root / lib / config.py @ 531baf8e
History | View | Annotate | Download (34.1 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 |
"""Configuration management for Ganeti
|
23 |
|
24 |
This module provides the interface to the Ganeti cluster configuration.
|
25 |
|
26 |
The configuration data is stored on every node but is updated on the master
|
27 |
only. After each update, the master distributes the data to the other nodes.
|
28 |
|
29 |
Currently, the data storage format is JSON. YAML was slow and consuming too
|
30 |
much memory.
|
31 |
|
32 |
"""
|
33 |
|
34 |
import os |
35 |
import tempfile |
36 |
import random |
37 |
import logging |
38 |
|
39 |
from ganeti import errors |
40 |
from ganeti import locking |
41 |
from ganeti import utils |
42 |
from ganeti import constants |
43 |
from ganeti import rpc |
44 |
from ganeti import objects |
45 |
from ganeti import serializer |
46 |
|
47 |
|
48 |
_config_lock = locking.SharedLock() |
49 |
|
50 |
|
51 |
def _ValidateConfig(data): |
52 |
"""Verifies that a configuration objects looks valid.
|
53 |
|
54 |
This only verifies the version of the configuration.
|
55 |
|
56 |
@raise errors.ConfigurationError: if the version differs from what
|
57 |
we expect
|
58 |
|
59 |
"""
|
60 |
if data.version != constants.CONFIG_VERSION:
|
61 |
raise errors.ConfigurationError("Cluster configuration version" |
62 |
" mismatch, got %s instead of %s" %
|
63 |
(data.version, |
64 |
constants.CONFIG_VERSION)) |
65 |
|
66 |
|
67 |
class ConfigWriter: |
68 |
"""The interface to the cluster configuration.
|
69 |
|
70 |
"""
|
71 |
def __init__(self, cfg_file=None, offline=False): |
72 |
self.write_count = 0 |
73 |
self._lock = _config_lock
|
74 |
self._config_data = None |
75 |
self._offline = offline
|
76 |
if cfg_file is None: |
77 |
self._cfg_file = constants.CLUSTER_CONF_FILE
|
78 |
else:
|
79 |
self._cfg_file = cfg_file
|
80 |
self._temporary_ids = set() |
81 |
self._temporary_drbds = {}
|
82 |
# Note: in order to prevent errors when resolving our name in
|
83 |
# _DistributeConfig, we compute it here once and reuse it; it's
|
84 |
# better to raise an error before starting to modify the config
|
85 |
# file than after it was modified
|
86 |
self._my_hostname = utils.HostInfo().name
|
87 |
self._OpenConfig()
|
88 |
|
89 |
# this method needs to be static, so that we can call it on the class
|
90 |
@staticmethod
|
91 |
def IsCluster(): |
92 |
"""Check if the cluster is configured.
|
93 |
|
94 |
"""
|
95 |
return os.path.exists(constants.CLUSTER_CONF_FILE)
|
96 |
|
97 |
@locking.ssynchronized(_config_lock, shared=1) |
98 |
def GenerateMAC(self): |
99 |
"""Generate a MAC for an instance.
|
100 |
|
101 |
This should check the current instances for duplicates.
|
102 |
|
103 |
"""
|
104 |
prefix = self._config_data.cluster.mac_prefix
|
105 |
all_macs = self._AllMACs()
|
106 |
retries = 64
|
107 |
while retries > 0: |
108 |
byte1 = random.randrange(0, 256) |
109 |
byte2 = random.randrange(0, 256) |
110 |
byte3 = random.randrange(0, 256) |
111 |
mac = "%s:%02x:%02x:%02x" % (prefix, byte1, byte2, byte3)
|
112 |
if mac not in all_macs: |
113 |
break
|
114 |
retries -= 1
|
115 |
else:
|
116 |
raise errors.ConfigurationError("Can't generate unique MAC") |
117 |
return mac
|
118 |
|
119 |
@locking.ssynchronized(_config_lock, shared=1) |
120 |
def IsMacInUse(self, mac): |
121 |
"""Predicate: check if the specified MAC is in use in the Ganeti cluster.
|
122 |
|
123 |
This only checks instances managed by this cluster, it does not
|
124 |
check for potential collisions elsewhere.
|
125 |
|
126 |
"""
|
127 |
all_macs = self._AllMACs()
|
128 |
return mac in all_macs |
129 |
|
130 |
@locking.ssynchronized(_config_lock, shared=1) |
131 |
def GenerateDRBDSecret(self): |
132 |
"""Generate a DRBD secret.
|
133 |
|
134 |
This checks the current disks for duplicates.
|
135 |
|
136 |
"""
|
137 |
all_secrets = self._AllDRBDSecrets()
|
138 |
retries = 64
|
139 |
while retries > 0: |
140 |
secret = utils.GenerateSecret() |
141 |
if secret not in all_secrets: |
142 |
break
|
143 |
retries -= 1
|
144 |
else:
|
145 |
raise errors.ConfigurationError("Can't generate unique DRBD secret") |
146 |
return secret
|
147 |
|
148 |
def _ComputeAllLVs(self): |
149 |
"""Compute the list of all LVs.
|
150 |
|
151 |
"""
|
152 |
lvnames = set()
|
153 |
for instance in self._config_data.instances.values(): |
154 |
node_data = instance.MapLVsByNode() |
155 |
for lv_list in node_data.values(): |
156 |
lvnames.update(lv_list) |
157 |
return lvnames
|
158 |
|
159 |
@locking.ssynchronized(_config_lock, shared=1) |
160 |
def GenerateUniqueID(self, exceptions=None): |
161 |
"""Generate an unique disk name.
