4 # Copyright (C) 2006, 2007 Google Inc.
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.
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.
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
22 """Ganeti node daemon"""
24 # functions in this module need to have a given name structure, so:
25 # pylint: disable-msg=C0103
35 from optparse import OptionParser
37 from ganeti import backend
38 from ganeti import constants
39 from ganeti import objects
40 from ganeti import errors
41 from ganeti import jstore
42 from ganeti import daemon
43 from ganeti import http
44 from ganeti import utils
46 import ganeti.http.server
52 def _RequireJobQueueLock(fn):
53 """Decorator for job queue manipulating functions.
56 QUEUE_LOCK_TIMEOUT = 10
58 def wrapper(*args, **kwargs):
59 # Locking in exclusive, blocking mode because there could be several
60 # children running at the same time. Waiting up to 10 seconds.
61 queue_lock.Exclusive(blocking=True, timeout=QUEUE_LOCK_TIMEOUT)
63 return fn(*args, **kwargs)
70 class NodeHttpServer(http.server.HttpServer):
71 """The server implementation.
73 This class holds all methods exposed over the RPC interface.
76 def __init__(self, *args, **kwargs):
77 http.server.HttpServer.__init__(self, *args, **kwargs)
78 self.noded_pid = os.getpid()
80 def HandleRequest(self, req):
84 if req.request_method.upper() != http.HTTP_PUT:
85 raise http.HttpBadRequest()
87 path = req.request_path
88 if path.startswith("/"):
91 method = getattr(self, "perspective_%s" % path, None)
93 raise http.HttpNotFound()
97 return method(req.request_body)
99 logging.exception("Error in RPC call")
101 except errors.QuitGanetiException, err:
102 # Tell parent to quit
103 os.kill(self.noded_pid, signal.SIGTERM)
105 # the new block devices --------------------------
108 def perspective_blockdev_create(params):
109 """Create a block device.
112 bdev_s, size, owner, on_primary, info = params
113 bdev = objects.Disk.FromDict(bdev_s)
115 raise ValueError("can't unserialize data!")
116 return backend.BlockdevCreate(bdev, size, owner, on_primary, info)
119 def perspective_blockdev_remove(params):
120 """Remove a block device.
124 bdev = objects.Disk.FromDict(bdev_s)
125 return backend.BlockdevRemove(bdev)
128 def perspective_blockdev_rename(params):
129 """Remove a block device.
132 devlist = [(objects.Disk.FromDict(ds), uid) for ds, uid in params]
133 return backend.BlockdevRename(devlist)
136 def perspective_blockdev_assemble(params):
137 """Assemble a block device.
140 bdev_s, owner, on_primary = params
141 bdev = objects.Disk.FromDict(bdev_s)
143 raise ValueError("can't unserialize data!")
144 return backend.BlockdevAssemble(bdev, owner, on_primary)
147 def perspective_blockdev_shutdown(params):
148 """Shutdown a block device.
152 bdev = objects.Disk.FromDict(bdev_s)
154 raise ValueError("can't unserialize data!")
155 return backend.BlockdevShutdown(bdev)
158 def perspective_blockdev_addchildren(params):
159 """Add a child to a mirror device.
161 Note: this is only valid for mirror devices. It's the caller's duty
162 to send a correct disk, otherwise we raise an error.
165 bdev_s, ndev_s = params
166 bdev = objects.Disk.FromDict(bdev_s)
167 ndevs = [objects.Disk.FromDict(disk_s) for disk_s in ndev_s]
168 if bdev is None or ndevs.count(None) > 0:
169 raise ValueError("can't unserialize data!")
170 return backend.BlockdevAddchildren(bdev, ndevs)
173 def perspective_blockdev_removechildren(params):
174 """Remove a child from a mirror device.
176 This is only valid for mirror devices, of course. It's the callers
177 duty to send a correct disk, otherwise we raise an error.
180 bdev_s, ndev_s = params
181 bdev = objects.Disk.FromDict(bdev_s)
182 ndevs = [objects.Disk.FromDict(disk_s) for disk_s in ndev_s]
183 if bdev is None or ndevs.count(None) > 0:
184 raise ValueError("can't unserialize data!")
185 return backend.BlockdevRemovechildren(bdev, ndevs)
188 def perspective_blockdev_getmirrorstatus(params):
189 """Return the mirror status for a list of disks.
192 disks = [objects.Disk.FromDict(dsk_s)
194 return backend.BlockdevGetmirrorstatus(disks)
197 def perspective_blockdev_find(params):
198 """Expose the FindBlockDevice functionality for a disk.
