#
|
---|---|
2 |
#
|
3 |
|
4 |
# Copyright (C) 2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013 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 |
"""Logical units dealing with instance operations (start/stop/...).
|
23 |
|
24 |
Those operations have in common that they affect the operating system in a
|
25 |
running instance directly.
|
26 |
|
27 |
"""
|
28 |
|
29 |
import logging |
30 |
|
31 |
from ganeti import constants |
32 |
from ganeti import errors |
33 |
from ganeti import hypervisor |
34 |
from ganeti import locking |
35 |
from ganeti import objects |
36 |
from ganeti import utils |
37 |
from ganeti.cmdlib.base import LogicalUnit, NoHooksLU |
38 |
from ganeti.cmdlib.common import INSTANCE_ONLINE, INSTANCE_DOWN, \ |
39 |
CheckHVParams, CheckInstanceState, CheckNodeOnline, ExpandNodeName, \ |
40 |
GetUpdatedParams, CheckOSParams, ShareAll |
41 |
from ganeti.cmdlib.instance_storage import StartInstanceDisks, \ |
42 |
ShutdownInstanceDisks
|
43 |
from ganeti.cmdlib.instance_utils import BuildInstanceHookEnvByObject, \ |
44 |
CheckInstanceBridgesExist, CheckNodeFreeMemory, CheckNodeHasOS |
45 |
|
46 |
|
47 |
class LUInstanceStartup(LogicalUnit): |
48 |
"""Starts an instance.
|
49 |
|
50 |
"""
|
51 |
HPATH = "instance-start"
|
52 |
HTYPE = constants.HTYPE_INSTANCE |
53 |
REQ_BGL = False
|
54 |
|
55 |
def CheckArguments(self): |
56 |
# extra beparams
|
57 |
if self.op.beparams: |
58 |
# fill the beparams dict
|
59 |
objects.UpgradeBeParams(self.op.beparams)
|
60 |
utils.ForceDictType(self.op.beparams, constants.BES_PARAMETER_TYPES)
|
61 |
|
62 |
def ExpandNames(self): |
63 |
self._ExpandAndLockInstance()
|
64 |
self.recalculate_locks[locking.LEVEL_NODE_RES] = constants.LOCKS_REPLACE
|
65 |
|
66 |
def DeclareLocks(self, level): |
67 |
if level == locking.LEVEL_NODE_RES:
|
68 |
self._LockInstancesNodes(primary_only=True, level=locking.LEVEL_NODE_RES) |
69 |
|
70 |
def BuildHooksEnv(self): |
71 |
"""Build hooks env.
|
72 |
|
73 |
This runs on master, primary and secondary nodes of the instance.
|
74 |
|
75 |
"""
|
76 |
env = { |
77 |
"FORCE": self.op.force, |
78 |
} |
79 |
|
80 |
env.update(BuildInstanceHookEnvByObject(self, self.instance)) |
81 |
|
82 |
return env
|
83 |
|
84 |
def BuildHooksNodes(self): |
85 |
"""Build hooks nodes.
|
86 |
|
87 |
"""
|
88 |
nl = [self.cfg.GetMasterNode()] + list(self.instance.all_nodes) |
89 |
return (nl, nl)
|
90 |
|
91 |
def CheckPrereq(self): |
92 |
"""Check prerequisites.
|
93 |
|
94 |
This checks that the instance is in the cluster.
