root / snf-cyclades-app / synnefo / logic / backend.py @ a868c831
History | View | Annotate | Download (47.7 kB)
1 |
# Copyright 2011-2013 GRNET S.A. All rights reserved.
|
---|---|
2 |
#
|
3 |
# Redistribution and use in source and binary forms, with or
|
4 |
# without modification, are permitted provided that the following
|
5 |
# conditions are met:
|
6 |
#
|
7 |
# 1. Redistributions of source code must retain the above
|
8 |
# copyright notice, this list of conditions and the following
|
9 |
# disclaimer.
|
10 |
#
|
11 |
# 2. Redistributions in binary form must reproduce the above
|
12 |
# copyright notice, this list of conditions and the following
|
13 |
# disclaimer in the documentation and/or other materials
|
14 |
# provided with the distribution.
|
15 |
#
|
16 |
# THIS SOFTWARE IS PROVIDED BY GRNET S.A. ``AS IS'' AND ANY EXPRESS
|
17 |
# OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
|
18 |
# WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
|
19 |
# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL GRNET S.A OR
|
20 |
# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
|
21 |
# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
|
22 |
# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF
|
23 |
# USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
|
24 |
# AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
25 |
# LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
|
26 |
# ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
|
27 |
# POSSIBILITY OF SUCH DAMAGE.
|
28 |
#
|
29 |
# The views and conclusions contained in the software and
|
30 |
# documentation are those of the authors and should not be
|
31 |
# interpreted as representing official policies, either expressed
|
32 |
# or implied, of GRNET S.A.
|
33 |
from django.conf import settings |
34 |
from django.db import transaction |
35 |
from datetime import datetime, timedelta |
36 |
|
37 |
from synnefo.db.models import (VirtualMachine, Network, |
38 |
BackendNetwork, BACKEND_STATUSES, |
39 |
pooled_rapi_client, VirtualMachineDiagnostic, |
40 |
Flavor, IPAddress, IPAddressLog) |
41 |
from synnefo.logic import utils, ips |
42 |
from synnefo import quotas |
43 |
from synnefo.api.util import release_resource |
44 |
from synnefo.util.mac2eui64 import mac2eui64 |
45 |
from synnefo.logic import rapi |
46 |
|
47 |
from logging import getLogger |
48 |
log = getLogger(__name__) |
49 |
|
50 |
|
51 |
_firewall_tags = { |
52 |
'ENABLED': settings.GANETI_FIREWALL_ENABLED_TAG,
|
53 |
'DISABLED': settings.GANETI_FIREWALL_DISABLED_TAG,
|
54 |
'PROTECTED': settings.GANETI_FIREWALL_PROTECTED_TAG}
|
55 |
|
56 |
_reverse_tags = dict((v.split(':')[3], k) for k, v in _firewall_tags.items()) |
57 |
|
58 |
SIMPLE_NIC_FIELDS = ["state", "mac", "network", "firewall_profile", "index"] |
59 |
COMPLEX_NIC_FIELDS = ["ipv4_address", "ipv6_address"] |
60 |
NIC_FIELDS = SIMPLE_NIC_FIELDS + COMPLEX_NIC_FIELDS |
61 |
DISK_FIELDS = ["status", "size", "index"] |
62 |
UNKNOWN_NIC_PREFIX = "unknown-"
|
63 |
UNKNOWN_DISK_PREFIX = "unknown-"
|
64 |
|
65 |
|
66 |
def handle_vm_quotas(vm, job_id, job_opcode, job_status, job_fields): |
67 |
"""Handle quotas for updated VirtualMachine.
|
68 |
|
69 |
Update quotas for the updated VirtualMachine based on the job that run on
|
70 |
the Ganeti backend. If a commission has been already issued for this job,
|
71 |
then this commission is just accepted or rejected based on the job status.
|
72 |
Otherwise, a new commission for the given change is issued, that is also in
|
73 |
force and auto-accept mode. In this case, previous commissions are
|
74 |
rejected, since they reflect a previous state of the VM.
|
75 |
|
76 |
"""
|
77 |
if job_status not in rapi.JOB_STATUS_FINALIZED: |
78 |
return vm
|
79 |
|
80 |
# Check successful completion of a job will trigger any quotable change in
|
81 |
# the VM state.
|
82 |
action = utils.get_action_from_opcode(job_opcode, job_fields) |
83 |
if action == "BUILD": |
84 |
# Quotas for new VMs are automatically accepted by the API
|
85 |
return vm
|
86 |
|
87 |
if vm.task_job_id == job_id and vm.serial is not None: |
88 |
# Commission for this change has already been issued. So just
|
89 |
# accept/reject it. Special case is OP_INSTANCE_CREATE, which even
|
90 |
# if fails, must be accepted, as the user must manually remove the
|
91 |
# failed server
|
92 |
serial = vm.serial |
93 |
if job_status == rapi.JOB_STATUS_SUCCESS:
|
94 |
quotas.accept_resource_serial(vm) |
95 |
elif job_status in [rapi.JOB_STATUS_ERROR, rapi.JOB_STATUS_CANCELED]: |
96 |
log.debug("Job %s failed. Rejecting related serial %s", job_id,
|
97 |
serial) |
98 |
quotas.reject_resource_serial(vm) |
99 |
elif job_status == rapi.JOB_STATUS_SUCCESS:
|
100 |
commission_info = quotas.get_commission_info(resource=vm, |
101 |
action=action, |
102 |
action_fields=job_fields) |
103 |
if commission_info is not None: |
104 |
# Commission for this change has not been issued, or the issued
|
105 |
# commission was unaware of the current change. Reject all previous
|
106 |
# commissions and create a new one in forced mode!
|
107 |
log.debug("Expected job was %s. Processing job %s. "
|
108 |
"Attached serial %s",
|
109 |
vm.task_job_id, job_id, vm.serial) |
110 |
reason = ("client: dispatcher, resource: %s, ganeti_job: %s"
|
111 |
% (vm, job_id)) |
112 |
serial = quotas.handle_resource_commission( |
113 |
vm, action, |
114 |
action_fields=job_fields, |
115 |
commission_name=reason, |
116 |
force=True,
|
117 |
auto_accept=True)
|
118 |
log.debug("Issued new commission: %s", serial)
|
119 |
return vm
|
120 |
|
121 |
|
122 |
@transaction.commit_on_success
|
123 |
def process_op_status(vm, etime, jobid, opcode, status, logmsg, nics=None, |
124 |
disks=None, job_fields=None): |
125 |
"""Process a job progress notification from the backend
|
126 |
|
127 |
Process an incoming message from the backend (currently Ganeti).
|
128 |
Job notifications with a terminating status (sucess, error, or canceled),
|
129 |
also update the operating state of the VM.