|
162 |
|
163 |
This checks the current node, instances and disk names for
|
164 |
duplicates.
|
165 |
|
166 |
@param exceptions: a list with some other names which should be checked
|
167 |
for uniqueness (used for example when you want to get
|
168 |
more than one id at one time without adding each one in
|
169 |
turn to the config file)
|
170 |
|
171 |
@rtype: string
|
172 |
@return: the unique id
|
173 |
|
174 |
"""
|
175 |
existing = set()
|
176 |
existing.update(self._temporary_ids)
|
177 |
existing.update(self._ComputeAllLVs())
|
178 |
existing.update(self._config_data.instances.keys())
|
179 |
existing.update(self._config_data.nodes.keys())
|
180 |
if exceptions is not None: |
181 |
existing.update(exceptions) |
182 |
retries = 64
|
183 |
while retries > 0: |
184 |
unique_id = utils.NewUUID() |
185 |
if unique_id not in existing and unique_id is not None: |
186 |
break
|
187 |
else:
|
188 |
raise errors.ConfigurationError("Not able generate an unique ID" |
189 |
" (last tried ID: %s" % unique_id)
|
190 |
self._temporary_ids.add(unique_id)
|
191 |
return unique_id
|
192 |
|
193 |
def _AllMACs(self): |
194 |
"""Return all MACs present in the config.
|
195 |
|
196 |
@rtype: list
|
197 |
@return: the list of all MACs
|
198 |
|
199 |
"""
|
200 |
result = [] |
201 |
for instance in self._config_data.instances.values(): |
202 |
for nic in instance.nics: |
203 |
result.append(nic.mac) |
204 |
|
205 |
return result
|
206 |
|
207 |
def _AllDRBDSecrets(self): |
208 |
"""Return all DRBD secrets present in the config.
|
209 |
|
210 |
@rtype: list
|
211 |
@return: the list of all DRBD secrets
|
212 |
|
213 |
"""
|
214 |
def helper(disk, result): |
215 |
"""Recursively gather secrets from this disk."""
|
216 |
if disk.dev_type == constants.DT_DRBD8:
|
217 |
result.append(disk.logical_id[5])
|
218 |
if disk.children:
|
219 |
for child in disk.children: |
220 |
helper(child, result) |
221 |
|
222 |
result = [] |
223 |
for instance in self._config_data.instances.values(): |
224 |
for disk in instance.disks: |
225 |
helper(disk, result) |
226 |
|
227 |
return result
|
228 |
|
229 |
@locking.ssynchronized(_config_lock, shared=1) |
230 |
def VerifyConfig(self): |
231 |
"""Verify function.
|
232 |
|
233 |
"""
|
234 |
result = [] |
235 |
seen_macs = [] |
236 |
ports = {} |
237 |
data = self._config_data
|
238 |
for instance_name in data.instances: |
239 |
instance = data.instances[instance_name] |
240 |
if instance.primary_node not in data.nodes: |
241 |
result.append("instance '%s' has invalid primary node '%s'" %
|
242 |
(instance_name, instance.primary_node)) |
243 |
for snode in instance.secondary_nodes: |
244 |
if snode not in data.nodes: |
245 |
result.append("instance '%s' has invalid secondary node '%s'" %
|
246 |
(instance_name, snode)) |
247 |
for idx, nic in enumerate(instance.nics): |
248 |
if nic.mac in seen_macs: |
249 |
result.append("instance '%s' has NIC %d mac %s duplicate" %
|
250 |
(instance_name, idx, nic.mac)) |
251 |
else:
|
252 |
seen_macs.append(nic.mac) |
253 |
|
254 |
# gather the drbd ports for duplicate checks
|
255 |
for dsk in instance.disks: |
256 |
if dsk.dev_type in constants.LDS_DRBD: |
257 |
tcp_port = dsk.logical_id[2]
|
258 |
if tcp_port not in ports: |
259 |
ports[tcp_port] = [] |
260 |
ports[tcp_port].append((instance.name, "drbd disk %s" % dsk.iv_name))
|
261 |
# gather network port reservation
|
262 |
net_port = getattr(instance, "network_port", None) |
263 |
if net_port is not None: |
264 |
if net_port not in ports: |
265 |
ports[net_port] = [] |
266 |
ports[net_port].append((instance.name, "network port"))
|
267 |
|
268 |
# cluster-wide pool of free ports
|
269 |
for free_port in data.cluster.tcpudp_port_pool: |
270 |
if free_port not in ports: |
271 |
ports[free_port] = [] |
272 |
ports[free_port].append(("cluster", "port marked as free")) |
273 |
|
274 |
# compute tcp/udp duplicate ports
|
275 |
keys = ports.keys() |
276 |
keys.sort() |
277 |
for pnum in keys: |
278 |
pdata = ports[pnum] |
279 |
if len(pdata) > 1: |
280 |
txt = ", ".join(["%s/%s" % val for val in pdata]) |
281 |
result.append("tcp/udp port %s has duplicates: %s" % (pnum, txt))
|
282 |
|
283 |
# highest used tcp port check
|
284 |
if keys:
|
285 |
if keys[-1] > data.cluster.highest_used_port: |
286 |
result.append("Highest used port mismatch, saved %s, computed %s" %
|
287 |
(data.cluster.highest_used_port, keys[-1]))
|
288 |
|
289 |
if not data.nodes[data.cluster.master_node].master_candidate: |
290 |
result.append("Master node is not a master candidate")
|
291 |
|
292 |
mc_now, mc_max = self._UnlockedGetMasterCandidateStats()
|
293 |
if mc_now < mc_max:
|
294 |
result.append("Not enough master candidates: actual %d, target %d" %
|
295 |
(mc_now, mc_max)) |
296 |
|
297 |
return result
|
298 |
|
299 |
def _UnlockedSetDiskID(self, disk, node_name): |
300 |
"""Convert the unique ID to the ID needed on the target nodes.