|
95 |
|
96 |
"""
|
97 |
self.instance = instance = self.cfg.GetInstanceInfo(self.op.instance_name) |
98 |
assert self.instance is not None, \ |
99 |
"Cannot retrieve locked instance %s" % self.op.instance_name |
100 |
|
101 |
# extra hvparams
|
102 |
if self.op.hvparams: |
103 |
# check hypervisor parameter syntax (locally)
|
104 |
cluster = self.cfg.GetClusterInfo()
|
105 |
utils.ForceDictType(self.op.hvparams, constants.HVS_PARAMETER_TYPES)
|
106 |
filled_hvp = cluster.FillHV(instance) |
107 |
filled_hvp.update(self.op.hvparams)
|
108 |
hv_type = hypervisor.GetHypervisorClass(instance.hypervisor) |
109 |
hv_type.CheckParameterSyntax(filled_hvp) |
110 |
CheckHVParams(self, instance.all_nodes, instance.hypervisor, filled_hvp)
|
111 |
|
112 |
CheckInstanceState(self, instance, INSTANCE_ONLINE)
|
113 |
|
114 |
self.primary_offline = self.cfg.GetNodeInfo(instance.primary_node).offline |
115 |
|
116 |
if self.primary_offline and self.op.ignore_offline_nodes: |
117 |
self.LogWarning("Ignoring offline primary node") |
118 |
|
119 |
if self.op.hvparams or self.op.beparams: |
120 |
self.LogWarning("Overridden parameters are ignored") |
121 |
else:
|
122 |
CheckNodeOnline(self, instance.primary_node)
|
123 |
|
124 |
bep = self.cfg.GetClusterInfo().FillBE(instance)
|
125 |
bep.update(self.op.beparams)
|
126 |
|
127 |
# check bridges existence
|
128 |
CheckInstanceBridgesExist(self, instance)
|
129 |
|
130 |
remote_info = self.rpc.call_instance_info(instance.primary_node,
|
131 |
instance.name, |
132 |
instance.hypervisor) |
133 |
remote_info.Raise("Error checking node %s" % instance.primary_node,
|
134 |
prereq=True, ecode=errors.ECODE_ENVIRON)
|
135 |
if not remote_info.payload: # not running already |
136 |
CheckNodeFreeMemory(self, instance.primary_node,
|
137 |
"starting instance %s" % instance.name,
|
138 |
bep[constants.BE_MINMEM], instance.hypervisor) |
139 |
|
140 |
def Exec(self, feedback_fn): |
141 |
"""Start the instance.
|
142 |
|
143 |
"""
|
144 |
instance = self.instance
|
145 |
force = self.op.force
|
146 |
reason = self.op.reason
|
147 |
|
148 |
if not self.op.no_remember: |
149 |
self.cfg.MarkInstanceUp(instance.name)
|
150 |
|
151 |
if self.primary_offline: |
152 |
assert self.op.ignore_offline_nodes |
153 |
self.LogInfo("Primary node offline, marked instance as started") |
154 |
else:
|
155 |
node_current = instance.primary_node |
156 |
|
157 |
StartInstanceDisks(self, instance, force)
|
158 |
|
159 |
result = \ |
160 |
self.rpc.call_instance_start(node_current,
|
161 |
(instance, self.op.hvparams,
|
162 |
self.op.beparams),
|
163 |
self.op.startup_paused, reason)
|
164 |
msg = result.fail_msg |
165 |
if msg:
|
166 |
ShutdownInstanceDisks(self, instance)
|
167 |
raise errors.OpExecError("Could not start instance: %s" % msg) |
168 |
|
169 |
|
170 |
class LUInstanceShutdown(LogicalUnit): |
171 |
"""Shutdown an instance.
|
172 |
|
173 |
"""
|
174 |
HPATH = "instance-stop"
|
175 |
HTYPE = constants.HTYPE_INSTANCE |
176 |
REQ_BGL = False
|
177 |
|
178 |
def ExpandNames(self): |
179 |
self._ExpandAndLockInstance()
|
180 |
|
181 |
def BuildHooksEnv(self): |
182 |
"""Build hooks env.
|
183 |
|
184 |
This runs on master, primary and secondary nodes of the instance.
|
185 |
|
186 |
"""
|
187 |
env = BuildInstanceHookEnvByObject(self, self.instance) |
188 |
env["TIMEOUT"] = self.op.timeout |
189 |
return env
|
190 |
|
191 |
def BuildHooksNodes(self): |
192 |
"""Build hooks nodes.
|
193 |
|
194 |
"""
|
195 |
nl = [self.cfg.GetMasterNode()] + list(self.instance.all_nodes) |
196 |
return (nl, nl)
|
197 |
|
198 |
def CheckPrereq(self): |
199 |
"""Check prerequisites.
|
200 |
|
201 |
This checks that the instance is in the cluster.