|
130 |
|
131 |
"""
|
132 |
# See #1492, #1031, #1111 why this line has been removed
|
133 |
#if (opcode not in [x[0] for x in VirtualMachine.BACKEND_OPCODES] or
|
134 |
if status not in [x[0] for x in BACKEND_STATUSES]: |
135 |
raise VirtualMachine.InvalidBackendMsgError(opcode, status)
|
136 |
|
137 |
vm.backendjobid = jobid |
138 |
vm.backendjobstatus = status |
139 |
vm.backendopcode = opcode |
140 |
vm.backendlogmsg = logmsg |
141 |
|
142 |
if status not in rapi.JOB_STATUS_FINALIZED: |
143 |
vm.save() |
144 |
return
|
145 |
|
146 |
if job_fields is None: |
147 |
job_fields = {} |
148 |
|
149 |
new_operstate = None
|
150 |
new_flavor = None
|
151 |
state_for_success = VirtualMachine.OPER_STATE_FROM_OPCODE.get(opcode) |
152 |
|
153 |
if status == rapi.JOB_STATUS_SUCCESS:
|
154 |
# If job succeeds, change operating state if needed
|
155 |
if state_for_success is not None: |
156 |
new_operstate = state_for_success |
157 |
|
158 |
beparams = job_fields.get("beparams", None) |
159 |
if beparams:
|
160 |
# Change the flavor of the VM
|
161 |
new_flavor = _process_resize(vm, beparams) |
162 |
|
163 |
# Update backendtime only for jobs that have been successfully
|
164 |
# completed, since only these jobs update the state of the VM. Else a
|
165 |
# "race condition" may occur when a successful job (e.g.
|
166 |
# OP_INSTANCE_REMOVE) completes before an error job and messages arrive
|
167 |
# in reversed order.
|
168 |
vm.backendtime = etime |
169 |
|
170 |
if status in rapi.JOB_STATUS_FINALIZED: |
171 |
if nics is not None: # Update the NICs of the VM |
172 |
_process_net_status(vm, etime, nics) |
173 |
if disks is not None: # Update the disks of the VM |
174 |
_process_disks_status(vm, etime, disks) |
175 |
|
176 |
# Special case: if OP_INSTANCE_CREATE fails --> ERROR
|
177 |
if opcode == 'OP_INSTANCE_CREATE' and status in (rapi.JOB_STATUS_CANCELED, |
178 |
rapi.JOB_STATUS_ERROR): |
179 |
new_operstate = "ERROR"
|
180 |
vm.backendtime = etime |
181 |
# Update state of associated attachments
|
182 |
vm.nics.all().update(state="ERROR")
|
183 |
vm.volumes.all().update(status="ERROR")
|
184 |
elif opcode == 'OP_INSTANCE_REMOVE': |
185 |
# Special case: OP_INSTANCE_REMOVE fails for machines in ERROR,
|
186 |
# when no instance exists at the Ganeti backend.
|
187 |
# See ticket #799 for all the details.
|
188 |
if (status == rapi.JOB_STATUS_SUCCESS or |
189 |
(status == rapi.JOB_STATUS_ERROR and not vm_exists_in_backend(vm))): |
190 |
# server has been deleted, so delete the server's attachments
|
191 |
vm.volumes.all().update(deleted=True, machine=None) |
192 |
for nic in vm.nics.all(): |
193 |
# but first release the IP
|
194 |
remove_nic_ips(nic) |
195 |
nic.delete() |
196 |
vm.deleted = True
|
197 |
new_operstate = state_for_success |
198 |
vm.backendtime = etime |
199 |
status = rapi.JOB_STATUS_SUCCESS |
200 |
|
201 |
if status in rapi.JOB_STATUS_FINALIZED: |
202 |
# Job is finalized: Handle quotas/commissioning
|
203 |
vm = handle_vm_quotas(vm, job_id=jobid, job_opcode=opcode, |
204 |
job_status=status, job_fields=job_fields) |
205 |
# and clear task fields
|
206 |
if vm.task_job_id == jobid:
|
207 |
vm.task = None
|
208 |
vm.task_job_id = None
|
209 |
|
210 |
if new_operstate is not None: |
211 |
vm.operstate = new_operstate |
212 |
if new_flavor is not None: |
213 |
vm.flavor = new_flavor |
214 |
|
215 |
vm.save() |
216 |
|
217 |
|
218 |
def _process_resize(vm, beparams): |
219 |
"""Change flavor of a VirtualMachine based on new beparams."""
|
220 |
old_flavor = vm.flavor |
221 |
vcpus = beparams.get("vcpus", old_flavor.cpu)
|
222 |
ram = beparams.get("maxmem", old_flavor.ram)
|
223 |
if vcpus == old_flavor.cpu and ram == old_flavor.ram: |
224 |
return
|
225 |
try:
|
226 |
new_flavor = Flavor.objects.get(cpu=vcpus, ram=ram, |
227 |
disk=old_flavor.disk, |
228 |
disk_template=old_flavor.disk_template) |
229 |
except Flavor.DoesNotExist:
|
230 |
raise Exception("Cannot find flavor for VM") |
231 |
return new_flavor
|
232 |
|
233 |
|
234 |
@transaction.commit_on_success
|
235 |
def process_net_status(vm, etime, nics): |
236 |
"""Wrap _process_net_status inside transaction."""
|
237 |
_process_net_status(vm, etime, nics) |
238 |
|
239 |
|
240 |
def _process_net_status(vm, etime, nics): |
241 |
"""Process a net status notification from the backend
|
242 |
|
243 |
Process an incoming message from the Ganeti backend,
|
244 |
detailing the NIC configuration of a VM instance.
|
245 |
|
246 |
Update the state of the VM in the DB accordingly.
|
247 |
|
248 |
"""
|
249 |
ganeti_nics = process_ganeti_nics(nics) |
250 |
db_nics = dict([(nic.id, nic)
|
251 |
for nic in vm.nics.select_related("network") |
252 |
.prefetch_related("ips")])
|
253 |
|
254 |
for nic_name in set(db_nics.keys()) | set(ganeti_nics.keys()): |
255 |
db_nic = db_nics.get(nic_name) |
256 |
ganeti_nic = ganeti_nics.get(nic_name) |
257 |
if ganeti_nic is None: |
258 |
if nic_is_stale(vm, nic):
|
259 |
log.debug("Removing stale NIC '%s'" % db_nic)
|
260 |
remove_nic_ips(db_nic) |
261 |
db_nic.delete() |
262 |
else:
|
263 |
log.info("NIC '%s' is still being created" % db_nic)
|
264 |
elif db_nic is None: |
265 |
msg = ("NIC/%s of VM %s does not exist in DB! Cannot automatically"
|
266 |
" fix this issue!" % (nic_name, vm))
|
267 |
log.error(msg) |
268 |
continue
|
269 |
elif not nics_are_equal(db_nic, ganeti_nic): |
270 |
for f in SIMPLE_NIC_FIELDS: |
271 |
# Update the NIC in DB with the values from Ganeti NIC
|
272 |
setattr(db_nic, f, ganeti_nic[f])
|
273 |
db_nic.save() |
274 |
|
275 |
# Special case where the IPv4 address has changed, because you
|
276 |
# need to release the old IPv4 address and reserve the new one
|
277 |
gnt_ipv4_address = ganeti_nic["ipv4_address"]
|
278 |
db_ipv4_address = db_nic.ipv4_address |
279 |
if db_ipv4_address != gnt_ipv4_address:
|
280 |
change_address_of_port(db_nic, vm.userid, |
281 |
old_address=db_ipv4_address, |
282 |
new_address=gnt_ipv4_address, |
283 |
version=4)
|
284 |
|
285 |
gnt_ipv6_address = ganeti_nic["ipv6_address"]
|
286 |
db_ipv6_address = db_nic.ipv6_address |
287 |
if db_ipv6_address != gnt_ipv6_address:
|
288 |
change_address_of_port(db_nic, vm.userid, |
289 |
old_address=db_ipv6_address, |
290 |
new_address=gnt_ipv6_address, |
291 |
version=6)
|
292 |
|
293 |
vm.backendtime = etime |
294 |
vm.save() |
295 |
|
296 |
|
297 |
def change_address_of_port(port, userid, old_address, new_address, version): |
298 |
"""Change."""