200 This will try to find but not activate a disk.
203 disk = objects.Disk.FromDict(params[0])
204 return backend.BlockdevFind(disk)
207 def perspective_blockdev_snapshot(params):
208 """Create a snapshot device.
210 Note that this is only valid for LVM disks, if we get passed
211 something else we raise an exception. The snapshot device can be
212 remove by calling the generic block device remove call.
215 cfbd = objects.Disk.FromDict(params[0])
216 return backend.BlockdevSnapshot(cfbd)
219 def perspective_blockdev_grow(params):
220 """Grow a stack of devices.
223 cfbd = objects.Disk.FromDict(params[0])
225 return backend.BlockdevGrow(cfbd, amount)
228 def perspective_blockdev_close(params):
229 """Closes the given block devices.
232 disks = [objects.Disk.FromDict(cf) for cf in params[1]]
233 return backend.BlockdevClose(params[0], disks)
235 # blockdev/drbd specific methods ----------
238 def perspective_drbd_disconnect_net(params):
239 """Disconnects the network connection of drbd disks.
241 Note that this is only valid for drbd disks, so the members of the
242 disk list must all be drbd devices.
245 nodes_ip, disks = params
246 disks = [objects.Disk.FromDict(cf) for cf in disks]
247 return backend.DrbdDisconnectNet(nodes_ip, disks)
250 def perspective_drbd_attach_net(params):
251 """Attaches the network connection of drbd disks.
253 Note that this is only valid for drbd disks, so the members of the
254 disk list must all be drbd devices.
257 nodes_ip, disks, instance_name, multimaster = params
258 disks = [objects.Disk.FromDict(cf) for cf in disks]
259 return backend.DrbdAttachNet(nodes_ip, disks,
260 instance_name, multimaster)
263 def perspective_drbd_wait_sync(params):
264 """Wait until DRBD disks are synched.
266 Note that this is only valid for drbd disks, so the members of the
267 disk list must all be drbd devices.
270 nodes_ip, disks = params
271 disks = [objects.Disk.FromDict(cf) for cf in disks]
272 return backend.DrbdWaitSync(nodes_ip, disks)
274 # export/import --------------------------
277 def perspective_snapshot_export(params):
278 """Export a given snapshot.
281 disk = objects.Disk.FromDict(params[0])
282 dest_node = params[1]
283 instance = objects.Instance.FromDict(params[2])
284 cluster_name = params[3]
286 return backend.ExportSnapshot(disk, dest_node, instance,
287 cluster_name, dev_idx)
290 def perspective_finalize_export(params):
291 """Expose the finalize export functionality.
294 instance = objects.Instance.FromDict(params[0])
295 snap_disks = [objects.Disk.FromDict(str_data)
296 for str_data in params[1]]
297 return backend.FinalizeExport(instance, snap_disks)
300 def perspective_export_info(params):
301 """Query information about an existing export on this node.
303 The given path may not contain an export, in which case we return
308 einfo = backend.ExportInfo(path)
314 def perspective_export_list(params):
315 """List the available exports on this node.
317 Note that as opposed to export_info, which may query data about an
318 export in any path, this only queries the standard Ganeti path
319 (constants.EXPORT_DIR).
322 return backend.ListExports()
325 def perspective_export_remove(params):
330 return backend.RemoveExport(export)
332 # volume --------------------------
335 def perspective_volume_list(params):
336 """Query the list of logical volumes in a given volume group.
340 return backend.GetVolumeList(vgname)
343 def perspective_vg_list(params):
344 """Query the list of volume groups.
347 return backend.ListVolumeGroups()
349 # bridge --------------------------
352 def perspective_bridges_exist(params):
353 """Check if all bridges given exist on this node.
356 bridges_list = params[0]
357 return backend.BridgesExist(bridges_list)
359 # instance --------------------------
362 def perspective_instance_os_add(params):
363 """Install an OS on a given instance.
367 inst = objects.Instance.FromDict(inst_s)
368 return backend.InstanceOsAdd(inst)
371 def perspective_instance_run_rename(params):
372 """Runs the OS rename script for an instance.
375 inst_s, old_name = params
376 inst = objects.Instance.FromDict(inst_s)
377 return backend.RunRenameInstance(inst, old_name)
380 def perspective_instance_os_import(params):
381 """Run the import function of an OS onto a given instance.
