root / lib / cmdlib / instance_migration.py @ 18397489
History | View | Annotate | Download (37.8 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 migration an failover."""
|
23 |
|
24 |
import logging |
25 |
import time |
26 |
|
27 |
from ganeti import constants |
28 |
from ganeti import errors |
29 |
from ganeti import locking |
30 |
from ganeti.masterd import iallocator |
31 |
from ganeti import utils |
32 |
from ganeti.cmdlib.base import LogicalUnit, Tasklet |
33 |
from ganeti.cmdlib.common import ExpandInstanceUuidAndName, \ |
34 |
CheckIAllocatorOrNode, ExpandNodeUuidAndName |
35 |
from ganeti.cmdlib.instance_storage import CheckDiskConsistency, \ |
36 |
ExpandCheckDisks, ShutdownInstanceDisks, AssembleInstanceDisks |
37 |
from ganeti.cmdlib.instance_utils import BuildInstanceHookEnvByObject, \ |
38 |
CheckTargetNodeIPolicy, ReleaseLocks, CheckNodeNotDrained, \ |
39 |
CopyLockList, CheckNodeFreeMemory, CheckInstanceBridgesExist |
40 |
|
41 |
import ganeti.masterd.instance |
42 |
|
43 |
|
44 |
def _ExpandNamesForMigration(lu): |
45 |
"""Expands names for use with L{TLMigrateInstance}.
|
46 |
|
47 |
@type lu: L{LogicalUnit}
|
48 |
|
49 |
"""
|
50 |
if lu.op.target_node is not None: |
51 |
(lu.op.target_node_uuid, lu.op.target_node) = \ |
52 |
ExpandNodeUuidAndName(lu.cfg, lu.op.target_node_uuid, lu.op.target_node) |
53 |
|
54 |
lu.needed_locks[locking.LEVEL_NODE] = [] |
55 |
lu.recalculate_locks[locking.LEVEL_NODE] = constants.LOCKS_REPLACE |
56 |
|
57 |
lu.needed_locks[locking.LEVEL_NODE_RES] = [] |
58 |
lu.recalculate_locks[locking.LEVEL_NODE_RES] = constants.LOCKS_REPLACE |
59 |
|
60 |
# The node allocation lock is actually only needed for externally replicated
|
61 |
# instances (e.g. sharedfile or RBD) and if an iallocator is used.
|
62 |
lu.needed_locks[locking.LEVEL_NODE_ALLOC] = [] |
63 |
|
64 |
|
65 |
def _DeclareLocksForMigration(lu, level): |
66 |
"""Declares locks for L{TLMigrateInstance}.
|
67 |
|
68 |
@type lu: L{LogicalUnit}
|
69 |
@param level: Lock level
|
70 |
|
71 |
"""
|
72 |
if level == locking.LEVEL_NODE_ALLOC:
|
73 |
assert lu.op.instance_name in lu.owned_locks(locking.LEVEL_INSTANCE) |
74 |
|
75 |
instance = lu.cfg.GetInstanceInfo(lu.op.instance_uuid) |
76 |
|
77 |
# Node locks are already declared here rather than at LEVEL_NODE as we need
|
78 |
# the instance object anyway to declare the node allocation lock.
|
79 |
if instance.disk_template in constants.DTS_EXT_MIRROR: |
80 |
if lu.op.target_node is None: |
81 |
lu.needed_locks[locking.LEVEL_NODE] = locking.ALL_SET |
82 |
lu.needed_locks[locking.LEVEL_NODE_ALLOC] = locking.ALL_SET |
83 |
else:
|
84 |
lu.needed_locks[locking.LEVEL_NODE] = [instance.primary_node, |
85 |
lu.op.target_node_uuid] |
86 |
del lu.recalculate_locks[locking.LEVEL_NODE]
|
87 |
else:
|
88 |
lu._LockInstancesNodes() # pylint: disable=W0212
|
89 |
|
90 |
elif level == locking.LEVEL_NODE:
|
91 |
# Node locks are declared together with the node allocation lock
|
92 |
assert (lu.needed_locks[locking.LEVEL_NODE] or |
93 |
lu.needed_locks[locking.LEVEL_NODE] is locking.ALL_SET)
|
94 |
|
95 |
elif level == locking.LEVEL_NODE_RES:
|
96 |
# Copy node locks
|
97 |
lu.needed_locks[locking.LEVEL_NODE_RES] = \ |
98 |
CopyLockList(lu.needed_locks[locking.LEVEL_NODE]) |
99 |
|
100 |
|
101 |
class LUInstanceFailover(LogicalUnit): |
102 |
"""Failover an instance.
|
103 |
|
104 |
"""
|
105 |
HPATH = "instance-failover"
|
106 |
HTYPE = constants.HTYPE_INSTANCE |
107 |
REQ_BGL = False
|
108 |
|
109 |
def CheckArguments(self): |
110 |
"""Check the arguments.
|
111 |
|
112 |
"""
|
113 |
self.iallocator = getattr(self.op, "iallocator", None) |
114 |
self.target_node = getattr(self.op, "target_node", None) |
115 |
|
116 |
def ExpandNames(self): |
117 |
self._ExpandAndLockInstance()
|
118 |
_ExpandNamesForMigration(self)
|
119 |
|
120 |
self._migrater = \
|
121 |
TLMigrateInstance(self, self.op.instance_uuid, self.op.instance_name, |
122 |
self.op.cleanup, True, False, |
123 |
self.op.ignore_consistency, True, |
124 |
self.op.shutdown_timeout, self.op.ignore_ipolicy) |
125 |
|
126 |
self.tasklets = [self._migrater] |
127 |
|
128 |
def DeclareLocks(self, level): |
129 |
_DeclareLocksForMigration(self, level)
|
130 |
|
131 |
def BuildHooksEnv(self): |
132 |
"""Build hooks env.
|
133 |
|
134 |
This runs on master, primary and secondary nodes of the instance.