|
299 |
if old_address is not None: |
300 |
msg = ("IPv%s Address of server '%s' changed from '%s' to '%s'"
|
301 |
% (version, port.machine_id, old_address, new_address)) |
302 |
log.error(msg) |
303 |
|
304 |
# Remove the old IP address
|
305 |
remove_nic_ips(port, version=version) |
306 |
|
307 |
if version == 4: |
308 |
ipaddress = ips.allocate_ip(port.network, userid, address=new_address) |
309 |
ipaddress.nic = port |
310 |
ipaddress.save() |
311 |
elif version == 6: |
312 |
subnet6 = port.network.subnet6 |
313 |
ipaddress = IPAddress.objects.create(userid=userid, |
314 |
network=port.network, |
315 |
subnet=subnet6, |
316 |
nic=port, |
317 |
address=new_address, |
318 |
ipversion=6)
|
319 |
else:
|
320 |
raise ValueError("Unknown version: %s" % version) |
321 |
|
322 |
# New address log
|
323 |
ip_log = IPAddressLog.objects.create(server_id=port.machine_id, |
324 |
network_id=port.network_id, |
325 |
address=new_address, |
326 |
active=True)
|
327 |
log.info("Created IP log entry '%s' for address '%s' to server '%s'",
|
328 |
ip_log.id, new_address, port.machine_id) |
329 |
|
330 |
return ipaddress
|
331 |
|
332 |
|
333 |
def nics_are_equal(db_nic, gnt_nic): |
334 |
for field in NIC_FIELDS: |
335 |
if getattr(db_nic, field) != gnt_nic[field]: |
336 |
return False |
337 |
return True |
338 |
|
339 |
|
340 |
def process_ganeti_nics(ganeti_nics): |
341 |
"""Process NIC dict from ganeti"""
|
342 |
new_nics = [] |
343 |
for index, gnic in enumerate(ganeti_nics): |
344 |
nic_name = gnic.get("name", None) |
345 |
if nic_name is not None: |
346 |
nic_id = utils.id_from_nic_name(nic_name) |
347 |
else:
|
348 |
# Put as default value the index. If it is an unknown NIC to
|
349 |
# synnefo it will be created automaticaly.
|
350 |
nic_id = UNKNOWN_NIC_PREFIX + str(index)
|
351 |
network_name = gnic.get('network', '') |
352 |
network_id = utils.id_from_network_name(network_name) |
353 |
network = Network.objects.get(id=network_id) |
354 |
|
355 |
# Get the new nic info
|
356 |
mac = gnic.get('mac')
|
357 |
ipv4 = gnic.get('ip')
|
358 |
subnet6 = network.subnet6 |
359 |
ipv6 = mac2eui64(mac, subnet6.cidr) if subnet6 else None |
360 |
|
361 |
firewall = gnic.get('firewall')
|
362 |
firewall_profile = _reverse_tags.get(firewall) |
363 |
if not firewall_profile and network.public: |
364 |
firewall_profile = settings.DEFAULT_FIREWALL_PROFILE |
365 |
|
366 |
nic_info = { |
367 |
'index': index,
|
368 |
'network': network,
|
369 |
'mac': mac,
|
370 |
'ipv4_address': ipv4,
|
371 |
'ipv6_address': ipv6,
|
372 |
'firewall_profile': firewall_profile,
|
373 |
'state': 'ACTIVE'} |
374 |
|
375 |
new_nics.append((nic_id, nic_info)) |
376 |
return dict(new_nics) |
377 |
|
378 |
|
379 |
def remove_nic_ips(nic, version=None): |
380 |
"""Remove IP addresses associated with a NetworkInterface.
|
381 |
|
382 |
Remove all IP addresses that are associated with the NetworkInterface
|
383 |
object, by returning them to the pool and deleting the IPAddress object. If
|
384 |
the IP is a floating IP, then it is just disassociated from the NIC.
|
385 |
If version is specified, then only IP addressses of that version will be
|
386 |
removed.
|
387 |
|
388 |
"""
|
389 |
for ip in nic.ips.all(): |
390 |
if version and ip.ipversion != version: |
391 |
continue
|
392 |
|
393 |
# Update the DB table holding the logging of all IP addresses
|
394 |
terminate_active_ipaddress_log(nic, ip) |
395 |
|
396 |
if ip.floating_ip:
|
397 |
ip.nic = None
|
398 |
ip.save() |
399 |
else:
|
400 |
# Release the IPv4 address
|
401 |
ip.release_address() |
402 |
ip.delete() |
403 |
|
404 |
|
405 |
def terminate_active_ipaddress_log(nic, ip): |
406 |
"""Update DB logging entry for this IP address."""
|
407 |
if not ip.network.public or nic.machine is None: |
408 |
return
|
409 |
try:
|
410 |
ip_log, created = \ |
411 |
IPAddressLog.objects.get_or_create(server_id=nic.machine_id, |
412 |
network_id=ip.network_id, |
413 |
address=ip.address, |
414 |
active=True)
|
415 |
except IPAddressLog.MultipleObjectsReturned:
|
416 |
logmsg = ("Multiple active log entries for IP %s, Network %s,"
|
417 |
"Server %s. Cannot proceed!"
|
418 |
% (ip.address, ip.network, nic.machine)) |
419 |
log.error(logmsg) |
420 |
raise
|
421 |
|
422 |
if created:
|
423 |
logmsg = ("No log entry for IP %s, Network %s, Server %s. Created new"
|
424 |
" but with wrong creation timestamp."
|
425 |
% (ip.address, ip.network, nic.machine)) |
426 |
log.error(logmsg) |
427 |
ip_log.released_at = datetime.now() |
428 |
ip_log.active = False
|
429 |
ip_log.save() |
430 |
|
431 |
|
432 |
@transaction.commit_on_success
|
433 |
def process_disks_status(vm, etime, disks): |
434 |
"""Wrap _process_disks_status inside transaction."""
|
435 |
_process_disks_status(vm, etime, disks) |
436 |
|
437 |
|
438 |
def _process_disks_status(vm, etime, disks): |
439 |
"""Process a disks status notification from the backend
|
440 |
|
441 |
Process an incoming message from the Ganeti backend,
|
442 |
detailing the disk configuration of a VM instance.
|
443 |
|
444 |
Update the state of the VM in the DB accordingly.
|
445 |
|
446 |
"""
|
447 |
ganeti_disks = process_ganeti_disks(disks) |
448 |
db_disks = dict([(disk.id, disk)
|
449 |
for disk in vm.volumes.filter(deleted=False)]) |
450 |
|
451 |
for disk_name in set(db_disks.keys()) | set(ganeti_disks.keys()): |
452 |
db_disk = db_disks.get(disk_name) |
453 |
ganeti_disk = ganeti_disks.get(disk_name) |
454 |
if ganeti_disk is None: |
455 |
if disk_is_stale(vm, disk):
|
456 |
log.debug("Removing stale disk '%s'" % db_disk)
|
457 |
# TODO: Handle disk deletion
|
458 |
db_disk.deleted = True
|
459 |
db_disk.save() |
460 |
else:
|
461 |
log.info("disk '%s' is still being created" % db_disk)
|
462 |
elif db_disk is None: |
463 |
msg = ("disk/%s of VM %s does not exist in DB! Cannot"
|
464 |
" automatically fix this issue!" % (disk_name, vm))
|
465 |
log.error(msg) |
466 |
continue
|
467 |
elif not disks_are_equal(db_disk, ganeti_disk): |
468 |
for f in DISK_FIELDS: |
469 |
# Update the disk in DB with the values from Ganeti disk
|
470 |
setattr(db_disk, f, ganeti_disk[f])
|
471 |
db_disk.save() |
472 |
|
473 |
# TODO: Special case where the size of the disk has changed!!
|
474 |
assert(ganeti_disk["size"] == db_disk.size) |
475 |
|
476 |
vm.backendtime = etime |
477 |
vm.save() |
478 |
|
479 |
|
480 |
def disks_are_equal(db_disk, gnt_disk): |
481 |
for field in DISK_FIELDS: |
482 |
if getattr(db_disk, field) != gnt_disk[field]: |
483 |
return False |
484 |
return True |
485 |
|
486 |
|
487 |
def process_ganeti_disks(ganeti_disks): |
488 |
"""Process disk dict from ganeti"""
|
489 |
new_disks = [] |
490 |
for index, gdisk in enumerate(ganeti_disks): |
491 |
disk_name = gdisk.get("name", None) |
492 |
if disk_name is not None: |
493 |
disk_id = utils.id_from_disk_name(disk_name) |
494 |
else:
|
495 |
# Put as default value the index. If it is an unknown disk to
|
496 |
# synnefo it will be created automaticaly.
|
497 |
disk_id = UNKNOWN_DISK_PREFIX + str(index)
|
498 |
|
499 |
# Get disk size in GB
|
500 |
size = gdisk.get("size") >> 10 |
501 |
|
502 |
disk_info = { |
503 |
'index': index,
|
504 |
'size': size,
|
505 |
'status': "IN_USE"} |
506 |
|
507 |
new_disks.append((disk_id, disk_info)) |
508 |
return dict(new_disks) |
509 |
|
510 |
|
511 |
@transaction.commit_on_success
|
512 |
def process_network_status(back_network, etime, jobid, opcode, status, logmsg): |
513 |
if status not in [x[0] for x in BACKEND_STATUSES]: |
514 |
raise Network.InvalidBackendMsgError(opcode, status)
|
515 |
|
516 |
back_network.backendjobid = jobid |
517 |
back_network.backendjobstatus = status |
518 |
back_network.backendopcode = opcode |
519 |
back_network.backendlogmsg = logmsg |
520 |
|
521 |
# Note: Network is already locked!
|
522 |
network = back_network.network |
523 |
|
524 |
# Notifications of success change the operating state
|
525 |
state_for_success = BackendNetwork.OPER_STATE_FROM_OPCODE.get(opcode, None)
|
526 |
if status == rapi.JOB_STATUS_SUCCESS and state_for_success is not None: |
527 |
back_network.operstate = state_for_success |
528 |
|
529 |
if (status in (rapi.JOB_STATUS_CANCELED, rapi.JOB_STATUS_ERROR) |
530 |
and opcode == 'OP_NETWORK_ADD'): |
531 |
back_network.operstate = 'ERROR'
|
532 |
back_network.backendtime = etime |
533 |
|
534 |
if opcode == 'OP_NETWORK_REMOVE': |
535 |
network_is_deleted = (status == rapi.JOB_STATUS_SUCCESS) |
536 |
if network_is_deleted or (status == rapi.JOB_STATUS_ERROR and not |
537 |
network_exists_in_backend(back_network)): |
538 |
back_network.operstate = state_for_success |
539 |
back_network.deleted = True
|
540 |
back_network.backendtime = etime |
541 |
|
542 |
if status == rapi.JOB_STATUS_SUCCESS:
|
543 |
back_network.backendtime = etime |
544 |
back_network.save() |
545 |
# Also you must update the state of the Network!!
|
546 |
update_network_state(network) |
547 |
|
548 |
|
549 |
def update_network_state(network): |
550 |
"""Update the state of a Network based on BackendNetwork states.
|
551 |
|
552 |
Update the state of a Network based on the operstate of the networks in the
|
553 |
backends that network exists.
|
554 |
|
555 |
The state of the network is:
|
556 |
* ACTIVE: If it is 'ACTIVE' in at least one backend.
|
557 |
* DELETED: If it is is 'DELETED' in all backends that have been created.
|
558 |
|
559 |
This function also releases the resources (MAC prefix or Bridge) and the
|
560 |
quotas for the network.
|
561 |
|
562 |
"""
|
563 |
if network.deleted:
|
564 |
# Network has already been deleted. Just assert that state is also
|
565 |
# DELETED
|
566 |
if not network.state == "DELETED": |
567 |
network.state = "DELETED"
|
568 |
network.save() |
569 |
return
|
570 |
|
571 |
backend_states = [s.operstate for s in network.backend_networks.all()] |
572 |
if not backend_states and network.action != "DESTROY": |
573 |
if network.state != "ACTIVE": |
574 |
network.state = "ACTIVE"
|
575 |
network.save() |
576 |
return
|
577 |
|
578 |
# Network is deleted when all BackendNetworks go to "DELETED" operstate
|
579 |
deleted = reduce(lambda x, y: x == y and "DELETED", backend_states, |
580 |
"DELETED")
|
581 |
|
582 |
# Release the resources on the deletion of the Network
|
583 |
if deleted:
|
584 |
if network.ips.filter(deleted=False, floating_ip=True).exists(): |
585 |
msg = "Cannot delete network %s! Floating IPs still in use!"
|
586 |
log.error(msg % network) |
587 |
raise Exception(msg % network) |
588 |
log.info("Network %r deleted. Releasing link %r mac_prefix %r",
|
589 |
network.id, network.mac_prefix, network.link) |
590 |
network.deleted = True
|
591 |
network.state = "DELETED"
|
592 |
# Undrain the network, otherwise the network state will remain
|
593 |
# as 'SNF:DRAINED'
|
594 |
network.drained = False
|
595 |
if network.mac_prefix:
|
596 |
if network.FLAVORS[network.flavor]["mac_prefix"] == "pool": |
597 |
release_resource(res_type="mac_prefix",
|
598 |
value=network.mac_prefix) |
599 |
if network.link:
|
600 |
if network.FLAVORS[network.flavor]["link"] == "pool": |
601 |
release_resource(res_type="bridge", value=network.link)
|
602 |
|
603 |
# Set all subnets as deleted
|
604 |
network.subnets.update(deleted=True)
|
605 |
# And delete the IP pools
|
606 |
for subnet in network.subnets.all(): |
607 |
if subnet.ipversion == 4: |
608 |
subnet.ip_pools.all().delete() |
609 |
# And all the backend networks since there are useless
|
610 |
network.backend_networks.all().delete() |
611 |
|
612 |
# Issue commission
|
613 |
if network.userid:
|
614 |
quotas.issue_and_accept_commission(network, action="DESTROY")
|
615 |
# the above has already saved the object and committed;
|
616 |
# a second save would override others' changes, since the
|
617 |
# object is now unlocked
|
618 |
return
|
619 |
elif not network.public: |
620 |
log.warning("Network %s does not have an owner!", network.id)
|
621 |
network.save() |
622 |
|
623 |
|
624 |
@transaction.commit_on_success
|
625 |
def process_network_modify(back_network, etime, jobid, opcode, status, |
626 |
job_fields): |
627 |
assert (opcode == "OP_NETWORK_SET_PARAMS") |
628 |
if status not in [x[0] for x in BACKEND_STATUSES]: |
629 |
raise Network.InvalidBackendMsgError(opcode, status)
|
630 |
|
631 |
back_network.backendjobid = jobid |
632 |
back_network.backendjobstatus = status |
633 |
back_network.opcode = opcode |
634 |
|
635 |
add_reserved_ips = job_fields.get("add_reserved_ips")
|
636 |
if add_reserved_ips:
|
637 |
network = back_network.network |
638 |
for ip in add_reserved_ips: |
639 |
network.reserve_address(ip, external=True)
|
640 |
|
641 |
if status == rapi.JOB_STATUS_SUCCESS:
|
642 |
back_network.backendtime = etime |
643 |
back_network.save() |
644 |
|
645 |
|
646 |
@transaction.commit_on_success
|
647 |
def process_create_progress(vm, etime, progress): |
648 |
|
649 |
percentage = int(progress)
|
650 |
|
651 |
# The percentage may exceed 100%, due to the way
|
652 |
# snf-image:copy-progress tracks bytes read by image handling processes
|
653 |
percentage = 100 if percentage > 100 else percentage |
654 |
if percentage < 0: |
655 |
raise ValueError("Percentage cannot be negative") |
656 |
|
657 |
# FIXME: log a warning here, see #1033
|
658 |
# if last_update > percentage:
|
659 |
# raise ValueError("Build percentage should increase monotonically " \
|
660 |
# "(old = %d, new = %d)" % (last_update, percentage))
|
661 |
|
662 |
# This assumes that no message of type 'ganeti-create-progress' is going to
|
663 |
# arrive once OP_INSTANCE_CREATE has succeeded for a Ganeti instance and
|
664 |
# the instance is STARTED. What if the two messages are processed by two
|
665 |
# separate dispatcher threads, and the 'ganeti-op-status' message for
|
666 |
# successful creation gets processed before the 'ganeti-create-progress'
|
667 |
# message? [vkoukis]
|
668 |
#
|
669 |
#if not vm.operstate == 'BUILD':
|
670 |
# raise VirtualMachine.IllegalState("VM is not in building state")
|
671 |
|
672 |
vm.buildpercentage = percentage |
673 |
vm.backendtime = etime |
674 |
vm.save() |
675 |
|
676 |
|
677 |
@transaction.commit_on_success
|
678 |
def create_instance_diagnostic(vm, message, source, level="DEBUG", etime=None, |
679 |
details=None):
|
680 |
"""
|
681 |
Create virtual machine instance diagnostic entry.