|
202 |
|
203 |
"""
|
204 |
self.instance = self.cfg.GetInstanceInfo(self.op.instance_name) |
205 |
assert self.instance is not None, \ |
206 |
"Cannot retrieve locked instance %s" % self.op.instance_name |
207 |
|
208 |
if not self.op.force: |
209 |
CheckInstanceState(self, self.instance, INSTANCE_ONLINE) |
210 |
else:
|
211 |
self.LogWarning("Ignoring offline instance check") |
212 |
|
213 |
self.primary_offline = \
|
214 |
self.cfg.GetNodeInfo(self.instance.primary_node).offline |
215 |
|
216 |
if self.primary_offline and self.op.ignore_offline_nodes: |
217 |
self.LogWarning("Ignoring offline primary node") |
218 |
else:
|
219 |
CheckNodeOnline(self, self.instance.primary_node) |
220 |
|
221 |
def Exec(self, feedback_fn): |
222 |
"""Shutdown the instance.
|
223 |
|
224 |
"""
|
225 |
instance = self.instance
|
226 |
node_current = instance.primary_node |
227 |
timeout = self.op.timeout
|
228 |
reason = self.op.reason
|
229 |
|
230 |
# If the instance is offline we shouldn't mark it as down, as that
|
231 |
# resets the offline flag.
|
232 |
if not self.op.no_remember and instance.admin_state in INSTANCE_ONLINE: |
233 |
self.cfg.MarkInstanceDown(instance.name)
|
234 |
|
235 |
if self.primary_offline: |
236 |
assert self.op.ignore_offline_nodes |
237 |
self.LogInfo("Primary node offline, marked instance as stopped") |
238 |
else:
|
239 |
result = self.rpc.call_instance_shutdown(node_current, instance, timeout,
|
240 |
reason) |
241 |
msg = result.fail_msg |
242 |
if msg:
|
243 |
self.LogWarning("Could not shutdown instance: %s", msg) |
244 |
|
245 |
ShutdownInstanceDisks(self, instance)
|
246 |
|
247 |
|
248 |
class LUInstanceReinstall(LogicalUnit): |
249 |
"""Reinstall an instance.
|
250 |
|
251 |
"""
|
252 |
HPATH = "instance-reinstall"
|
253 |
HTYPE = constants.HTYPE_INSTANCE |
254 |
REQ_BGL = False
|
255 |
|
256 |
def ExpandNames(self): |
257 |
self._ExpandAndLockInstance()
|
258 |
|
259 |
def BuildHooksEnv(self): |
260 |
"""Build hooks env.
|
261 |
|
262 |
This runs on master, primary and secondary nodes of the instance.
|
263 |
|
264 |
"""
|
265 |
return BuildInstanceHookEnvByObject(self, self.instance) |
266 |
|
267 |
def BuildHooksNodes(self): |
268 |
"""Build hooks nodes.
|
269 |
|
270 |
"""
|
271 |
nl = [self.cfg.GetMasterNode()] + list(self.instance.all_nodes) |
272 |
return (nl, nl)
|
273 |
|
274 |
def CheckPrereq(self): |
275 |
"""Check prerequisites.
|
276 |
|
277 |
This checks that the instance is in the cluster and is not running.
|
278 |
|
279 |
"""
|
280 |
instance = self.cfg.GetInstanceInfo(self.op.instance_name) |
281 |
assert instance is not None, \ |
282 |
"Cannot retrieve locked instance %s" % self.op.instance_name |
283 |
CheckNodeOnline(self, instance.primary_node, "Instance primary node" |
284 |
" offline, cannot reinstall")
|
285 |
|
286 |
if instance.disk_template == constants.DT_DISKLESS:
|
287 |
raise errors.OpPrereqError("Instance '%s' has no disks" % |
288 |
self.op.instance_name,
|
289 |
errors.ECODE_INVAL) |
290 |
CheckInstanceState(self, instance, INSTANCE_DOWN, msg="cannot reinstall") |
291 |
|
292 |
if self.op.os_type is not None: |
293 |
# OS verification
|
294 |
pnode = ExpandNodeName(self.cfg, instance.primary_node)
|
295 |
CheckNodeHasOS(self, pnode, self.op.os_type, self.op.force_variant) |
296 |
instance_os = self.op.os_type
|
297 |
else:
|
298 |
instance_os = instance.os |
299 |
|
300 |
nodelist = list(instance.all_nodes)
|
301 |
|
302 |
if self.op.osparams: |
303 |
i_osdict = GetUpdatedParams(instance.osparams, self.op.osparams)
|
304 |
CheckOSParams(self, True, nodelist, instance_os, i_osdict) |
305 |
self.os_inst = i_osdict # the new dict (without defaults) |
306 |
else:
|
307 |
self.os_inst = None |
308 |
|
309 |
self.instance = instance
|
310 |
|
311 |
def Exec(self, feedback_fn): |
312 |
"""Reinstall the instance.