|
301 |
|
302 |
This is used only for drbd, which needs ip/port configuration.
|
303 |
|
304 |
The routine descends down and updates its children also, because
|
305 |
this helps when the only the top device is passed to the remote
|
306 |
node.
|
307 |
|
308 |
This function is for internal use, when the config lock is already held.
|
309 |
|
310 |
"""
|
311 |
if disk.children:
|
312 |
for child in disk.children: |
313 |
self._UnlockedSetDiskID(child, node_name)
|
314 |
|
315 |
if disk.logical_id is None and disk.physical_id is not None: |
316 |
return
|
317 |
if disk.dev_type == constants.LD_DRBD8:
|
318 |
pnode, snode, port, pminor, sminor, secret = disk.logical_id |
319 |
if node_name not in (pnode, snode): |
320 |
raise errors.ConfigurationError("DRBD device not knowing node %s" % |
321 |
node_name) |
322 |
pnode_info = self._UnlockedGetNodeInfo(pnode)
|
323 |
snode_info = self._UnlockedGetNodeInfo(snode)
|
324 |
if pnode_info is None or snode_info is None: |
325 |
raise errors.ConfigurationError("Can't find primary or secondary node" |
326 |
" for %s" % str(disk)) |
327 |
p_data = (pnode_info.secondary_ip, port) |
328 |
s_data = (snode_info.secondary_ip, port) |
329 |
if pnode == node_name:
|
330 |
disk.physical_id = p_data + s_data + (pminor, secret) |
331 |
else: # it must be secondary, we tested above |
332 |
disk.physical_id = s_data + p_data + (sminor, secret) |
333 |
else:
|
334 |
disk.physical_id = disk.logical_id |
335 |
return
|
336 |
|
337 |
@locking.ssynchronized(_config_lock)
|
338 |
def SetDiskID(self, disk, node_name): |
339 |
"""Convert the unique ID to the ID needed on the target nodes.
|
340 |
|
341 |
This is used only for drbd, which needs ip/port configuration.
|
342 |
|
343 |
The routine descends down and updates its children also, because
|
344 |
this helps when the only the top device is passed to the remote
|
345 |
node.
|
346 |
|
347 |
"""
|
348 |
return self._UnlockedSetDiskID(disk, node_name) |
349 |
|
350 |
@locking.ssynchronized(_config_lock)
|
351 |
def AddTcpUdpPort(self, port): |
352 |
"""Adds a new port to the available port pool.
|
353 |
|
354 |
"""
|
355 |
if not isinstance(port, int): |
356 |
raise errors.ProgrammerError("Invalid type passed for port") |
357 |
|
358 |
self._config_data.cluster.tcpudp_port_pool.add(port)
|
359 |
self._WriteConfig()
|
360 |
|
361 |
@locking.ssynchronized(_config_lock, shared=1) |
362 |
def GetPortList(self): |
363 |
"""Returns a copy of the current port list.
|
364 |
|
365 |
"""
|
366 |
return self._config_data.cluster.tcpudp_port_pool.copy() |
367 |
|
368 |
@locking.ssynchronized(_config_lock)
|
369 |
def AllocatePort(self): |
370 |
"""Allocate a port.
|
371 |
|
372 |
The port will be taken from the available port pool or from the
|
373 |
default port range (and in this case we increase
|
374 |
highest_used_port).
|
375 |
|
376 |
"""
|
377 |
# If there are TCP/IP ports configured, we use them first.
|
378 |
if self._config_data.cluster.tcpudp_port_pool: |
379 |
port = self._config_data.cluster.tcpudp_port_pool.pop()
|
380 |
else:
|
381 |
port = self._config_data.cluster.highest_used_port + 1 |
382 |
if port >= constants.LAST_DRBD_PORT:
|
383 |
raise errors.ConfigurationError("The highest used port is greater" |
384 |
" than %s. Aborting." %
|
385 |
constants.LAST_DRBD_PORT) |
386 |
self._config_data.cluster.highest_used_port = port
|
387 |
|
388 |
self._WriteConfig()
|
389 |
return port
|
390 |
|
391 |
def _ComputeDRBDMap(self, instance): |
392 |
"""Compute the used DRBD minor/nodes.
|
393 |
|
394 |
@return: dictionary of node_name: dict of minor: instance_name;
|
395 |
the returned dict will have all the nodes in it (even if with
|
396 |
an empty list).
|
397 |
|
398 |
"""
|
399 |
def _AppendUsedPorts(instance_name, disk, used): |
400 |
if disk.dev_type == constants.LD_DRBD8 and len(disk.logical_id) >= 5: |
401 |
nodeA, nodeB, dummy, minorA, minorB = disk.logical_id[:5]
|
402 |
for node, port in ((nodeA, minorA), (nodeB, minorB)): |
403 |
assert node in used, "Instance node not found in node list" |
404 |
if port in used[node]: |
405 |
raise errors.ProgrammerError("DRBD minor already used:" |
406 |
" %s/%s, %s/%s" %
|
407 |
(node, port, instance_name, |
408 |
used[node][port])) |
409 |
|
410 |
used[node][port] = instance_name |
411 |
if disk.children:
|
412 |
for child in disk.children: |
413 |
_AppendUsedPorts(instance_name, child, used) |
414 |
|
415 |
my_dict = dict((node, {}) for node in self._config_data.nodes) |
416 |
for (node, minor), instance in self._temporary_drbds.iteritems(): |
417 |
my_dict[node][minor] = instance |
418 |
for instance in self._config_data.instances.itervalues(): |
419 |
for disk in instance.disks: |
420 |
_AppendUsedPorts(instance.name, disk, my_dict) |
421 |
return my_dict
|
422 |
|
423 |
@locking.ssynchronized(_config_lock)
|
424 |
def AllocateDRBDMinor(self, nodes, instance): |
425 |
"""Allocate a drbd minor.