|
682 |
|
683 |
:param vm: VirtualMachine instance to create diagnostic for.
|
684 |
:param message: Diagnostic message.
|
685 |
:param source: Diagnostic source identifier (e.g. image-helper).
|
686 |
:param level: Diagnostic level (`DEBUG`, `INFO`, `WARNING`, `ERROR`).
|
687 |
:param etime: The time the message occured (if available).
|
688 |
:param details: Additional details or debug information.
|
689 |
"""
|
690 |
VirtualMachineDiagnostic.objects.create_for_vm(vm, level, source=source, |
691 |
source_date=etime, |
692 |
message=message, |
693 |
details=details) |
694 |
|
695 |
|
696 |
def create_instance(vm, nics, volumes, flavor, image): |
697 |
"""`image` is a dictionary which should contain the keys:
|
698 |
'backend_id', 'format' and 'metadata'
|
699 |
|
700 |
metadata value should be a dictionary.
|
701 |
"""
|
702 |
|
703 |
# Handle arguments to CreateInstance() as a dictionary,
|
704 |
# initialize it based on a deployment-specific value.
|
705 |
# This enables the administrator to override deployment-specific
|
706 |
# arguments, such as the disk template to use, name of os provider
|
707 |
# and hypervisor-specific parameters at will (see Synnefo #785, #835).
|
708 |
#
|
709 |
kw = vm.backend.get_create_params() |
710 |
kw['mode'] = 'create' |
711 |
kw['name'] = vm.backend_vm_id
|
712 |
# Defined in settings.GANETI_CREATEINSTANCE_KWARGS
|
713 |
|
714 |
kw['disk_template'] = volumes[0].template |
715 |
disks = [] |
716 |
for volume in volumes: |
717 |
disk = {"name": volume.backend_volume_uuid,
|
718 |
"size": volume.size * 1024} |
719 |
provider = volume.provider |
720 |
if provider is not None: |
721 |
disk["provider"] = provider
|
722 |
disk["origin"] = volume.origin
|
723 |
extra_disk_params = settings.GANETI_DISK_PROVIDER_KWARGS\ |
724 |
.get(provider) |
725 |
if extra_disk_params is not None: |
726 |
disk.update(extra_disk_params) |
727 |
disks.append(disk) |
728 |
|
729 |
kw["disks"] = disks
|
730 |
|
731 |
kw['nics'] = [{"name": nic.backend_uuid, |
732 |
"network": nic.network.backend_id,
|
733 |
"ip": nic.ipv4_address}
|
734 |
for nic in nics] |
735 |
|
736 |
backend = vm.backend |
737 |
depend_jobs = [] |
738 |
for nic in nics: |
739 |
bnet, job_ids = ensure_network_is_active(backend, nic.network_id) |
740 |
depend_jobs.extend(job_ids) |
741 |
|
742 |
kw["depends"] = create_job_dependencies(depend_jobs)
|
743 |
|
744 |
# Defined in settings.GANETI_CREATEINSTANCE_KWARGS
|
745 |
# kw['os'] = settings.GANETI_OS_PROVIDER
|
746 |
kw['ip_check'] = False |
747 |
kw['name_check'] = False |
748 |
|
749 |
# Do not specific a node explicitly, have
|
750 |
# Ganeti use an iallocator instead
|
751 |
#kw['pnode'] = rapi.GetNodes()[0]
|
752 |
|
753 |
kw['dry_run'] = settings.TEST
|
754 |
|
755 |
kw['beparams'] = {
|
756 |
'auto_balance': True, |
757 |
'vcpus': flavor.cpu,
|
758 |
'memory': flavor.ram}
|
759 |
|
760 |
kw['osparams'] = {
|
761 |
'config_url': vm.config_url,
|
762 |
# Store image id and format to Ganeti
|
763 |
'img_id': image['backend_id'], |
764 |
'img_format': image['format']} |
765 |
|
766 |
# Use opportunistic locking
|
767 |
kw['opportunistic_locking'] = settings.GANETI_USE_OPPORTUNISTIC_LOCKING
|
768 |
|
769 |
# Defined in settings.GANETI_CREATEINSTANCE_KWARGS
|
770 |
# kw['hvparams'] = dict(serial_console=False)
|
771 |
|
772 |
log.debug("Creating instance %s", utils.hide_pass(kw))
|
773 |
with pooled_rapi_client(vm) as client: |
774 |
return client.CreateInstance(**kw)
|
775 |
|
776 |
|
777 |
def delete_instance(vm, shutdown_timeout=None): |
778 |
with pooled_rapi_client(vm) as client: |
779 |
return client.DeleteInstance(vm.backend_vm_id,
|
780 |
shutdown_timeout=shutdown_timeout, |
781 |
dry_run=settings.TEST) |
782 |
|
783 |
|
784 |
def reboot_instance(vm, reboot_type, shutdown_timeout=None): |
785 |
assert reboot_type in ('soft', 'hard') |
786 |
# Note that reboot type of Ganeti job must be always hard. The 'soft' and
|
787 |
# 'hard' type of OS API is different from the one in Ganeti, and maps to
|
788 |
# 'shutdown_timeout'.
|
789 |
kwargs = {"instance": vm.backend_vm_id,
|
790 |
"reboot_type": "hard"} |
791 |
# 'shutdown_timeout' parameter is only support from snf-ganeti>=2.8.2 and
|
792 |
# Ganeti > 2.10. In other versions this parameter will be ignored and
|
793 |
# we will fallback to default timeout of Ganeti (120s).
|
794 |
if shutdown_timeout is not None: |
795 |
kwargs["shutdown_timeout"] = shutdown_timeout
|
796 |
if reboot_type == "hard": |
797 |
kwargs["shutdown_timeout"] = 0 |
798 |
if settings.TEST:
|
799 |
kwargs["dry_run"] = True |
800 |
with pooled_rapi_client(vm) as client: |
801 |
return client.RebootInstance(**kwargs)
|
802 |
|
803 |
|
804 |
def startup_instance(vm): |
805 |
with pooled_rapi_client(vm) as client: |
806 |
return client.StartupInstance(vm.backend_vm_id, dry_run=settings.TEST)
|
807 |
|
808 |
|
809 |
def shutdown_instance(vm, shutdown_timeout=None): |
810 |
with pooled_rapi_client(vm) as client: |
811 |
return client.ShutdownInstance(vm.backend_vm_id,
|
812 |
timeout=shutdown_timeout, |
813 |
dry_run=settings.TEST) |
814 |
|
815 |
|
816 |
def resize_instance(vm, vcpus, memory): |
817 |
beparams = {"vcpus": int(vcpus), |
818 |
"minmem": int(memory), |
819 |
"maxmem": int(memory)} |
820 |
with pooled_rapi_client(vm) as client: |
821 |
return client.ModifyInstance(vm.backend_vm_id, beparams=beparams)