|
313 |
|
314 |
"""
|
315 |
inst = self.instance
|
316 |
|
317 |
if self.op.os_type is not None: |
318 |
feedback_fn("Changing OS to '%s'..." % self.op.os_type) |
319 |
inst.os = self.op.os_type
|
320 |
# Write to configuration
|
321 |
self.cfg.Update(inst, feedback_fn)
|
322 |
|
323 |
StartInstanceDisks(self, inst, None) |
324 |
try:
|
325 |
feedback_fn("Running the instance OS create scripts...")
|
326 |
# FIXME: pass debug option from opcode to backend
|
327 |
result = self.rpc.call_instance_os_add(inst.primary_node,
|
328 |
(inst, self.os_inst), True, |
329 |
self.op.debug_level)
|
330 |
result.Raise("Could not install OS for instance %s on node %s" %
|
331 |
(inst.name, inst.primary_node)) |
332 |
finally:
|
333 |
ShutdownInstanceDisks(self, inst)
|
334 |
|
335 |
|
336 |
class LUInstanceReboot(LogicalUnit): |
337 |
"""Reboot an instance.
|
338 |
|
339 |
"""
|
340 |
HPATH = "instance-reboot"
|
341 |
HTYPE = constants.HTYPE_INSTANCE |
342 |
REQ_BGL = False
|
343 |
|
344 |
def ExpandNames(self): |
345 |
self._ExpandAndLockInstance()
|
346 |
|
347 |
def BuildHooksEnv(self): |
348 |
"""Build hooks env.
|
349 |
|
350 |
This runs on master, primary and secondary nodes of the instance.
|
351 |
|
352 |
"""
|
353 |
env = { |
354 |
"IGNORE_SECONDARIES": self.op.ignore_secondaries, |
355 |
"REBOOT_TYPE": self.op.reboot_type, |
356 |
"SHUTDOWN_TIMEOUT": self.op.shutdown_timeout, |
357 |
} |
358 |
|
359 |
env.update(BuildInstanceHookEnvByObject(self, self.instance)) |
360 |
|
361 |
return env
|
362 |
|
363 |
def BuildHooksNodes(self): |
364 |
"""Build hooks nodes.
|
365 |
|
366 |
"""
|
367 |
nl = [self.cfg.GetMasterNode()] + list(self.instance.all_nodes) |
368 |
return (nl, nl)
|
369 |
|
370 |
def CheckPrereq(self): |
371 |
"""Check prerequisites.
|
372 |
|
373 |
This checks that the instance is in the cluster.
|
374 |
|
375 |
"""
|
376 |
self.instance = instance = self.cfg.GetInstanceInfo(self.op.instance_name) |
377 |
assert self.instance is not None, \ |
378 |
"Cannot retrieve locked instance %s" % self.op.instance_name |
379 |
CheckInstanceState(self, instance, INSTANCE_ONLINE)
|
380 |
CheckNodeOnline(self, instance.primary_node)
|
381 |
|
382 |
# check bridges existence
|
383 |
CheckInstanceBridgesExist(self, instance)
|
384 |
|
385 |
def Exec(self, feedback_fn): |
386 |
"""Reboot the instance.