|
135 |
|
136 |
"""
|
137 |
instance = self._migrater.instance
|
138 |
source_node_uuid = instance.primary_node |
139 |
env = { |
140 |
"IGNORE_CONSISTENCY": self.op.ignore_consistency, |
141 |
"SHUTDOWN_TIMEOUT": self.op.shutdown_timeout, |
142 |
"OLD_PRIMARY": self.cfg.GetNodeName(source_node_uuid), |
143 |
"NEW_PRIMARY": self.op.target_node, |
144 |
"FAILOVER_CLEANUP": self.op.cleanup, |
145 |
} |
146 |
|
147 |
if instance.disk_template in constants.DTS_INT_MIRROR: |
148 |
env["OLD_SECONDARY"] = self.cfg.GetNodeName(instance.secondary_nodes[0]) |
149 |
env["NEW_SECONDARY"] = self.cfg.GetNodeName(source_node_uuid) |
150 |
else:
|
151 |
env["OLD_SECONDARY"] = env["NEW_SECONDARY"] = "" |
152 |
|
153 |
env.update(BuildInstanceHookEnvByObject(self, instance))
|
154 |
|
155 |
return env
|
156 |
|
157 |
def BuildHooksNodes(self): |
158 |
"""Build hooks nodes.
|
159 |
|
160 |
"""
|
161 |
instance = self._migrater.instance
|
162 |
nl = [self.cfg.GetMasterNode()] + list(instance.secondary_nodes) |
163 |
return (nl, nl + [instance.primary_node])
|
164 |
|
165 |
|
166 |
class LUInstanceMigrate(LogicalUnit): |
167 |
"""Migrate an instance.
|
168 |
|
169 |
This is migration without shutting down, compared to the failover,
|
170 |
which is done with shutdown.
|
171 |
|
172 |
"""
|
173 |
HPATH = "instance-migrate"
|
174 |
HTYPE = constants.HTYPE_INSTANCE |
175 |
REQ_BGL = False
|
176 |
|
177 |
def ExpandNames(self): |
178 |
self._ExpandAndLockInstance()
|
179 |
_ExpandNamesForMigration(self)
|
180 |
|
181 |
self._migrater = \
|
182 |
TLMigrateInstance(self, self.op.instance_uuid, self.op.instance_name, |
183 |
self.op.cleanup, False, self.op.allow_failover, False, |
184 |
self.op.allow_runtime_changes,
|
185 |
constants.DEFAULT_SHUTDOWN_TIMEOUT, |
186 |
self.op.ignore_ipolicy)
|
187 |
|
188 |
self.tasklets = [self._migrater] |
189 |
|
190 |
def DeclareLocks(self, level): |
191 |
_DeclareLocksForMigration(self, level)
|
192 |
|
193 |
def BuildHooksEnv(self): |
194 |
"""Build hooks env.
|
195 |
|
196 |
This runs on master, primary and secondary nodes of the instance.
|
197 |
|
198 |
"""
|
199 |
instance = self._migrater.instance
|
200 |
source_node_uuid = instance.primary_node |
201 |
env = BuildInstanceHookEnvByObject(self, instance)
|
202 |
env.update({ |
203 |
"MIGRATE_LIVE": self._migrater.live, |
204 |
"MIGRATE_CLEANUP": self.op.cleanup, |
205 |
"OLD_PRIMARY": self.cfg.GetNodeName(source_node_uuid), |
206 |
"NEW_PRIMARY": self.op.target_node, |
207 |
"ALLOW_RUNTIME_CHANGES": self.op.allow_runtime_changes, |
208 |
}) |
209 |
|
210 |
if instance.disk_template in constants.DTS_INT_MIRROR: |
211 |
env["OLD_SECONDARY"] = self.cfg.GetNodeName(instance.secondary_nodes[0]) |
212 |
env["NEW_SECONDARY"] = self.cfg.GetNodeName(source_node_uuid) |
213 |
else:
|
214 |
env["OLD_SECONDARY"] = env["NEW_SECONDARY"] = None |
215 |
|
216 |
return env
|
217 |
|
218 |
def BuildHooksNodes(self): |
219 |
"""Build hooks nodes.
|
220 |
|
221 |
"""
|
222 |
instance = self._migrater.instance
|
223 |
snode_uuids = list(instance.secondary_nodes)
|
224 |
nl = [self.cfg.GetMasterNode(), instance.primary_node] + snode_uuids
|
225 |
return (nl, nl)
|
226 |
|
227 |
|
228 |
class TLMigrateInstance(Tasklet): |
229 |
"""Tasklet class for instance migration.
|
230 |
|
231 |
@type live: boolean
|
232 |
@ivar live: whether the migration will be done live or non-live;
|
233 |
this variable is initalized only after CheckPrereq has run
|
234 |
@type cleanup: boolean
|
235 |
@ivar cleanup: Wheater we cleanup from a failed migration
|
236 |
@type iallocator: string
|
237 |
@ivar iallocator: The iallocator used to determine target_node
|
238 |
@type target_node_uuid: string
|
239 |
@ivar target_node_uuid: If given, the target node UUID to reallocate the
|
240 |
instance to
|
241 |
@type failover: boolean
|
242 |
@ivar failover: Whether operation results in failover or migration
|
243 |
@type fallback: boolean
|
244 |
@ivar fallback: Whether fallback to failover is allowed if migration not
|
245 |
possible
|
246 |
@type ignore_consistency: boolean
|
247 |
@ivar ignore_consistency: Wheter we should ignore consistency between source
|
248 |
and target node
|
249 |
@type shutdown_timeout: int
|
250 |
@ivar shutdown_timeout: In case of failover timeout of the shutdown
|
251 |
@type ignore_ipolicy: bool
|
252 |
@ivar ignore_ipolicy: If true, we can ignore instance policy when migrating
|
253 |
|
254 |
"""
|
255 |
|
256 |
# Constants
|
257 |
_MIGRATION_POLL_INTERVAL = 1 # seconds |
258 |
_MIGRATION_FEEDBACK_INTERVAL = 10 # seconds |
259 |
|
260 |
def __init__(self, lu, instance_uuid, instance_name, cleanup, failover, |
261 |
fallback, ignore_consistency, allow_runtime_changes, |
262 |
shutdown_timeout, ignore_ipolicy): |
263 |
"""Initializes this class.