|
822 |
|
823 |
|
824 |
def get_instance_console(vm): |
825 |
# RAPI GetInstanceConsole() returns endpoints to the vnc_bind_address,
|
826 |
# which is a cluster-wide setting, either 0.0.0.0 or 127.0.0.1, and pretty
|
827 |
# useless (see #783).
|
828 |
#
|
829 |
# Until this is fixed on the Ganeti side, construct a console info reply
|
830 |
# directly.
|
831 |
#
|
832 |
# WARNING: This assumes that VNC runs on port network_port on
|
833 |
# the instance's primary node, and is probably
|
834 |
# hypervisor-specific.
|
835 |
#
|
836 |
log.debug("Getting console for vm %s", vm)
|
837 |
|
838 |
console = {} |
839 |
console['kind'] = 'vnc' |
840 |
|
841 |
with pooled_rapi_client(vm) as client: |
842 |
i = client.GetInstance(vm.backend_vm_id) |
843 |
|
844 |
if vm.backend.hypervisor == "kvm" and i['hvparams']['serial_console']: |
845 |
raise Exception("hv parameter serial_console cannot be true") |
846 |
console['host'] = i['pnode'] |
847 |
console['port'] = i['network_port'] |
848 |
|
849 |
return console
|
850 |
|
851 |
|
852 |
def get_instance_info(vm): |
853 |
with pooled_rapi_client(vm) as client: |
854 |
return client.GetInstance(vm.backend_vm_id)
|
855 |
|
856 |
|
857 |
def vm_exists_in_backend(vm): |
858 |
try:
|
859 |
get_instance_info(vm) |
860 |
return True |
861 |
except rapi.GanetiApiError as e: |
862 |
if e.code == 404: |
863 |
return False |
864 |
raise e
|
865 |
|
866 |
|
867 |
def get_network_info(backend_network): |
868 |
with pooled_rapi_client(backend_network) as client: |
869 |
return client.GetNetwork(backend_network.network.backend_id)
|
870 |
|
871 |
|
872 |
def network_exists_in_backend(backend_network): |
873 |
try:
|
874 |
get_network_info(backend_network) |
875 |
return True |
876 |
except rapi.GanetiApiError as e: |
877 |
if e.code == 404: |
878 |
return False |
879 |
|
880 |
|
881 |
def job_is_still_running(vm, job_id=None): |
882 |
with pooled_rapi_client(vm) as c: |
883 |
try:
|
884 |
if job_id is None: |
885 |
job_id = vm.backendjobid |
886 |
job_info = c.GetJobStatus(job_id) |
887 |
return not (job_info["status"] in rapi.JOB_STATUS_FINALIZED) |
888 |
except rapi.GanetiApiError:
|
889 |
return False |
890 |
|
891 |
|
892 |
def disk_is_stale(vm, disk, timeout=60): |
893 |
"""Check if a disk is stale or exists in the Ganeti backend."""
|
894 |
# First check the state of the disk
|
895 |
if disk.status == "CREATING": |
896 |
if datetime.now() < disk.created + timedelta(seconds=timeout):
|
897 |
# Do not check for too recent disks to avoid the time overhead
|
898 |
return False |
899 |
if job_is_still_running(vm, job_id=disk.backendjobid):
|
900 |
return False |
901 |
else:
|
902 |
# If job has finished, check that the disk exists, because the
|
903 |
# message may have been lost or stuck in the queue.
|
904 |
vm_info = get_instance_info(vm) |
905 |
if disk.backend_volume_uuid in vm_info["disk.names"]: |
906 |
return False |
907 |
return True |
908 |
|
909 |
|
910 |
def nic_is_stale(vm, nic, timeout=60): |
911 |
"""Check if a NIC is stale or exists in the Ganeti backend."""
|
912 |
# First check the state of the NIC and if there is a pending CONNECT
|
913 |
if nic.state == "BUILD" and vm.task == "CONNECT": |
914 |
if datetime.now() < nic.created + timedelta(seconds=timeout):
|
915 |
# Do not check for too recent NICs to avoid the time overhead
|
916 |
return False |
917 |
if job_is_still_running(vm, job_id=vm.task_job_id):
|
918 |
return False |
919 |
else:
|
920 |
# If job has finished, check that the NIC exists, because the
|
921 |
# message may have been lost or stuck in the queue.
|
922 |
vm_info = get_instance_info(vm) |
923 |
if nic.backend_uuid in vm_info["nic.names"]: |
924 |
return False |
925 |
return True |
926 |
|
927 |
|
928 |
def ensure_network_is_active(backend, network_id): |
929 |
"""Ensure that a network is active in the specified backend
|
930 |
|
931 |
Check that a network exists and is active in the specified backend. If not
|
932 |
(re-)create the network. Return the corresponding BackendNetwork object
|
933 |
and the IDs of the Ganeti job to create the network.
|
934 |
|
935 |
"""
|
936 |
job_ids = [] |
937 |
try:
|
938 |
bnet = BackendNetwork.objects.select_related("network")\
|
939 |
.get(backend=backend, network=network_id) |
940 |
if bnet.operstate != "ACTIVE": |
941 |
job_ids = create_network(bnet.network, backend, connect=True)
|
942 |
except BackendNetwork.DoesNotExist:
|
943 |
network = Network.objects.select_for_update().get(id=network_id) |
944 |
bnet = BackendNetwork.objects.create(backend=backend, network=network) |
945 |
job_ids = create_network(network, backend, connect=True)
|
946 |
|
947 |
return bnet, job_ids
|
948 |
|
949 |
|
950 |
def create_network(network, backend, connect=True): |
951 |
"""Create a network in a Ganeti backend"""
|
952 |
log.debug("Creating network %s in backend %s", network, backend)
|
953 |
|
954 |
job_id = _create_network(network, backend) |
955 |
|
956 |
if connect:
|
957 |
job_ids = connect_network(network, backend, depends=[job_id]) |
958 |
return job_ids
|
959 |
else:
|
960 |
return [job_id]
|
961 |
|
962 |
|
963 |
def _create_network(network, backend): |
964 |
"""Create a network."""
|
965 |
|
966 |
tags = network.backend_tag |
967 |
subnet = None
|
968 |
subnet6 = None
|
969 |
gateway = None
|
970 |
gateway6 = None
|
971 |
for _subnet in network.subnets.all(): |
972 |
if _subnet.dhcp and not "nfdhcpd" in tags: |
973 |
tags.append("nfdhcpd")
|
974 |
if _subnet.ipversion == 4: |
975 |
subnet = _subnet.cidr |
976 |
gateway = _subnet.gateway |
977 |
elif _subnet.ipversion == 6: |
978 |
subnet6 = _subnet.cidr |
979 |
gateway6 = _subnet.gateway |
980 |
|
981 |
conflicts_check = False
|
982 |
if network.public:
|
983 |
tags.append('public')
|
984 |
if subnet is not None: |
985 |
conflicts_check = True
|
986 |
else:
|
987 |
tags.append('private')
|
988 |
|
989 |
# Use a dummy network subnet for IPv6 only networks. Currently Ganeti does
|
990 |
# not support IPv6 only networks. To bypass this limitation, we create the
|
991 |
# network with a dummy network subnet, and make Cyclades connect instances
|
992 |
# to such networks, with address=None.
|
993 |
if subnet is None: |
994 |
subnet = "10.0.0.0/29"
|
995 |
|
996 |
try:
|
997 |
bn = BackendNetwork.objects.get(network=network, backend=backend) |
998 |
mac_prefix = bn.mac_prefix |
999 |
except BackendNetwork.DoesNotExist:
|
1000 |
raise Exception("BackendNetwork for network '%s' in backend '%s'" |
1001 |
" does not exist" % (network.id, backend.id))
|
1002 |
|
1003 |
with pooled_rapi_client(backend) as client: |
1004 |
return client.CreateNetwork(network_name=network.backend_id,
|
1005 |
network=subnet, |
1006 |
network6=subnet6, |
1007 |
gateway=gateway, |
1008 |
gateway6=gateway6, |
1009 |
mac_prefix=mac_prefix, |
1010 |
conflicts_check=conflicts_check, |
1011 |
tags=tags) |
1012 |
|
1013 |
|
1014 |
def connect_network(network, backend, depends=[], group=None): |
1015 |
"""Connect a network to nodegroups."""