|
426 |
|
427 |
The free minor will be automatically computed from the existing
|
428 |
devices. A node can be given multiple times in order to allocate
|
429 |
multiple minors. The result is the list of minors, in the same
|
430 |
order as the passed nodes.
|
431 |
|
432 |
"""
|
433 |
d_map = self._ComputeDRBDMap(instance)
|
434 |
result = [] |
435 |
for nname in nodes: |
436 |
ndata = d_map[nname] |
437 |
if not ndata: |
438 |
# no minors used, we can start at 0
|
439 |
result.append(0)
|
440 |
ndata[0] = instance
|
441 |
self._temporary_drbds[(nname, 0)] = instance |
442 |
continue
|
443 |
keys = ndata.keys() |
444 |
keys.sort() |
445 |
ffree = utils.FirstFree(keys) |
446 |
if ffree is None: |
447 |
# return the next minor
|
448 |
# TODO: implement high-limit check
|
449 |
minor = keys[-1] + 1 |
450 |
else:
|
451 |
minor = ffree |
452 |
result.append(minor) |
453 |
ndata[minor] = instance |
454 |
assert (nname, minor) not in self._temporary_drbds, \ |
455 |
"Attempt to reuse reserved DRBD minor"
|
456 |
self._temporary_drbds[(nname, minor)] = instance
|
457 |
logging.debug("Request to allocate drbd minors, input: %s, returning %s",
|
458 |
nodes, result) |
459 |
return result
|
460 |
|
461 |
@locking.ssynchronized(_config_lock)
|
462 |
def ReleaseDRBDMinors(self, instance): |
463 |
"""Release temporary drbd minors allocated for a given instance.
|
464 |
|
465 |
This should be called on both the error paths and on the success
|
466 |
paths (after the instance has been added or updated).
|
467 |
|
468 |
@type instance: string
|
469 |
@param instance: the instance for which temporary minors should be
|
470 |
released
|
471 |
|
472 |
"""
|
473 |
for key, name in self._temporary_drbds.items(): |
474 |
if name == instance:
|
475 |
del self._temporary_drbds[key] |
476 |
|
477 |
@locking.ssynchronized(_config_lock, shared=1) |
478 |
def GetConfigVersion(self): |
479 |
"""Get the configuration version.
|
480 |
|
481 |
@return: Config version
|
482 |
|
483 |
"""
|
484 |
return self._config_data.version |
485 |
|
486 |
@locking.ssynchronized(_config_lock, shared=1) |
487 |
def GetClusterName(self): |
488 |
"""Get cluster name.
|
489 |
|
490 |
@return: Cluster name
|
491 |
|
492 |
"""
|
493 |
return self._config_data.cluster.cluster_name |
494 |
|
495 |
@locking.ssynchronized(_config_lock, shared=1) |
496 |
def GetMasterNode(self): |
497 |
"""Get the hostname of the master node for this cluster.
|
498 |
|
499 |
@return: Master hostname
|
500 |
|
501 |
"""
|
502 |
return self._config_data.cluster.master_node |
503 |
|
504 |
@locking.ssynchronized(_config_lock, shared=1) |
505 |
def GetMasterIP(self): |
506 |
"""Get the IP of the master node for this cluster.
|
507 |
|
508 |
@return: Master IP
|
509 |
|
510 |
"""
|
511 |
return self._config_data.cluster.master_ip |
512 |
|
513 |
@locking.ssynchronized(_config_lock, shared=1) |
514 |
def GetMasterNetdev(self): |
515 |
"""Get the master network device for this cluster.
|
516 |
|
517 |
"""
|
518 |
return self._config_data.cluster.master_netdev |
519 |
|
520 |
@locking.ssynchronized(_config_lock, shared=1) |
521 |
def GetFileStorageDir(self): |
522 |
"""Get the file storage dir for this cluster.
|
523 |
|
524 |
"""
|
525 |
return self._config_data.cluster.file_storage_dir |
526 |
|
527 |
@locking.ssynchronized(_config_lock, shared=1) |
528 |
def GetHypervisorType(self): |
529 |
"""Get the hypervisor type for this cluster.
|
530 |
|
531 |
"""
|
532 |
return self._config_data.cluster.default_hypervisor |
533 |
|
534 |
@locking.ssynchronized(_config_lock, shared=1) |
535 |
def GetHostKey(self): |
536 |
"""Return the rsa hostkey from the config.
|
537 |
|
538 |
@rtype: string
|
539 |
@return: the rsa hostkey
|
540 |
|
541 |
"""
|
542 |
return self._config_data.cluster.rsahostkeypub |
543 |
|
544 |
@locking.ssynchronized(_config_lock)
|
545 |
def AddInstance(self, instance): |
546 |
"""Add an instance to the config.
|
547 |
|
548 |
This should be used after creating a new instance.
|
549 |
|
550 |
@type instance: L{objects.Instance}
|
551 |
@param instance: the instance object
|
552 |
|
553 |
"""
|
554 |
if not isinstance(instance, objects.Instance): |
555 |
raise errors.ProgrammerError("Invalid type passed to AddInstance") |
556 |
|
557 |
if instance.disk_template != constants.DT_DISKLESS:
|
558 |
all_lvs = instance.MapLVsByNode() |
559 |
logging.info("Instance '%s' DISK_LAYOUT: %s", instance.name, all_lvs)
|
560 |
|
561 |
instance.serial_no = 1
|
562 |
self._config_data.instances[instance.name] = instance
|
563 |
self._WriteConfig()
|
564 |
|
565 |
def _SetInstanceStatus(self, instance_name, status): |
566 |
"""Set the instance's status to a given value.