|
264 |
|
265 |
"""
|
266 |
Tasklet.__init__(self, lu)
|
267 |
|
268 |
# Parameters
|
269 |
self.instance_uuid = instance_uuid
|
270 |
self.instance_name = instance_name
|
271 |
self.cleanup = cleanup
|
272 |
self.live = False # will be overridden later |
273 |
self.failover = failover
|
274 |
self.fallback = fallback
|
275 |
self.ignore_consistency = ignore_consistency
|
276 |
self.shutdown_timeout = shutdown_timeout
|
277 |
self.ignore_ipolicy = ignore_ipolicy
|
278 |
self.allow_runtime_changes = allow_runtime_changes
|
279 |
|
280 |
def CheckPrereq(self): |
281 |
"""Check prerequisites.
|
282 |
|
283 |
This checks that the instance is in the cluster.
|
284 |
|
285 |
"""
|
286 |
(self.instance_uuid, self.instance_name) = \ |
287 |
ExpandInstanceUuidAndName(self.lu.cfg, self.instance_uuid, |
288 |
self.instance_name)
|
289 |
self.instance = self.cfg.GetInstanceInfo(self.instance_uuid) |
290 |
assert self.instance is not None |
291 |
cluster = self.cfg.GetClusterInfo()
|
292 |
|
293 |
if (not self.cleanup and |
294 |
not self.instance.admin_state == constants.ADMINST_UP and |
295 |
not self.failover and self.fallback): |
296 |
self.lu.LogInfo("Instance is marked down or offline, fallback allowed," |
297 |
" switching to failover")
|
298 |
self.failover = True |
299 |
|
300 |
if self.instance.disk_template not in constants.DTS_MIRRORED: |
301 |
if self.failover: |
302 |
text = "failovers"
|
303 |
else:
|
304 |
text = "migrations"
|
305 |
raise errors.OpPrereqError("Instance's disk layout '%s' does not allow" |
306 |
" %s" % (self.instance.disk_template, text), |
307 |
errors.ECODE_STATE) |
308 |
|
309 |
if self.instance.disk_template in constants.DTS_EXT_MIRROR: |
310 |
CheckIAllocatorOrNode(self.lu, "iallocator", "target_node") |
311 |
|
312 |
if self.lu.op.iallocator: |
313 |
assert locking.NAL in self.lu.owned_locks(locking.LEVEL_NODE_ALLOC) |
314 |
self._RunAllocator()
|
315 |
else:
|
316 |
# We set set self.target_node_uuid as it is required by
|
317 |
# BuildHooksEnv
|
318 |
self.target_node_uuid = self.lu.op.target_node_uuid |
319 |
|
320 |
# Check that the target node is correct in terms of instance policy
|
321 |
nodeinfo = self.cfg.GetNodeInfo(self.target_node_uuid) |
322 |
group_info = self.cfg.GetNodeGroup(nodeinfo.group)
|
323 |
ipolicy = ganeti.masterd.instance.CalculateGroupIPolicy(cluster, |
324 |
group_info) |
325 |
CheckTargetNodeIPolicy(self.lu, ipolicy, self.instance, nodeinfo, |
326 |
self.cfg, ignore=self.ignore_ipolicy) |
327 |
|
328 |
# self.target_node is already populated, either directly or by the
|
329 |
# iallocator run
|
330 |
target_node_uuid = self.target_node_uuid
|
331 |
if self.target_node_uuid == self.instance.primary_node: |
332 |
raise errors.OpPrereqError(
|
333 |
"Cannot migrate instance %s to its primary (%s)" %
|
334 |
(self.instance.name,
|
335 |
self.cfg.GetNodeName(self.instance.primary_node)), |
336 |
errors.ECODE_STATE) |
337 |
|
338 |
if len(self.lu.tasklets) == 1: |
339 |
# It is safe to release locks only when we're the only tasklet
|
340 |
# in the LU
|
341 |
ReleaseLocks(self.lu, locking.LEVEL_NODE,
|
342 |
keep=[self.instance.primary_node, self.target_node_uuid]) |
343 |
ReleaseLocks(self.lu, locking.LEVEL_NODE_ALLOC)
|
344 |
|
345 |
else:
|
346 |
assert not self.lu.glm.is_owned(locking.LEVEL_NODE_ALLOC) |
347 |
|
348 |
secondary_node_uuids = self.instance.secondary_nodes
|
349 |
if not secondary_node_uuids: |
350 |
raise errors.ConfigurationError("No secondary node but using" |
351 |
" %s disk template" %
|
352 |
self.instance.disk_template)
|
353 |
target_node_uuid = secondary_node_uuids[0]
|
354 |
if self.lu.op.iallocator or \ |
355 |
(self.lu.op.target_node_uuid and |
356 |
self.lu.op.target_node_uuid != target_node_uuid):
|
357 |
if self.failover: |
358 |
text = "failed over"
|
359 |
else:
|
360 |
text = "migrated"
|
361 |
raise errors.OpPrereqError("Instances with disk template %s cannot" |
362 |
" be %s to arbitrary nodes"
|
363 |
" (neither an iallocator nor a target"
|
364 |
" node can be passed)" %
|
365 |
(self.instance.disk_template, text),
|
366 |
errors.ECODE_INVAL) |
367 |
nodeinfo = self.cfg.GetNodeInfo(target_node_uuid)
|
368 |
group_info = self.cfg.GetNodeGroup(nodeinfo.group)
|
369 |
ipolicy = ganeti.masterd.instance.CalculateGroupIPolicy(cluster, |
370 |
group_info) |