384 inst_s, src_node, src_images, cluster_name = params
385 inst = objects.Instance.FromDict(inst_s)
386 return backend.ImportOSIntoInstance(inst, src_node, src_images,
390 def perspective_instance_shutdown(params):
391 """Shutdown an instance.
394 instance = objects.Instance.FromDict(params[0])
395 return backend.InstanceShutdown(instance)
398 def perspective_instance_start(params):
399 """Start an instance.
402 instance = objects.Instance.FromDict(params[0])
403 return backend.StartInstance(instance)
406 def perspective_migration_info(params):
407 """Gather information about an instance to be migrated.
410 instance = objects.Instance.FromDict(params[0])
411 return backend.MigrationInfo(instance)
414 def perspective_accept_instance(params):
415 """Prepare the node to accept an instance.
418 instance, info, target = params
419 instance = objects.Instance.FromDict(instance)
420 return backend.AcceptInstance(instance, info, target)
423 def perspective_finalize_migration(params):
424 """Finalize the instance migration.
427 instance, info, success = params
428 instance = objects.Instance.FromDict(instance)
429 return backend.FinalizeMigration(instance, info, success)
432 def perspective_instance_migrate(params):
433 """Migrates an instance.
436 instance, target, live = params
437 instance = objects.Instance.FromDict(instance)
438 return backend.MigrateInstance(instance, target, live)
441 def perspective_instance_reboot(params):
442 """Reboot an instance.
445 instance = objects.Instance.FromDict(params[0])
446 reboot_type = params[1]
447 return backend.InstanceReboot(instance, reboot_type)
450 def perspective_instance_info(params):
451 """Query instance information.
454 return backend.GetInstanceInfo(params[0], params[1])
457 def perspective_instance_migratable(params):
458 """Query whether the specified instance can be migrated.
461 instance = objects.Instance.FromDict(params[0])
462 return backend.GetInstanceMigratable(instance)
465 def perspective_all_instances_info(params):
466 """Query information about all instances.
469 return backend.GetAllInstancesInfo(params[0])
472 def perspective_instance_list(params):
473 """Query the list of running instances.
476 return backend.GetInstanceList(params[0])
478 # node --------------------------
481 def perspective_node_tcp_ping(params):
482 """Do a TcpPing on the remote node.
485 return utils.TcpPing(params[1], params[2], timeout=params[3],
486 live_port_needed=params[4], source=params[0])
489 def perspective_node_has_ip_address(params):
490 """Checks if a node has the given ip address.
493 return utils.OwnIpAddress(params[0])
496 def perspective_node_info(params):
497 """Query node information.
500 vgname, hypervisor_type = params
501 return backend.GetNodeInfo(vgname, hypervisor_type)
504 def perspective_node_add(params):
505 """Complete the registration of this node in the cluster.
508 return backend.AddNode(params[0], params[1], params[2],
509 params[3], params[4], params[5])
512 def perspective_node_verify(params):
513 """Run a verify sequence on this node.
516 return backend.VerifyNode(params[0], params[1])
519 def perspective_node_start_master(params):
520 """Promote this node to master status.
523 return backend.StartMaster(params[0])
526 def perspective_node_stop_master(params):
527 """Demote this node from master status.
530 return backend.StopMaster(params[0])
533 def perspective_node_leave_cluster(params):
534 """Cleanup after leaving a cluster.
537 return backend.LeaveCluster()
540 def perspective_node_volumes(params):
541 """Query the list of all logical volume groups.
544 return backend.NodeVolumes()
547 def perspective_node_demote_from_mc(params):
548 """Demote a node from the master candidate role.
551 return backend.DemoteFromMC()
554 # cluster --------------------------
557 def perspective_version(params):
558 """Query version information.
561 return constants.PROTOCOL_VERSION
564 def perspective_upload_file(params):
567 Note that the backend implementation imposes strict rules on which
571 return backend.UploadFile(*params)
574 def perspective_master_info(params):
575 """Query master information.
578 return backend.GetMasterInfo()
581 def perspective_write_ssconf_files(params):
582 """Write ssconf files.
586 return backend.WriteSsconfFiles(values)
588 # os -----------------------
591 def perspective_os_diagnose(params):
592 """Query detailed information about existing OSes.
595 return [os_obj.ToDict() for os_obj in backend.DiagnoseOS()]
598 def perspective_os_get(params):
599 """Query information about a given OS.