|
567 |
|
568 |
"""
|
569 |
if status not in ("up", "down"): |
570 |
raise errors.ProgrammerError("Invalid status '%s' passed to" |
571 |
" ConfigWriter._SetInstanceStatus()" %
|
572 |
status) |
573 |
|
574 |
if instance_name not in self._config_data.instances: |
575 |
raise errors.ConfigurationError("Unknown instance '%s'" % |
576 |
instance_name) |
577 |
instance = self._config_data.instances[instance_name]
|
578 |
if instance.status != status:
|
579 |
instance.status = status |
580 |
instance.serial_no += 1
|
581 |
self._WriteConfig()
|
582 |
|
583 |
@locking.ssynchronized(_config_lock)
|
584 |
def MarkInstanceUp(self, instance_name): |
585 |
"""Mark the instance status to up in the config.
|
586 |
|
587 |
"""
|
588 |
self._SetInstanceStatus(instance_name, "up") |
589 |
|
590 |
@locking.ssynchronized(_config_lock)
|
591 |
def RemoveInstance(self, instance_name): |
592 |
"""Remove the instance from the configuration.
|
593 |
|
594 |
"""
|
595 |
if instance_name not in self._config_data.instances: |
596 |
raise errors.ConfigurationError("Unknown instance '%s'" % instance_name) |
597 |
del self._config_data.instances[instance_name] |
598 |
self._WriteConfig()
|
599 |
|
600 |
@locking.ssynchronized(_config_lock)
|
601 |
def RenameInstance(self, old_name, new_name): |
602 |
"""Rename an instance.
|
603 |
|
604 |
This needs to be done in ConfigWriter and not by RemoveInstance
|
605 |
combined with AddInstance as only we can guarantee an atomic
|
606 |
rename.
|
607 |
|
608 |
"""
|
609 |
if old_name not in self._config_data.instances: |
610 |
raise errors.ConfigurationError("Unknown instance '%s'" % old_name) |
611 |
inst = self._config_data.instances[old_name]
|
612 |
del self._config_data.instances[old_name] |
613 |
inst.name = new_name |
614 |
|
615 |
for disk in inst.disks: |
616 |
if disk.dev_type == constants.LD_FILE:
|
617 |
# rename the file paths in logical and physical id
|
618 |
file_storage_dir = os.path.dirname(os.path.dirname(disk.logical_id[1]))
|
619 |
disk.physical_id = disk.logical_id = (disk.logical_id[0],
|
620 |
os.path.join(file_storage_dir, |
621 |
inst.name, |
622 |
disk.iv_name)) |
623 |
|
624 |
self._config_data.instances[inst.name] = inst
|
625 |
self._WriteConfig()
|
626 |
|
627 |
@locking.ssynchronized(_config_lock)
|
628 |
def MarkInstanceDown(self, instance_name): |
629 |
"""Mark the status of an instance to down in the configuration.
|
630 |
|
631 |
"""
|
632 |
self._SetInstanceStatus(instance_name, "down") |
633 |
|
634 |
def _UnlockedGetInstanceList(self): |
635 |
"""Get the list of instances.
|
636 |
|
637 |
This function is for internal use, when the config lock is already held.
|
638 |
|
639 |
"""
|
640 |
return self._config_data.instances.keys() |
641 |
|
642 |
@locking.ssynchronized(_config_lock, shared=1) |
643 |
def GetInstanceList(self): |
644 |
"""Get the list of instances.
|
645 |
|
646 |
@return: array of instances, ex. ['instance2.example.com',
|
647 |
'instance1.example.com']
|
648 |
|
649 |
"""
|
650 |
return self._UnlockedGetInstanceList() |
651 |
|
652 |
@locking.ssynchronized(_config_lock, shared=1) |
653 |
def ExpandInstanceName(self, short_name): |
654 |
"""Attempt to expand an incomplete instance name.
|
655 |
|
656 |
"""
|
657 |
return utils.MatchNameComponent(short_name,
|
658 |
self._config_data.instances.keys())
|
659 |
|
660 |
def _UnlockedGetInstanceInfo(self, instance_name): |
661 |
"""Returns informations about an instance.
|
662 |
|
663 |
This function is for internal use, when the config lock is already held.
|
664 |
|
665 |
"""
|
666 |
if instance_name not in self._config_data.instances: |
667 |
return None |
668 |
|
669 |
return self._config_data.instances[instance_name] |
670 |
|
671 |
@locking.ssynchronized(_config_lock, shared=1) |
672 |
def GetInstanceInfo(self, instance_name): |
673 |
"""Returns informations about an instance.
|
674 |
|
675 |
It takes the information from the configuration file. Other informations of
|
676 |
an instance are taken from the live systems.
|
677 |
|
678 |
@param instance_name: name of the instance, e.g.
|
679 |
I{instance1.example.com}
|
680 |
|
681 |
@rtype: L{objects.Instance}
|
682 |
@return: the instance object
|
683 |
|
684 |
"""
|
685 |
return self._UnlockedGetInstanceInfo(instance_name) |
686 |
|
687 |
@locking.ssynchronized(_config_lock, shared=1) |
688 |
def GetAllInstancesInfo(self): |
689 |
"""Get the configuration of all instances.