371 |
CheckTargetNodeIPolicy(self.lu, ipolicy, self.instance, nodeinfo, |
372 |
self.cfg, ignore=self.ignore_ipolicy) |
373 |
|
374 |
i_be = cluster.FillBE(self.instance)
|
375 |
|
376 |
# check memory requirements on the secondary node
|
377 |
if (not self.cleanup and |
378 |
(not self.failover or |
379 |
self.instance.admin_state == constants.ADMINST_UP)):
|
380 |
self.tgt_free_mem = CheckNodeFreeMemory(
|
381 |
self.lu, target_node_uuid,
|
382 |
"migrating instance %s" % self.instance.name, |
383 |
i_be[constants.BE_MINMEM], self.instance.hypervisor,
|
384 |
self.cfg.GetClusterInfo().hvparams[self.instance.hypervisor]) |
385 |
else:
|
386 |
self.lu.LogInfo("Not checking memory on the secondary node as" |
387 |
" instance will not be started")
|
388 |
|
389 |
# check if failover must be forced instead of migration
|
390 |
if (not self.cleanup and not self.failover and |
391 |
i_be[constants.BE_ALWAYS_FAILOVER]): |
392 |
self.lu.LogInfo("Instance configured to always failover; fallback" |
393 |
" to failover")
|
394 |
self.failover = True |
395 |
|
396 |
# check bridge existance
|
397 |
CheckInstanceBridgesExist(self.lu, self.instance, |
398 |
node_uuid=target_node_uuid) |
399 |
|
400 |
if not self.cleanup: |
401 |
CheckNodeNotDrained(self.lu, target_node_uuid)
|
402 |
if not self.failover: |
403 |
result = self.rpc.call_instance_migratable(self.instance.primary_node, |
404 |
self.instance)
|
405 |
if result.fail_msg and self.fallback: |
406 |
self.lu.LogInfo("Can't migrate, instance offline, fallback to" |
407 |
" failover")
|
408 |
self.failover = True |
409 |
else:
|
410 |
result.Raise("Can't migrate, please use failover",
|
411 |
prereq=True, ecode=errors.ECODE_STATE)
|
412 |
|
413 |
assert not (self.failover and self.cleanup) |
414 |
|
415 |
if not self.failover: |
416 |
if self.lu.op.live is not None and self.lu.op.mode is not None: |
417 |
raise errors.OpPrereqError("Only one of the 'live' and 'mode'" |
418 |
" parameters are accepted",
|
419 |
errors.ECODE_INVAL) |
420 |
if self.lu.op.live is not None: |
421 |
if self.lu.op.live: |
422 |
self.lu.op.mode = constants.HT_MIGRATION_LIVE
|
423 |
else:
|
424 |
self.lu.op.mode = constants.HT_MIGRATION_NONLIVE
|
425 |
# reset the 'live' parameter to None so that repeated
|
426 |
# invocations of CheckPrereq do not raise an exception
|
427 |
self.lu.op.live = None |
428 |
elif self.lu.op.mode is None: |
429 |
# read the default value from the hypervisor
|
430 |
i_hv = cluster.FillHV(self.instance, skip_globals=False) |
431 |
self.lu.op.mode = i_hv[constants.HV_MIGRATION_MODE]
|
432 |
|
433 |
self.live = self.lu.op.mode == constants.HT_MIGRATION_LIVE |
434 |
else:
|
435 |
# Failover is never live
|
436 |
self.live = False |
437 |
|
438 |
if not (self.failover or self.cleanup): |
439 |
remote_info = self.rpc.call_instance_info(
|
440 |
self.instance.primary_node, self.instance.name, |
441 |
self.instance.hypervisor, cluster.hvparams[self.instance.hypervisor]) |
442 |
remote_info.Raise("Error checking instance on node %s" %
|
443 |
self.cfg.GetNodeName(self.instance.primary_node)) |
444 |
instance_running = bool(remote_info.payload)
|
445 |
if instance_running:
|
446 |
self.current_mem = int(remote_info.payload["memory"]) |
447 |
|
448 |
def _RunAllocator(self): |
449 |
"""Run the allocator based on input opcode.
|
450 |
|
451 |
"""
|
452 |
assert locking.NAL in self.lu.owned_locks(locking.LEVEL_NODE_ALLOC) |
453 |
|
454 |
# FIXME: add a self.ignore_ipolicy option
|
455 |
req = iallocator.IAReqRelocate( |
456 |
inst_uuid=self.instance_uuid,
|
457 |
relocate_from_node_uuids=[self.instance.primary_node])
|
458 |
ial = iallocator.IAllocator(self.cfg, self.rpc, req) |
459 |
|
460 |
ial.Run(self.lu.op.iallocator)
|
461 |
|
462 |
if not ial.success: |
463 |
raise errors.OpPrereqError("Can't compute nodes using" |
464 |
" iallocator '%s': %s" %
|
465 |
(self.lu.op.iallocator, ial.info),
|
466 |
errors.ECODE_NORES) |
467 |
self.target_node_uuid = self.cfg.GetNodeInfoByName(ial.result[0]).uuid |
468 |
self.lu.LogInfo("Selected nodes for instance %s via iallocator %s: %s", |
469 |
self.instance_name, self.lu.op.iallocator, |
470 |
utils.CommaJoin(ial.result)) |
471 |
|
472 |
def _WaitUntilSync(self): |
473 |
"""Poll with custom rpc for disk sync.
|
474 |
|
475 |
This uses our own step-based rpc call.