|
1016 |
log.debug("Connecting network %s to backend %s", network, backend)
|
1017 |
|
1018 |
conflicts_check = False
|
1019 |
if network.public and (network.subnet4 is not None): |
1020 |
conflicts_check = True
|
1021 |
|
1022 |
depends = create_job_dependencies(depends) |
1023 |
with pooled_rapi_client(backend) as client: |
1024 |
groups = [group] if group is not None else client.GetGroups() |
1025 |
job_ids = [] |
1026 |
for group in groups: |
1027 |
job_id = client.ConnectNetwork(network.backend_id, group, |
1028 |
network.mode, network.link, |
1029 |
conflicts_check, |
1030 |
depends=depends) |
1031 |
job_ids.append(job_id) |
1032 |
return job_ids
|
1033 |
|
1034 |
|
1035 |
def delete_network(network, backend, disconnect=True): |
1036 |
log.debug("Deleting network %s from backend %s", network, backend)
|
1037 |
|
1038 |
depends = [] |
1039 |
if disconnect:
|
1040 |
depends = disconnect_network(network, backend) |
1041 |
_delete_network(network, backend, depends=depends) |
1042 |
|
1043 |
|
1044 |
def _delete_network(network, backend, depends=[]): |
1045 |
depends = create_job_dependencies(depends) |
1046 |
with pooled_rapi_client(backend) as client: |
1047 |
return client.DeleteNetwork(network.backend_id, depends)
|
1048 |
|
1049 |
|
1050 |
def disconnect_network(network, backend, group=None): |
1051 |
log.debug("Disconnecting network %s to backend %s", network, backend)
|
1052 |
|
1053 |
with pooled_rapi_client(backend) as client: |
1054 |
groups = [group] if group is not None else client.GetGroups() |
1055 |
job_ids = [] |
1056 |
for group in groups: |
1057 |
job_id = client.DisconnectNetwork(network.backend_id, group) |
1058 |
job_ids.append(job_id) |
1059 |
return job_ids
|
1060 |
|
1061 |
|
1062 |
def connect_to_network(vm, nic): |
1063 |
network = nic.network |
1064 |
backend = vm.backend |
1065 |
bnet, depend_jobs = ensure_network_is_active(backend, network.id) |
1066 |
|
1067 |
depends = create_job_dependencies(depend_jobs) |
1068 |
|
1069 |
nic = {'name': nic.backend_uuid,
|
1070 |
'network': network.backend_id,
|
1071 |
'ip': nic.ipv4_address}
|
1072 |
|
1073 |
log.debug("Adding NIC %s to VM %s", nic, vm)
|
1074 |
|
1075 |
kwargs = { |
1076 |
"instance": vm.backend_vm_id,
|
1077 |
"nics": [("add", "-1", nic)], |
1078 |
"depends": depends,
|
1079 |
} |
1080 |
if vm.backend.use_hotplug():
|
1081 |
kwargs["hotplug_if_possible"] = True |
1082 |
if settings.TEST:
|
1083 |
kwargs["dry_run"] = True |
1084 |
|
1085 |
with pooled_rapi_client(vm) as client: |
1086 |
return client.ModifyInstance(**kwargs)
|
1087 |
|
1088 |
|
1089 |
def disconnect_from_network(vm, nic): |
1090 |
log.debug("Removing NIC %s of VM %s", nic, vm)
|
1091 |
|
1092 |
kwargs = { |
1093 |
"instance": vm.backend_vm_id,
|
1094 |
"nics": [("remove", nic.backend_uuid, {})], |
1095 |
} |
1096 |
if vm.backend.use_hotplug():
|
1097 |
kwargs["hotplug_if_possible"] = True |
1098 |
if settings.TEST:
|
1099 |
kwargs["dry_run"] = True |
1100 |
|
1101 |
with pooled_rapi_client(vm) as client: |
1102 |
jobID = client.ModifyInstance(**kwargs) |
1103 |
firewall_profile = nic.firewall_profile |
1104 |
if firewall_profile and firewall_profile != "DISABLED": |
1105 |
tag = _firewall_tags[firewall_profile] % nic.backend_uuid |
1106 |
client.DeleteInstanceTags(vm.backend_vm_id, [tag], |
1107 |
dry_run=settings.TEST) |
1108 |
|
1109 |
return jobID
|
1110 |
|
1111 |
|
1112 |
def set_firewall_profile(vm, profile, nic): |
1113 |
uuid = nic.backend_uuid |
1114 |
try:
|
1115 |
tag = _firewall_tags[profile] % uuid |
1116 |
except KeyError: |
1117 |
raise ValueError("Unsopported Firewall Profile: %s" % profile) |
1118 |
|
1119 |
log.debug("Setting tag of VM %s, NIC %s, to %s", vm, nic, profile)
|
1120 |
|
1121 |
with pooled_rapi_client(vm) as client: |
1122 |
# Delete previous firewall tags
|
1123 |
old_tags = client.GetInstanceTags(vm.backend_vm_id) |
1124 |
delete_tags = [(t % uuid) for t in _firewall_tags.values() |
1125 |
if (t % uuid) in old_tags] |
1126 |
if delete_tags:
|
1127 |
client.DeleteInstanceTags(vm.backend_vm_id, delete_tags, |
1128 |
dry_run=settings.TEST) |
1129 |
|
1130 |
if profile != "DISABLED": |
1131 |
client.AddInstanceTags(vm.backend_vm_id, [tag], |
1132 |
dry_run=settings.TEST) |
1133 |
|
1134 |
# XXX NOP ModifyInstance call to force process_net_status to run
|
1135 |
# on the dispatcher
|
1136 |
os_name = settings.GANETI_CREATEINSTANCE_KWARGS['os']
|
1137 |
client.ModifyInstance(vm.backend_vm_id, |
1138 |
os_name=os_name) |
1139 |
return None |
1140 |
|
1141 |
|
1142 |
def attach_volume(vm, volume, depends=[]): |
1143 |
log.debug("Attaching volume %s to vm %s", volume, vm)
|
1144 |
|
1145 |
disk = {"size": int(volume.size) << 10, |
1146 |
"name": volume.backend_volume_uuid,
|
1147 |
"volume_name": volume.backend_volume_uuid}
|
1148 |
|
1149 |
disk_provider = volume.provider |
1150 |
if disk_provider is not None: |
1151 |
disk["provider"] = disk_provider
|
1152 |
|
1153 |
if volume.origin is not None: |
1154 |
disk["origin"] = volume.origin
|
1155 |
|
1156 |
kwargs = { |
1157 |
"instance": vm.backend_vm_id,
|
1158 |
"disks": [("add", "-1", disk)], |
1159 |
"depends": depends,
|
1160 |
} |
1161 |
if vm.backend.use_hotplug():
|
1162 |
kwargs["hotplug_if_possible"] = True |
1163 |
if settings.TEST:
|
1164 |
kwargs["dry_run"] = True |
1165 |
|
1166 |
with pooled_rapi_client(vm) as client: |
1167 |
return client.ModifyInstance(**kwargs)
|
1168 |
|
1169 |
|
1170 |
def detach_volume(vm, volume, depends=[]): |
1171 |
log.debug("Removing volume %s from vm %s", volume, vm)
|
1172 |
kwargs = { |
1173 |
"instance": vm.backend_vm_id,
|
1174 |
"disks": [("remove", volume.backend_volume_uuid, {})], |
1175 |
"depends": depends,
|
1176 |
} |
1177 |
if vm.backend.use_hotplug():
|
1178 |
kwargs["hotplug_if_possible"] = True |
1179 |
if settings.TEST:
|
1180 |
kwargs["dry_run"] = True |
1181 |
|
1182 |
with pooled_rapi_client(vm) as client: |
1183 |
return client.ModifyInstance(**kwargs)
|
1184 |
|
1185 |
|
1186 |
def snapshot_instance(vm, snapshot_name): |
1187 |
#volume = instance.volumes.all()[0]
|
1188 |
with pooled_rapi_client(vm) as client: |
1189 |
return client.SnapshotInstance(instance=vm.backend_vm_id,
|
1190 |
snapshot_name=snapshot_name) |
1191 |
|
1192 |
|
1193 |
def get_instances(backend, bulk=True): |
1194 |
with pooled_rapi_client(backend) as c: |
1195 |
return c.GetInstances(bulk=bulk)
|
1196 |
|
1197 |
|
1198 |
def get_nodes(backend, bulk=True): |
1199 |
with pooled_rapi_client(backend) as c: |
1200 |
return c.GetNodes(bulk=bulk)
|
1201 |
|
1202 |
|
1203 |
def get_jobs(backend, bulk=True): |
1204 |
with pooled_rapi_client(backend) as c: |
1205 |
return c.GetJobs(bulk=bulk)
|
1206 |
|
1207 |
|
1208 |
def get_physical_resources(backend): |
1209 |
""" Get the physical resources of a backend.