|
690 |
|
691 |
@rtype: dict
|
692 |
@returns: dict of (instance, instance_info), where instance_info is what
|
693 |
would GetInstanceInfo return for the node
|
694 |
|
695 |
"""
|
696 |
my_dict = dict([(instance, self._UnlockedGetInstanceInfo(instance)) |
697 |
for instance in self._UnlockedGetInstanceList()]) |
698 |
return my_dict
|
699 |
|
700 |
@locking.ssynchronized(_config_lock)
|
701 |
def AddNode(self, node): |
702 |
"""Add a node to the configuration.
|
703 |
|
704 |
@type node: L{objects.Node}
|
705 |
@param node: a Node instance
|
706 |
|
707 |
"""
|
708 |
logging.info("Adding node %s to configuration" % node.name)
|
709 |
|
710 |
node.serial_no = 1
|
711 |
self._config_data.nodes[node.name] = node
|
712 |
self._config_data.cluster.serial_no += 1 |
713 |
self._WriteConfig()
|
714 |
|
715 |
@locking.ssynchronized(_config_lock)
|
716 |
def RemoveNode(self, node_name): |
717 |
"""Remove a node from the configuration.
|
718 |
|
719 |
"""
|
720 |
logging.info("Removing node %s from configuration" % node_name)
|
721 |
|
722 |
if node_name not in self._config_data.nodes: |
723 |
raise errors.ConfigurationError("Unknown node '%s'" % node_name) |
724 |
|
725 |
del self._config_data.nodes[node_name] |
726 |
self._config_data.cluster.serial_no += 1 |
727 |
self._WriteConfig()
|
728 |
|
729 |
@locking.ssynchronized(_config_lock, shared=1) |
730 |
def ExpandNodeName(self, short_name): |
731 |
"""Attempt to expand an incomplete instance name.
|
732 |
|
733 |
"""
|
734 |
return utils.MatchNameComponent(short_name,
|
735 |
self._config_data.nodes.keys())
|
736 |
|
737 |
def _UnlockedGetNodeInfo(self, node_name): |
738 |
"""Get the configuration of a node, as stored in the config.
|
739 |
|
740 |
This function is for internal use, when the config lock is already
|
741 |
held.
|
742 |
|
743 |
@param node_name: the node name, e.g. I{node1.example.com}
|
744 |
|
745 |
@rtype: L{objects.Node}
|
746 |
@return: the node object
|
747 |
|
748 |
"""
|
749 |
if node_name not in self._config_data.nodes: |
750 |
return None |
751 |
|
752 |
return self._config_data.nodes[node_name] |
753 |
|
754 |
|
755 |
@locking.ssynchronized(_config_lock, shared=1) |
756 |
def GetNodeInfo(self, node_name): |
757 |
"""Get the configuration of a node, as stored in the config.
|
758 |
|
759 |
This is just a locked wrapper over L{_UnlockedGetNodeInfo}.
|
760 |
|
761 |
@param node_name: the node name, e.g. I{node1.example.com}
|
762 |
|
763 |
@rtype: L{objects.Node}
|
764 |
@return: the node object
|
765 |
|
766 |
"""
|
767 |
return self._UnlockedGetNodeInfo(node_name) |
768 |
|
769 |
def _UnlockedGetNodeList(self): |
770 |
"""Return the list of nodes which are in the configuration.
|
771 |
|
772 |
This function is for internal use, when the config lock is already
|
773 |
held.
|
774 |
|
775 |
@rtype: list
|
776 |
|
777 |
"""
|
778 |
return self._config_data.nodes.keys() |
779 |
|
780 |
|
781 |
@locking.ssynchronized(_config_lock, shared=1) |
782 |
def GetNodeList(self): |
783 |
"""Return the list of nodes which are in the configuration.
|
784 |
|
785 |
"""
|
786 |
return self._UnlockedGetNodeList() |
787 |
|
788 |
@locking.ssynchronized(_config_lock, shared=1) |
789 |
def GetAllNodesInfo(self): |
790 |
"""Get the configuration of all nodes.
|
791 |
|
792 |
@rtype: dict
|
793 |
@return: dict of (node, node_info), where node_info is what
|
794 |
would GetNodeInfo return for the node
|
795 |
|
796 |
"""
|
797 |
my_dict = dict([(node, self._UnlockedGetNodeInfo(node)) |
798 |
for node in self._UnlockedGetNodeList()]) |
799 |
return my_dict
|
800 |
|
801 |
def _UnlockedGetMasterCandidateStats(self): |
802 |
"""Get the number of current and maximum desired and possible candidates.
|
803 |
|
804 |
@rtype: tuple
|
805 |
@return: tuple of (current, desired and possible)
|
806 |
|
807 |
"""
|
808 |
mc_now = mc_max = 0
|
809 |
for node in self._config_data.nodes.itervalues(): |
810 |
if not node.offline: |
811 |
mc_max += 1
|
812 |
if node.master_candidate:
|
813 |
mc_now += 1
|
814 |
mc_max = min(mc_max, self._config_data.cluster.candidate_pool_size) |
815 |
return (mc_now, mc_max)
|
816 |
|
817 |
@locking.ssynchronized(_config_lock, shared=1) |
818 |
def GetMasterCandidateStats(self): |
819 |
"""Get the number of current and maximum possible candidates.
|
820 |
|
821 |
This is just a wrapper over L{_UnlockedGetMasterCandidateStats}.
|
822 |
|
823 |
@rtype: tuple
|
824 |
@return: tuple of (current, max)
|
825 |
|
826 |
"""
|
827 |
return self._UnlockedGetMasterCandidateStats() |
828 |
|
829 |
@locking.ssynchronized(_config_lock)
|
830 |
def MaintainCandidatePool(self): |
831 |
"""Try to grow the candidate pool to the desired size.