|
476 |
|
477 |
"""
|
478 |
self.feedback_fn("* wait until resync is done") |
479 |
all_done = False
|
480 |
while not all_done: |
481 |
all_done = True
|
482 |
result = self.rpc.call_drbd_wait_sync(self.all_node_uuids, |
483 |
self.nodes_ip,
|
484 |
(self.instance.disks,
|
485 |
self.instance))
|
486 |
min_percent = 100
|
487 |
for node_uuid, nres in result.items(): |
488 |
nres.Raise("Cannot resync disks on node %s" %
|
489 |
self.cfg.GetNodeName(node_uuid))
|
490 |
node_done, node_percent = nres.payload |
491 |
all_done = all_done and node_done
|
492 |
if node_percent is not None: |
493 |
min_percent = min(min_percent, node_percent)
|
494 |
if not all_done: |
495 |
if min_percent < 100: |
496 |
self.feedback_fn(" - progress: %.1f%%" % min_percent) |
497 |
time.sleep(2)
|
498 |
|
499 |
def _EnsureSecondary(self, node_uuid): |
500 |
"""Demote a node to secondary.
|
501 |
|
502 |
"""
|
503 |
self.feedback_fn("* switching node %s to secondary mode" % |
504 |
self.cfg.GetNodeName(node_uuid))
|
505 |
|
506 |
for dev in self.instance.disks: |
507 |
self.cfg.SetDiskID(dev, node_uuid)
|
508 |
|
509 |
result = self.rpc.call_blockdev_close(node_uuid, self.instance.name, |
510 |
self.instance.disks)
|
511 |
result.Raise("Cannot change disk to secondary on node %s" %
|
512 |
self.cfg.GetNodeName(node_uuid))
|
513 |
|
514 |
def _GoStandalone(self): |
515 |
"""Disconnect from the network.
|
516 |
|
517 |
"""
|
518 |
self.feedback_fn("* changing into standalone mode") |
519 |
result = self.rpc.call_drbd_disconnect_net(self.all_node_uuids, |
520 |
self.nodes_ip,
|
521 |
self.instance.disks)
|
522 |
for node_uuid, nres in result.items(): |
523 |
nres.Raise("Cannot disconnect disks node %s" %
|
524 |
self.cfg.GetNodeName(node_uuid))
|
525 |
|
526 |
def _GoReconnect(self, multimaster): |
527 |
"""Reconnect to the network.
|
528 |
|
529 |
"""
|
530 |
if multimaster:
|
531 |
msg = "dual-master"
|
532 |
else:
|
533 |
msg = "single-master"
|
534 |
self.feedback_fn("* changing disks into %s mode" % msg) |
535 |
result = self.rpc.call_drbd_attach_net(self.all_node_uuids, self.nodes_ip, |
536 |
(self.instance.disks, self.instance), |
537 |
self.instance.name, multimaster)
|
538 |
for node_uuid, nres in result.items(): |
539 |
nres.Raise("Cannot change disks config on node %s" %
|
540 |
self.cfg.GetNodeName(node_uuid))
|
541 |
|
542 |
def _ExecCleanup(self): |
543 |
"""Try to cleanup after a failed migration.
|
544 |
|
545 |
The cleanup is done by:
|
546 |
- check that the instance is running only on one node
|
547 |
(and update the config if needed)
|
548 |
- change disks on its secondary node to secondary
|
549 |
- wait until disks are fully synchronized
|
550 |
- disconnect from the network
|
551 |
- change disks into single-master mode
|
552 |
- wait again until disks are fully synchronized
|
553 |
|
554 |
"""
|
555 |
# check running on only one node
|
556 |
self.feedback_fn("* checking where the instance actually runs" |
557 |
" (if this hangs, the hypervisor might be in"
|
558 |
" a bad state)")
|
559 |
cluster_hvparams = self.cfg.GetClusterInfo().hvparams
|
560 |
ins_l = self.rpc.call_instance_list(self.all_node_uuids, |
561 |
[self.instance.hypervisor],
|
562 |
cluster_hvparams) |
563 |
for node_uuid, result in ins_l.items(): |
564 |
result.Raise("Can't contact node %s" % node_uuid)
|
565 |
|
566 |
runningon_source = self.instance.name in \ |
567 |
ins_l[self.source_node_uuid].payload
|
568 |
runningon_target = self.instance.name in \ |
569 |
ins_l[self.target_node_uuid].payload
|
570 |
|
571 |
if runningon_source and runningon_target: |
572 |
raise errors.OpExecError("Instance seems to be running on two nodes," |
573 |
" or the hypervisor is confused; you will have"
|
574 |
" to ensure manually that it runs only on one"
|
575 |
" and restart this operation")
|
576 |
|
577 |
if not (runningon_source or runningon_target): |
578 |
raise errors.OpExecError("Instance does not seem to be running at all;" |
579 |
" in this case it's safer to repair by"
|
580 |
" running 'gnt-instance stop' to ensure disk"
|
581 |
" shutdown, and then restarting it")
|
582 |
|
583 |
if runningon_target:
|
584 |
# the migration has actually succeeded, we need to update the config
|
585 |
self.feedback_fn("* instance running on secondary node (%s)," |
586 |
" updating config" %
|
587 |
self.cfg.GetNodeName(self.target_node_uuid)) |
588 |
self.instance.primary_node = self.target_node_uuid |
589 |
self.cfg.Update(self.instance, self.feedback_fn) |
590 |
demoted_node_uuid = self.source_node_uuid
|
591 |
else:
|
592 |
self.feedback_fn("* instance confirmed to be running on its" |
593 |
" primary node (%s)" %
|
594 |
self.cfg.GetNodeName(self.source_node_uuid)) |
595 |
demoted_node_uuid = self.target_node_uuid
|
596 |
|
597 |
if self.instance.disk_template in constants.DTS_INT_MIRROR: |
598 |
self._EnsureSecondary(demoted_node_uuid)
|
599 |
try:
|
600 |
self._WaitUntilSync()
|
601 |
except errors.OpExecError:
|
602 |
# we ignore here errors, since if the device is standalone, it
|
603 |
# won't be able to sync
|
604 |
pass
|
605 |
self._GoStandalone()
|
606 |
self._GoReconnect(False) |
607 |
self._WaitUntilSync()
|
608 |
|
609 |
self.feedback_fn("* done") |
610 |
|
611 |
def _RevertDiskStatus(self): |
612 |
"""Try to revert the disk status after a failed migration.