604 os_obj = backend.OSFromDisk(name)
605 except errors.InvalidOS, err:
606 os_obj = objects.OS.FromInvalidOS(err)
607 return os_obj.ToDict()
609 # hooks -----------------------
612 def perspective_hooks_runner(params):
616 hpath, phase, env = params
617 hr = backend.HooksRunner()
618 return hr.RunHooks(hpath, phase, env)
620 # iallocator -----------------
623 def perspective_iallocator_runner(params):
624 """Run an iallocator script.
628 iar = backend.IAllocatorRunner()
629 return iar.Run(name, idata)
631 # test -----------------------
634 def perspective_test_delay(params):
639 return utils.TestDelay(duration)
641 # file storage ---------------
644 def perspective_file_storage_dir_create(params):
645 """Create the file storage directory.
648 file_storage_dir = params[0]
649 return backend.CreateFileStorageDir(file_storage_dir)
652 def perspective_file_storage_dir_remove(params):
653 """Remove the file storage directory.
656 file_storage_dir = params[0]
657 return backend.RemoveFileStorageDir(file_storage_dir)
660 def perspective_file_storage_dir_rename(params):
661 """Rename the file storage directory.
664 old_file_storage_dir = params[0]
665 new_file_storage_dir = params[1]
666 return backend.RenameFileStorageDir(old_file_storage_dir,
667 new_file_storage_dir)
669 # jobs ------------------------
672 @_RequireJobQueueLock
673 def perspective_jobqueue_update(params):
677 (file_name, content) = params
678 return backend.JobQueueUpdate(file_name, content)
681 @_RequireJobQueueLock
682 def perspective_jobqueue_purge(params):
686 return backend.JobQueuePurge()
689 @_RequireJobQueueLock
690 def perspective_jobqueue_rename(params):
691 """Rename a job queue file.
694 # TODO: What if a file fails to rename?
695 return [backend.JobQueueRename(old, new) for old, new in params]
698 def perspective_jobqueue_set_drain(params):
699 """Set/unset the queue drain flag.
702 drain_flag = params[0]
703 return backend.JobQueueSetDrainFlag(drain_flag)
706 # hypervisor ---------------
709 def perspective_hypervisor_validate_params(params):
710 """Validate the hypervisor parameters.
713 (hvname, hvparams) = params
714 return backend.ValidateHVParams(hvname, hvparams)
718 """Parse the command line options.
720 @return: (options, args) as from OptionParser.parse_args()
723 parser = OptionParser(description="Ganeti node daemon",
724 usage="%prog [-f] [-d] [-b ADDRESS]",
725 version="%%prog (ganeti) %s" %
726 constants.RELEASE_VERSION)
728 parser.add_option("-f", "--foreground", dest="fork",
729 help="Don't detach from the current terminal",
730 default=True, action="store_false")
731 parser.add_option("-d", "--debug", dest="debug",
732 help="Enable some debug messages",
733 default=False, action="store_true")
734 parser.add_option("-b", "--bind", dest="bind_address",
736 default="", metavar="ADDRESS")
738 options, args = parser.parse_args()
743 """Main function for the node daemon.
748 options, args = ParseOptions()
749 utils.debug = options.debug
754 for fname in (constants.SSL_CERT_FILE,):
755 if not os.path.isfile(fname):
756 print "config %s not there, will not run." % fname
757 sys.exit(constants.EXIT_NOTCLUSTER)
759 port = utils.GetNodeDaemonPort()
761 dirs = [(val, constants.RUN_DIRS_MODE) for val in constants.SUB_RUN_DIRS]
762 dirs.append((constants.LOG_OS_DIR, 0750))
763 dirs.append((constants.LOCK_DIR, 1777))
764 utils.EnsureDirs(dirs)
768 utils.Daemonize(logfile=constants.LOG_NODESERVER)
770 utils.WritePidFile(constants.NODED_PID)
772 utils.SetupLogging(logfile=constants.LOG_NODESERVER, debug=options.debug,
773 stderr_logging=not options.fork)
774 logging.info("ganeti node daemon startup")
776 # Read SSL certificate
777 ssl_params = http.HttpSslParams(ssl_key_path=constants.SSL_CERT_FILE,
778 ssl_cert_path=constants.SSL_CERT_FILE)
781 queue_lock = jstore.InitAndVerifyQueue(must_lock=False)
783 mainloop = daemon.Mainloop()
784 server = NodeHttpServer(mainloop, options.bind_address, port,
785 ssl_params=ssl_params, ssl_verify_peer=True)
792 utils.RemovePidFile(constants.NODED_PID)
795 if __name__ == '__main__':