|
832 |
|
833 |
@rtype: list
|
834 |
@return: list with the adjusted nodes (L{objects.Node} instances)
|
835 |
|
836 |
"""
|
837 |
mc_now, mc_max = self._UnlockedGetMasterCandidateStats()
|
838 |
mod_list = [] |
839 |
if mc_now < mc_max:
|
840 |
node_list = self._config_data.nodes.keys()
|
841 |
random.shuffle(node_list) |
842 |
for name in node_list: |
843 |
if mc_now >= mc_max:
|
844 |
break
|
845 |
node = self._config_data.nodes[name]
|
846 |
if node.master_candidate or node.offline: |
847 |
continue
|
848 |
mod_list.append(node) |
849 |
node.master_candidate = True
|
850 |
node.serial_no += 1
|
851 |
mc_now += 1
|
852 |
if mc_now != mc_max:
|
853 |
# this should not happen
|
854 |
logging.warning("Warning: MaintainCandidatePool didn't manage to"
|
855 |
" fill the candidate pool (%d/%d)", mc_now, mc_max)
|
856 |
if mod_list:
|
857 |
self._config_data.cluster.serial_no += 1 |
858 |
self._WriteConfig()
|
859 |
|
860 |
return mod_list
|
861 |
|
862 |
def _BumpSerialNo(self): |
863 |
"""Bump up the serial number of the config.
|
864 |
|
865 |
"""
|
866 |
self._config_data.serial_no += 1 |
867 |
|
868 |
def _OpenConfig(self): |
869 |
"""Read the config data from disk.
|
870 |
|
871 |
"""
|
872 |
f = open(self._cfg_file, 'r') |
873 |
try:
|
874 |
try:
|
875 |
data = objects.ConfigData.FromDict(serializer.Load(f.read())) |
876 |
except Exception, err: |
877 |
raise errors.ConfigurationError(err)
|
878 |
finally:
|
879 |
f.close() |
880 |
|
881 |
# Make sure the configuration has the right version
|
882 |
_ValidateConfig(data) |
883 |
|
884 |
if (not hasattr(data, 'cluster') or |
885 |
not hasattr(data.cluster, 'rsahostkeypub')): |
886 |
raise errors.ConfigurationError("Incomplete configuration" |
887 |
" (missing cluster.rsahostkeypub)")
|
888 |
self._config_data = data
|
889 |
# init the last serial as -1 so that the next write will cause
|
890 |
# ssconf update
|
891 |
self._last_cluster_serial = -1 |
892 |
|
893 |
def _DistributeConfig(self): |
894 |
"""Distribute the configuration to the other nodes.
|
895 |
|
896 |
Currently, this only copies the configuration file. In the future,
|
897 |
it could be used to encapsulate the 2/3-phase update mechanism.
|
898 |
|
899 |
"""
|
900 |
if self._offline: |
901 |
return True |
902 |
bad = False
|
903 |
|
904 |
node_list = [] |
905 |
addr_list = [] |
906 |
myhostname = self._my_hostname
|
907 |
# we can skip checking whether _UnlockedGetNodeInfo returns None
|
908 |
# since the node list comes from _UnlocketGetNodeList, and we are
|
909 |
# called with the lock held, so no modifications should take place
|
910 |
# in between
|
911 |
for node_name in self._UnlockedGetNodeList(): |
912 |
if node_name == myhostname:
|
913 |
continue
|
914 |
node_info = self._UnlockedGetNodeInfo(node_name)
|
915 |
if not node_info.master_candidate: |
916 |
continue
|
917 |
node_list.append(node_info.name) |
918 |
addr_list.append(node_info.primary_ip) |
919 |
|
920 |
result = rpc.RpcRunner.call_upload_file(node_list, self._cfg_file,
|
921 |
address_list=addr_list) |
922 |
for node in node_list: |
923 |
if not result[node]: |
924 |
logging.error("copy of file %s to node %s failed",
|
925 |
self._cfg_file, node)
|
926 |
bad = True
|
927 |
return not bad |
928 |
|
929 |
def _WriteConfig(self, destination=None): |
930 |
"""Write the configuration data to persistent storage.
|
931 |
|
932 |
"""
|
933 |
if destination is None: |
934 |
destination = self._cfg_file
|
935 |
self._BumpSerialNo()
|
936 |
txt = serializer.Dump(self._config_data.ToDict())
|
937 |
dir_name, file_name = os.path.split(destination) |
938 |
fd, name = tempfile.mkstemp('.newconfig', file_name, dir_name)
|
939 |
f = os.fdopen(fd, 'w')
|
940 |
try:
|
941 |
f.write(txt) |
942 |
os.fsync(f.fileno()) |
943 |
finally:
|
944 |
f.close() |
945 |
# we don't need to do os.close(fd) as f.close() did it
|
946 |
os.rename(name, destination) |
947 |
self.write_count += 1 |
948 |
|
949 |
# and redistribute the config file to master candidates
|
950 |
self._DistributeConfig()
|
951 |
|
952 |
# Write ssconf files on all nodes (including locally)
|
953 |
if self._last_cluster_serial < self._config_data.cluster.serial_no: |
954 |
if not self._offline: |
955 |
rpc.RpcRunner.call_write_ssconf_files(self._UnlockedGetNodeList(),
|
956 |
self._UnlockedGetSsconfValues())
|
957 |
self._last_cluster_serial = self._config_data.cluster.serial_no |
958 |
|
959 |
def _UnlockedGetSsconfValues(self): |
960 |
"""Return the values needed by ssconf.