|
613 |
|
614 |
"""
|
615 |
if self.instance.disk_template in constants.DTS_EXT_MIRROR: |
616 |
return
|
617 |
|
618 |
try:
|
619 |
self._EnsureSecondary(self.target_node_uuid) |
620 |
self._GoStandalone()
|
621 |
self._GoReconnect(False) |
622 |
self._WaitUntilSync()
|
623 |
except errors.OpExecError, err:
|
624 |
self.lu.LogWarning("Migration failed and I can't reconnect the drives," |
625 |
" please try to recover the instance manually;"
|
626 |
" error '%s'" % str(err)) |
627 |
|
628 |
def _AbortMigration(self): |
629 |
"""Call the hypervisor code to abort a started migration.
|
630 |
|
631 |
"""
|
632 |
abort_result = self.rpc.call_instance_finalize_migration_dst(
|
633 |
self.target_node_uuid, self.instance, self.migration_info, |
634 |
False)
|
635 |
abort_msg = abort_result.fail_msg |
636 |
if abort_msg:
|
637 |
logging.error("Aborting migration failed on target node %s: %s",
|
638 |
self.cfg.GetNodeName(self.target_node_uuid), abort_msg) |
639 |
# Don't raise an exception here, as we stil have to try to revert the
|
640 |
# disk status, even if this step failed.
|
641 |
|
642 |
abort_result = self.rpc.call_instance_finalize_migration_src(
|
643 |
self.source_node_uuid, self.instance, False, self.live) |
644 |
abort_msg = abort_result.fail_msg |
645 |
if abort_msg:
|
646 |
logging.error("Aborting migration failed on source node %s: %s",
|
647 |
self.cfg.GetNodeName(self.source_node_uuid), abort_msg) |
648 |
|
649 |
def _ExecMigration(self): |
650 |
"""Migrate an instance.
|
651 |
|
652 |
The migrate is done by:
|
653 |
- change the disks into dual-master mode
|
654 |
- wait until disks are fully synchronized again
|
655 |
- migrate the instance
|
656 |
- change disks on the new secondary node (the old primary) to secondary
|
657 |
- wait until disks are fully synchronized
|
658 |
- change disks into single-master mode
|
659 |
|
660 |
"""
|
661 |
# Check for hypervisor version mismatch and warn the user.
|
662 |
hvspecs = [(self.instance.hypervisor,
|
663 |
self.cfg.GetClusterInfo().hvparams[self.instance.hypervisor])] |
664 |
nodeinfo = self.rpc.call_node_info(
|
665 |
[self.source_node_uuid, self.target_node_uuid], None, hvspecs) |
666 |
for ninfo in nodeinfo.values(): |
667 |
ninfo.Raise("Unable to retrieve node information from node '%s'" %
|
668 |
ninfo.node) |
669 |
(_, _, (src_info, )) = nodeinfo[self.source_node_uuid].payload
|
670 |
(_, _, (dst_info, )) = nodeinfo[self.target_node_uuid].payload
|
671 |
|
672 |
if ((constants.HV_NODEINFO_KEY_VERSION in src_info) and |
673 |
(constants.HV_NODEINFO_KEY_VERSION in dst_info)):
|
674 |
src_version = src_info[constants.HV_NODEINFO_KEY_VERSION] |
675 |
dst_version = dst_info[constants.HV_NODEINFO_KEY_VERSION] |
676 |
if src_version != dst_version:
|
677 |
self.feedback_fn("* warning: hypervisor version mismatch between" |
678 |
" source (%s) and target (%s) node" %
|
679 |
(src_version, dst_version)) |
680 |
|
681 |
self.feedback_fn("* checking disk consistency between source and target") |
682 |
for (idx, dev) in enumerate(self.instance.disks): |
683 |
if not CheckDiskConsistency(self.lu, self.instance, dev, |
684 |
self.target_node_uuid,
|
685 |
False):
|
686 |
raise errors.OpExecError("Disk %s is degraded or not fully" |
687 |
" synchronized on target node,"
|
688 |
" aborting migration" % idx)
|
689 |
|
690 |
if self.current_mem > self.tgt_free_mem: |
691 |
if not self.allow_runtime_changes: |
692 |
raise errors.OpExecError("Memory ballooning not allowed and not enough" |
693 |
" free memory to fit instance %s on target"
|
694 |
" node %s (have %dMB, need %dMB)" %
|
695 |
(self.instance.name,
|
696 |
self.cfg.GetNodeName(self.target_node_uuid), |
697 |
self.tgt_free_mem, self.current_mem)) |
698 |
self.feedback_fn("* setting instance memory to %s" % self.tgt_free_mem) |
699 |
rpcres = self.rpc.call_instance_balloon_memory(self.instance.primary_node, |
700 |
self.instance,
|
701 |
self.tgt_free_mem)
|
702 |
rpcres.Raise("Cannot modify instance runtime memory")
|
703 |
|
704 |
# First get the migration information from the remote node
|
705 |
result = self.rpc.call_migration_info(self.source_node_uuid, self.instance) |
706 |
msg = result.fail_msg |
707 |
if msg:
|
708 |
log_err = ("Failed fetching source migration information from %s: %s" %
|
709 |
(self.cfg.GetNodeName(self.source_node_uuid), msg)) |
710 |
logging.error(log_err) |
711 |
raise errors.OpExecError(log_err)
|
712 |
|
713 |
self.migration_info = migration_info = result.payload
|
714 |
|
715 |
if self.instance.disk_template not in constants.DTS_EXT_MIRROR: |
716 |
# Then switch the disks to master/master mode
|
717 |
self._EnsureSecondary(self.target_node_uuid) |
718 |
self._GoStandalone()
|
719 |
self._GoReconnect(True) |
720 |
self._WaitUntilSync()
|
721 |
|
722 |
self.feedback_fn("* preparing %s to accept the instance" % |
723 |
self.cfg.GetNodeName(self.target_node_uuid)) |