|
1210 |
|
1211 |
Get the resources of a backend as reported by the backend (not the db).
|
1212 |
|
1213 |
"""
|
1214 |
nodes = get_nodes(backend, bulk=True)
|
1215 |
attr = ['mfree', 'mtotal', 'dfree', 'dtotal', 'pinst_cnt', 'ctotal'] |
1216 |
res = {} |
1217 |
for a in attr: |
1218 |
res[a] = 0
|
1219 |
for n in nodes: |
1220 |
# Filter out drained, offline and not vm_capable nodes since they will
|
1221 |
# not take part in the vm allocation process
|
1222 |
can_host_vms = n['vm_capable'] and not (n['drained'] or n['offline']) |
1223 |
if can_host_vms and n['cnodes']: |
1224 |
for a in attr: |
1225 |
res[a] += int(n[a] or 0) |
1226 |
return res
|
1227 |
|
1228 |
|
1229 |
def update_backend_resources(backend, resources=None): |
1230 |
""" Update the state of the backend resources in db.
|
1231 |
|
1232 |
"""
|
1233 |
|
1234 |
if not resources: |
1235 |
resources = get_physical_resources(backend) |
1236 |
|
1237 |
backend.mfree = resources['mfree']
|
1238 |
backend.mtotal = resources['mtotal']
|
1239 |
backend.dfree = resources['dfree']
|
1240 |
backend.dtotal = resources['dtotal']
|
1241 |
backend.pinst_cnt = resources['pinst_cnt']
|
1242 |
backend.ctotal = resources['ctotal']
|
1243 |
backend.updated = datetime.now() |
1244 |
backend.save() |
1245 |
|
1246 |
|
1247 |
def get_memory_from_instances(backend): |
1248 |
""" Get the memory that is used from instances.
|
1249 |
|
1250 |
Get the used memory of a backend. Note: This is different for
|
1251 |
the real memory used, due to kvm's memory de-duplication.
|
1252 |
|
1253 |
"""
|
1254 |
with pooled_rapi_client(backend) as client: |
1255 |
instances = client.GetInstances(bulk=True)
|
1256 |
mem = 0
|
1257 |
for i in instances: |
1258 |
mem += i['oper_ram']
|
1259 |
return mem
|
1260 |
|
1261 |
|
1262 |
def get_available_disk_templates(backend): |
1263 |
"""Get the list of available disk templates of a Ganeti backend.
|
1264 |
|
1265 |
The list contains the disk templates that are enabled in the Ganeti backend
|
1266 |
and also included in ipolicy-disk-templates.
|
1267 |
|
1268 |
"""
|
1269 |
with pooled_rapi_client(backend) as c: |
1270 |
info = c.GetInfo() |
1271 |
ipolicy_disk_templates = info["ipolicy"]["disk-templates"] |
1272 |
try:
|
1273 |
enabled_disk_templates = info["enabled_disk_templates"]
|
1274 |
return [dp for dp in enabled_disk_templates |
1275 |
if dp in ipolicy_disk_templates] |
1276 |
except KeyError: |
1277 |
# Ganeti < 2.8 does not have 'enabled_disk_templates'
|
1278 |
return ipolicy_disk_templates
|
1279 |
|
1280 |
|
1281 |
def update_backend_disk_templates(backend): |
1282 |
disk_templates = get_available_disk_templates(backend) |
1283 |
backend.disk_templates = disk_templates |
1284 |
backend.save() |
1285 |
|
1286 |
|
1287 |
##
|
1288 |
## Synchronized operations for reconciliation
|
1289 |
##
|
1290 |
|
1291 |
|
1292 |
def create_network_synced(network, backend): |
1293 |
result = _create_network_synced(network, backend) |
1294 |
if result[0] != rapi.JOB_STATUS_SUCCESS: |
1295 |
return result
|
1296 |
result = connect_network_synced(network, backend) |
1297 |
return result
|
1298 |
|
1299 |
|
1300 |
def _create_network_synced(network, backend): |
1301 |
with pooled_rapi_client(backend) as client: |
1302 |
job = _create_network(network, backend) |
1303 |
result = wait_for_job(client, job) |
1304 |
return result
|
1305 |
|
1306 |
|
1307 |
def connect_network_synced(network, backend): |
1308 |
with pooled_rapi_client(backend) as client: |
1309 |
for group in client.GetGroups(): |
1310 |
job = client.ConnectNetwork(network.backend_id, group, |
1311 |
network.mode, network.link) |
1312 |
result = wait_for_job(client, job) |
1313 |
if result[0] != rapi.JOB_STATUS_SUCCESS: |
1314 |
return result
|
1315 |
|
1316 |
return result
|
1317 |
|
1318 |
|
1319 |
def wait_for_job(client, jobid): |
1320 |
result = client.WaitForJobChange(jobid, ['status', 'opresult'], None, None) |
1321 |
status = result['job_info'][0] |
1322 |
while status not in rapi.JOB_STATUS_FINALIZED: |
1323 |
result = client.WaitForJobChange(jobid, ['status', 'opresult'], |
1324 |
[result], None)
|
1325 |
status = result['job_info'][0] |
1326 |
|
1327 |
if status == rapi.JOB_STATUS_SUCCESS:
|
1328 |
return (status, None) |
1329 |
else:
|
1330 |
error = result['job_info'][1] |
1331 |
return (status, error)
|
1332 |
|
1333 |
|
1334 |
def create_job_dependencies(job_ids=[], job_states=None): |
1335 |
"""Transform a list of job IDs to Ganeti 'depends' attribute."""
|
1336 |
if job_states is None: |
1337 |
job_states = list(rapi.JOB_STATUS_FINALIZED)
|
1338 |
assert(type(job_states) == list) |
1339 |
return [[job_id, job_states] for job_id in job_ids] |