|
961 |
|
962 |
@rtype: dict
|
963 |
@return: a dictionary with keys the ssconf names and values their
|
964 |
associated value
|
965 |
|
966 |
"""
|
967 |
fn = "\n".join
|
968 |
node_names = utils.NiceSort(self._UnlockedGetNodeList())
|
969 |
node_info = [self._UnlockedGetNodeInfo(name) for name in node_names] |
970 |
|
971 |
off_data = fn(node.name for node in node_info if node.offline) |
972 |
mc_data = fn(node.name for node in node_info if node.master_candidate) |
973 |
node_data = fn(node_names) |
974 |
|
975 |
cluster = self._config_data.cluster
|
976 |
return {
|
977 |
constants.SS_CLUSTER_NAME: cluster.cluster_name, |
978 |
constants.SS_FILE_STORAGE_DIR: cluster.file_storage_dir, |
979 |
constants.SS_MASTER_CANDIDATES: mc_data, |
980 |
constants.SS_MASTER_IP: cluster.master_ip, |
981 |
constants.SS_MASTER_NETDEV: cluster.master_netdev, |
982 |
constants.SS_MASTER_NODE: cluster.master_node, |
983 |
constants.SS_NODE_LIST: node_data, |
984 |
constants.SS_OFFLINE_NODES: off_data, |
985 |
} |
986 |
|
987 |
@locking.ssynchronized(_config_lock)
|
988 |
def InitConfig(self, version, cluster_config, master_node_config): |
989 |
"""Create the initial cluster configuration.
|
990 |
|
991 |
It will contain the current node, which will also be the master
|
992 |
node, and no instances.
|
993 |
|
994 |
@type version: int
|
995 |
@param version: Configuration version
|
996 |
@type cluster_config: objects.Cluster
|
997 |
@param cluster_config: Cluster configuration
|
998 |
@type master_node_config: objects.Node
|
999 |
@param master_node_config: Master node configuration
|
1000 |
|
1001 |
"""
|
1002 |
nodes = { |
1003 |
master_node_config.name: master_node_config, |
1004 |
} |
1005 |
|
1006 |
self._config_data = objects.ConfigData(version=version,
|
1007 |
cluster=cluster_config, |
1008 |
nodes=nodes, |
1009 |
instances={}, |
1010 |
serial_no=1)
|
1011 |
self._WriteConfig()
|
1012 |
|
1013 |
@locking.ssynchronized(_config_lock, shared=1) |
1014 |
def GetVGName(self): |
1015 |
"""Return the volume group name.
|
1016 |
|
1017 |
"""
|
1018 |
return self._config_data.cluster.volume_group_name |
1019 |
|
1020 |
@locking.ssynchronized(_config_lock)
|
1021 |
def SetVGName(self, vg_name): |
1022 |
"""Set the volume group name.
|
1023 |
|
1024 |
"""
|
1025 |
self._config_data.cluster.volume_group_name = vg_name
|
1026 |
self._config_data.cluster.serial_no += 1 |
1027 |
self._WriteConfig()
|
1028 |
|
1029 |
@locking.ssynchronized(_config_lock, shared=1) |
1030 |
def GetDefBridge(self): |
1031 |
"""Return the default bridge.
|
1032 |
|
1033 |
"""
|
1034 |
return self._config_data.cluster.default_bridge |
1035 |
|
1036 |
@locking.ssynchronized(_config_lock, shared=1) |
1037 |
def GetMACPrefix(self): |
1038 |
"""Return the mac prefix.
|
1039 |
|
1040 |
"""
|
1041 |
return self._config_data.cluster.mac_prefix |
1042 |
|
1043 |
@locking.ssynchronized(_config_lock, shared=1) |
1044 |
def GetClusterInfo(self): |
1045 |
"""Returns informations about the cluster
|
1046 |
|
1047 |
@rtype: L{objects.Cluster}
|
1048 |
@return: the cluster object
|
1049 |
|
1050 |
"""
|
1051 |
return self._config_data.cluster |
1052 |
|
1053 |
@locking.ssynchronized(_config_lock)
|
1054 |
def Update(self, target): |
1055 |
"""Notify function to be called after updates.
|
1056 |
|
1057 |
This function must be called when an object (as returned by
|
1058 |
GetInstanceInfo, GetNodeInfo, GetCluster) has been updated and the
|
1059 |
caller wants the modifications saved to the backing store. Note
|
1060 |
that all modified objects will be saved, but the target argument
|
1061 |
is the one the caller wants to ensure that it's saved.
|
1062 |
|
1063 |
@param target: an instance of either L{objects.Cluster},
|
1064 |
L{objects.Node} or L{objects.Instance} which is existing in
|
1065 |
the cluster
|
1066 |
|
1067 |
"""
|
1068 |
if self._config_data is None: |
1069 |
raise errors.ProgrammerError("Configuration file not read," |
1070 |
" cannot save.")
|
1071 |
update_serial = False
|
1072 |
if isinstance(target, objects.Cluster): |
1073 |
test = target == self._config_data.cluster
|
1074 |
elif isinstance(target, objects.Node): |
1075 |
test = target in self._config_data.nodes.values() |
1076 |
update_serial = True
|
1077 |
elif isinstance(target, objects.Instance): |
1078 |
test = target in self._config_data.instances.values() |
1079 |
else:
|
1080 |
raise errors.ProgrammerError("Invalid object type (%s) passed to" |
1081 |
" ConfigWriter.Update" % type(target)) |
1082 |
if not test: |
1083 |
raise errors.ConfigurationError("Configuration updated since object" |
1084 |
" has been read or unknown object")
|
1085 |
target.serial_no += 1
|
1086 |
|
1087 |
if update_serial:
|
1088 |
# for node updates, we need to increase the cluster serial too
|
1089 |
self._config_data.cluster.serial_no += 1 |
1090 |
|
1091 |
self._WriteConfig()
|