724 |
result = self.rpc.call_accept_instance(self.target_node_uuid, |
725 |
self.instance,
|
726 |
migration_info, |
727 |
self.nodes_ip[self.target_node_uuid]) |
728 |
|
729 |
msg = result.fail_msg |
730 |
if msg:
|
731 |
logging.error("Instance pre-migration failed, trying to revert"
|
732 |
" disk status: %s", msg)
|
733 |
self.feedback_fn("Pre-migration failed, aborting") |
734 |
self._AbortMigration()
|
735 |
self._RevertDiskStatus()
|
736 |
raise errors.OpExecError("Could not pre-migrate instance %s: %s" % |
737 |
(self.instance.name, msg))
|
738 |
|
739 |
self.feedback_fn("* migrating instance to %s" % |
740 |
self.cfg.GetNodeName(self.target_node_uuid)) |
741 |
cluster = self.cfg.GetClusterInfo()
|
742 |
result = self.rpc.call_instance_migrate(
|
743 |
self.source_node_uuid, cluster.cluster_name, self.instance, |
744 |
self.nodes_ip[self.target_node_uuid], self.live) |
745 |
msg = result.fail_msg |
746 |
if msg:
|
747 |
logging.error("Instance migration failed, trying to revert"
|
748 |
" disk status: %s", msg)
|
749 |
self.feedback_fn("Migration failed, aborting") |
750 |
self._AbortMigration()
|
751 |
self._RevertDiskStatus()
|
752 |
raise errors.OpExecError("Could not migrate instance %s: %s" % |
753 |
(self.instance.name, msg))
|
754 |
|
755 |
self.feedback_fn("* starting memory transfer") |
756 |
last_feedback = time.time() |
757 |
while True: |
758 |
result = self.rpc.call_instance_get_migration_status(
|
759 |
self.source_node_uuid, self.instance) |
760 |
msg = result.fail_msg |
761 |
ms = result.payload # MigrationStatus instance
|
762 |
if msg or (ms.status in constants.HV_MIGRATION_FAILED_STATUSES): |
763 |
logging.error("Instance migration failed, trying to revert"
|
764 |
" disk status: %s", msg)
|
765 |
self.feedback_fn("Migration failed, aborting") |
766 |
self._AbortMigration()
|
767 |
self._RevertDiskStatus()
|
768 |
if not msg: |
769 |
msg = "hypervisor returned failure"
|
770 |
raise errors.OpExecError("Could not migrate instance %s: %s" % |
771 |
(self.instance.name, msg))
|
772 |
|
773 |
if result.payload.status != constants.HV_MIGRATION_ACTIVE:
|
774 |
self.feedback_fn("* memory transfer complete") |
775 |
break
|
776 |
|
777 |
if (utils.TimeoutExpired(last_feedback,
|
778 |
self._MIGRATION_FEEDBACK_INTERVAL) and |
779 |
ms.transferred_ram is not None): |
780 |
mem_progress = 100 * float(ms.transferred_ram) / float(ms.total_ram) |
781 |
self.feedback_fn("* memory transfer progress: %.2f %%" % mem_progress) |
782 |
last_feedback = time.time() |
783 |
|
784 |
time.sleep(self._MIGRATION_POLL_INTERVAL)
|
785 |
|
786 |
result = self.rpc.call_instance_finalize_migration_src(
|
787 |
self.source_node_uuid, self.instance, True, self.live) |
788 |
msg = result.fail_msg |
789 |
if msg:
|
790 |
logging.error("Instance migration succeeded, but finalization failed"
|
791 |
" on the source node: %s", msg)
|
792 |
raise errors.OpExecError("Could not finalize instance migration: %s" % |
793 |
msg) |
794 |
|
795 |
self.instance.primary_node = self.target_node_uuid |
796 |
|
797 |
# distribute new instance config to the other nodes
|
798 |
self.cfg.Update(self.instance, self.feedback_fn) |
799 |
|
800 |
result = self.rpc.call_instance_finalize_migration_dst(
|
801 |
self.target_node_uuid, self.instance, migration_info, True) |
802 |
msg = result.fail_msg |
803 |
if msg:
|
804 |
logging.error("Instance migration succeeded, but finalization failed"
|
805 |
" on the target node: %s", msg)
|
806 |
raise errors.OpExecError("Could not finalize instance migration: %s" % |
807 |
msg) |
808 |
|
809 |
if self.instance.disk_template not in constants.DTS_EXT_MIRROR: |
810 |
self._EnsureSecondary(self.source_node_uuid) |
811 |
self._WaitUntilSync()
|
812 |
self._GoStandalone()
|
813 |
self._GoReconnect(False) |
814 |
self._WaitUntilSync()
|
815 |
|
816 |
# If the instance's disk template is `rbd' or `ext' and there was a
|
817 |
# successful migration, unmap the device from the source node.
|
818 |
if self.instance.disk_template in (constants.DT_RBD, constants.DT_EXT): |
819 |
disks = ExpandCheckDisks(self.instance, self.instance.disks) |
820 |
self.feedback_fn("* unmapping instance's disks from %s" % |
821 |
self.cfg.GetNodeName(self.source_node_uuid)) |
822 |
for disk in disks: |
823 |
result = self.rpc.call_blockdev_shutdown(self.source_node_uuid, |
824 |
(disk, self.instance))
|
825 |
msg = result.fail_msg |
826 |
if msg:
|
827 |
logging.error("Migration was successful, but couldn't unmap the"
|
828 |
" block device %s on source node %s: %s",
|
829 |
disk.iv_name, |
830 |
self.cfg.GetNodeName(self.source_node_uuid), msg) |
831 |
logging.error("You need to unmap the device %s manually on %s",
|
832 |
disk.iv_name, |
833 |
self.cfg.GetNodeName(self.source_node_uuid)) |
834 |
|
835 |
self.feedback_fn("* done") |
836 |
|
837 |
def _ExecFailover(self): |
838 |
"""Failover an instance.