|
387 |
|
388 |
"""
|
389 |
instance = self.instance
|
390 |
ignore_secondaries = self.op.ignore_secondaries
|
391 |
reboot_type = self.op.reboot_type
|
392 |
reason = self.op.reason
|
393 |
|
394 |
remote_info = self.rpc.call_instance_info(instance.primary_node,
|
395 |
instance.name, |
396 |
instance.hypervisor) |
397 |
remote_info.Raise("Error checking node %s" % instance.primary_node)
|
398 |
instance_running = bool(remote_info.payload)
|
399 |
|
400 |
node_current = instance.primary_node |
401 |
|
402 |
if instance_running and reboot_type in [constants.INSTANCE_REBOOT_SOFT, |
403 |
constants.INSTANCE_REBOOT_HARD]: |
404 |
for disk in instance.disks: |
405 |
self.cfg.SetDiskID(disk, node_current)
|
406 |
result = self.rpc.call_instance_reboot(node_current, instance,
|
407 |
reboot_type, |
408 |
self.op.shutdown_timeout, reason)
|
409 |
result.Raise("Could not reboot instance")
|
410 |
else:
|
411 |
if instance_running:
|
412 |
result = self.rpc.call_instance_shutdown(node_current, instance,
|
413 |
self.op.shutdown_timeout,
|
414 |
reason) |
415 |
result.Raise("Could not shutdown instance for full reboot")
|
416 |
ShutdownInstanceDisks(self, instance)
|
417 |
else:
|
418 |
self.LogInfo("Instance %s was already stopped, starting now", |
419 |
instance.name) |
420 |
StartInstanceDisks(self, instance, ignore_secondaries)
|
421 |
result = self.rpc.call_instance_start(node_current,
|
422 |
(instance, None, None), False, |
423 |
reason) |
424 |
msg = result.fail_msg |
425 |
if msg:
|
426 |
ShutdownInstanceDisks(self, instance)
|
427 |
raise errors.OpExecError("Could not start instance for" |
428 |
" full reboot: %s" % msg)
|
429 |
|
430 |
self.cfg.MarkInstanceUp(instance.name)
|
431 |
|
432 |
|
433 |
def GetInstanceConsole(cluster, instance): |
434 |
"""Returns console information for an instance.
|
435 |
|
436 |
@type cluster: L{objects.Cluster}
|
437 |
@type instance: L{objects.Instance}
|
438 |
@rtype: dict
|
439 |
|
440 |
"""
|
441 |
hyper = hypervisor.GetHypervisorClass(instance.hypervisor) |
442 |
# beparams and hvparams are passed separately, to avoid editing the
|
443 |
# instance and then saving the defaults in the instance itself.
|
444 |
hvparams = cluster.FillHV(instance) |
445 |
beparams = cluster.FillBE(instance) |
446 |
console = hyper.GetInstanceConsole(instance, hvparams, beparams) |
447 |
|
448 |
assert console.instance == instance.name
|
449 |
assert console.Validate()
|
450 |
|
451 |
return console.ToDict()
|
452 |
|
453 |
|
454 |
class LUInstanceConsole(NoHooksLU): |
455 |
"""Connect to an instance's console.
|
456 |
|
457 |
This is somewhat special in that it returns the command line that
|
458 |
you need to run on the master node in order to connect to the
|
459 |
console.
|
460 |
|
461 |
"""
|
462 |
REQ_BGL = False
|
463 |
|
464 |
def ExpandNames(self): |
465 |
self.share_locks = ShareAll()
|
466 |
self._ExpandAndLockInstance()
|
467 |
|
468 |
def CheckPrereq(self): |
469 |
"""Check prerequisites.
|
470 |
|
471 |
This checks that the instance is in the cluster.
|
472 |
|
473 |
"""
|
474 |
self.instance = self.cfg.GetInstanceInfo(self.op.instance_name) |
475 |
assert self.instance is not None, \ |
476 |
"Cannot retrieve locked instance %s" % self.op.instance_name |
477 |
CheckNodeOnline(self, self.instance.primary_node) |
478 |
|
479 |
def Exec(self, feedback_fn): |
480 |
"""Connect to the console of an instance
|
481 |
|
482 |
"""
|
483 |
instance = self.instance
|
484 |
node = instance.primary_node |
485 |
|
486 |
cluster_hvparams = self.cfg.GetClusterInfo().hvparams
|
487 |
node_insts = self.rpc.call_instance_list([node],
|
488 |
[instance.hypervisor], |
489 |
cluster_hvparams)[node] |
490 |
node_insts.Raise("Can't get node information from %s" % node)
|
491 |
|
492 |
if instance.name not in node_insts.payload: |
493 |
if instance.admin_state == constants.ADMINST_UP:
|
494 |
state = constants.INSTST_ERRORDOWN |
495 |
elif instance.admin_state == constants.ADMINST_DOWN:
|
496 |
state = constants.INSTST_ADMINDOWN |
497 |
else:
|
498 |
state = constants.INSTST_ADMINOFFLINE |
499 |
raise errors.OpExecError("Instance %s is not running (state %s)" % |
500 |
(instance.name, state)) |
501 |
|
502 |
logging.debug("Connecting to console of %s on %s", instance.name, node)
|
503 |
|
504 |
return GetInstanceConsole(self.cfg.GetClusterInfo(), instance) |