|
839 |
|
840 |
The failover is done by shutting it down on its present node and
|
841 |
starting it on the secondary.
|
842 |
|
843 |
"""
|
844 |
primary_node = self.cfg.GetNodeInfo(self.instance.primary_node) |
845 |
|
846 |
source_node_uuid = self.instance.primary_node
|
847 |
|
848 |
if self.instance.disks_active: |
849 |
self.feedback_fn("* checking disk consistency between source and target") |
850 |
for (idx, dev) in enumerate(self.instance.disks): |
851 |
# for drbd, these are drbd over lvm
|
852 |
if not CheckDiskConsistency(self.lu, self.instance, dev, |
853 |
self.target_node_uuid, False): |
854 |
if primary_node.offline:
|
855 |
self.feedback_fn("Node %s is offline, ignoring degraded disk %s on" |
856 |
" target node %s" %
|
857 |
(primary_node.name, idx, |
858 |
self.cfg.GetNodeName(self.target_node_uuid))) |
859 |
elif not self.ignore_consistency: |
860 |
raise errors.OpExecError("Disk %s is degraded on target node," |
861 |
" aborting failover" % idx)
|
862 |
else:
|
863 |
self.feedback_fn("* not checking disk consistency as instance is not" |
864 |
" running")
|
865 |
|
866 |
self.feedback_fn("* shutting down instance on source node") |
867 |
logging.info("Shutting down instance %s on node %s",
|
868 |
self.instance.name, self.cfg.GetNodeName(source_node_uuid)) |
869 |
|
870 |
result = self.rpc.call_instance_shutdown(source_node_uuid, self.instance, |
871 |
self.shutdown_timeout,
|
872 |
self.lu.op.reason)
|
873 |
msg = result.fail_msg |
874 |
if msg:
|
875 |
if self.ignore_consistency or primary_node.offline: |
876 |
self.lu.LogWarning("Could not shutdown instance %s on node %s," |
877 |
" proceeding anyway; please make sure node"
|
878 |
" %s is down; error details: %s",
|
879 |
self.instance.name,
|
880 |
self.cfg.GetNodeName(source_node_uuid),
|
881 |
self.cfg.GetNodeName(source_node_uuid), msg)
|
882 |
else:
|
883 |
raise errors.OpExecError("Could not shutdown instance %s on" |
884 |
" node %s: %s" %
|
885 |
(self.instance.name,
|
886 |
self.cfg.GetNodeName(source_node_uuid), msg))
|
887 |
|
888 |
self.feedback_fn("* deactivating the instance's disks on source node") |
889 |
if not ShutdownInstanceDisks(self.lu, self.instance, ignore_primary=True): |
890 |
raise errors.OpExecError("Can't shut down the instance's disks") |
891 |
|
892 |
self.instance.primary_node = self.target_node_uuid |
893 |
# distribute new instance config to the other nodes
|
894 |
self.cfg.Update(self.instance, self.feedback_fn) |
895 |
|
896 |
# Only start the instance if it's marked as up
|
897 |
if self.instance.admin_state == constants.ADMINST_UP: |
898 |
self.feedback_fn("* activating the instance's disks on target node %s" % |
899 |
self.cfg.GetNodeName(self.target_node_uuid)) |
900 |
logging.info("Starting instance %s on node %s", self.instance.name, |
901 |
self.cfg.GetNodeName(self.target_node_uuid)) |
902 |
|
903 |
disks_ok, _ = AssembleInstanceDisks(self.lu, self.instance, |
904 |
ignore_secondaries=True)
|
905 |
if not disks_ok: |
906 |
ShutdownInstanceDisks(self.lu, self.instance) |
907 |
raise errors.OpExecError("Can't activate the instance's disks") |
908 |
|
909 |
self.feedback_fn("* starting the instance on the target node %s" % |
910 |
self.cfg.GetNodeName(self.target_node_uuid)) |
911 |
result = self.rpc.call_instance_start(self.target_node_uuid, |
912 |
(self.instance, None, None), False, |
913 |
self.lu.op.reason)
|
914 |
msg = result.fail_msg |
915 |
if msg:
|
916 |
ShutdownInstanceDisks(self.lu, self.instance) |
917 |
raise errors.OpExecError("Could not start instance %s on node %s: %s" % |
918 |
(self.instance.name,
|
919 |
self.cfg.GetNodeName(self.target_node_uuid), |
920 |
msg)) |
921 |
|
922 |
def Exec(self, feedback_fn): |
923 |
"""Perform the migration.
|
924 |
|
925 |
"""
|
926 |
self.feedback_fn = feedback_fn
|
927 |
self.source_node_uuid = self.instance.primary_node |
928 |
|
929 |
# FIXME: if we implement migrate-to-any in DRBD, this needs fixing
|
930 |
if self.instance.disk_template in constants.DTS_INT_MIRROR: |
931 |
self.target_node_uuid = self.instance.secondary_nodes[0] |
932 |
# Otherwise self.target_node has been populated either
|
933 |
# directly, or through an iallocator.
|
934 |
|
935 |
self.all_node_uuids = [self.source_node_uuid, self.target_node_uuid] |
936 |
self.nodes_ip = dict((uuid, node.secondary_ip) for (uuid, node) |
937 |
in self.cfg.GetMultiNodeInfo(self.all_node_uuids)) |
938 |
|
939 |
if self.failover: |
940 |
feedback_fn("Failover instance %s" % self.instance.name) |
941 |
self._ExecFailover()
|
942 |
else:
|
943 |
feedback_fn("Migrating instance %s" % self.instance.name) |
944 |
|
945 |
if self.cleanup: |
946 |
return self._ExecCleanup() |
947 |
else:
|
948 |
return self._ExecMigration() |