X-Git-Url: https://code.grnet.gr/git/ganeti-local/blobdiff_plain/57c177af0052760a14f36db9fb1881655bbb1873..bd1e4562b3951edb0815823e2505121b8bcaae2b:/lib/backend.py diff --git a/lib/backend.py b/lib/backend.py index a6b676d..ca86245 100644 --- a/lib/backend.py +++ b/lib/backend.py @@ -30,8 +30,10 @@ import stat import errno import re import subprocess +import random +import logging +import tempfile -from ganeti import logger from ganeti import errors from ganeti import utils from ganeti import ssh @@ -46,36 +48,127 @@ def _GetSshRunner(): return ssh.SshRunner() -def StartMaster(): - """Activate local node as master node. +def _CleanDirectory(path, exclude=[]): + """Removes all regular files in a directory. - There are two needed steps for this: - - run the master script - - register the cron script + @param exclude: List of files to be excluded. + @type exclude: list """ - result = utils.RunCmd([constants.MASTER_SCRIPT, "-d", "start"]) + if not os.path.isdir(path): + return - if result.failed: - logger.Error("could not activate cluster interface with command %s," - " error: '%s'" % (result.cmd, result.output)) - return False + # Normalize excluded paths + exclude = [os.path.normpath(i) for i in exclude] - return True + for rel_name in utils.ListVisibleFiles(path): + full_name = os.path.normpath(os.path.join(path, rel_name)) + if full_name in exclude: + continue + if os.path.isfile(full_name) and not os.path.islink(full_name): + utils.RemoveFile(full_name) + + +def _JobQueuePurge(keep_lock): + """Removes job queue files and archived jobs + + """ + if keep_lock: + exclude = [constants.JOB_QUEUE_LOCK_FILE] + else: + exclude = [] + + _CleanDirectory(constants.QUEUE_DIR, exclude=exclude) + _CleanDirectory(constants.JOB_QUEUE_ARCHIVE_DIR) + + +def GetMasterInfo(): + """Returns master information. + This is an utility function to compute master information, either + for consumption here or from the node daemon. -def StopMaster(): + @rtype: tuple + @return: (master_netdev, master_ip, master_name) + + """ + try: + ss = ssconf.SimpleStore() + master_netdev = ss.GetMasterNetdev() + master_ip = ss.GetMasterIP() + master_node = ss.GetMasterNode() + except errors.ConfigurationError, err: + logging.exception("Cluster configuration incomplete") + return (None, None) + return (master_netdev, master_ip, master_node) + + +def StartMaster(start_daemons): + """Activate local node as master node. + + The function will always try activate the IP address of the master + (if someone else has it, then it won't). Then, if the start_daemons + parameter is True, it will also start the master daemons + (ganet-masterd and ganeti-rapi). + + """ + ok = True + master_netdev, master_ip, _ = GetMasterInfo() + if not master_netdev: + return False + + if utils.TcpPing(master_ip, constants.DEFAULT_NODED_PORT): + if utils.TcpPing(master_ip, constants.DEFAULT_NODED_PORT, + source=constants.LOCALHOST_IP_ADDRESS): + # we already have the ip: + logging.debug("Already started") + else: + logging.error("Someone else has the master ip, not activating") + ok = False + else: + result = utils.RunCmd(["ip", "address", "add", "%s/32" % master_ip, + "dev", master_netdev, "label", + "%s:0" % master_netdev]) + if result.failed: + logging.error("Can't activate master IP: %s", result.output) + ok = False + + result = utils.RunCmd(["arping", "-q", "-U", "-c 3", "-I", master_netdev, + "-s", master_ip, master_ip]) + # we'll ignore the exit code of arping + + # and now start the master and rapi daemons + if start_daemons: + for daemon in 'ganeti-masterd', 'ganeti-rapi': + result = utils.RunCmd([daemon]) + if result.failed: + logging.error("Can't start daemon %s: %s", daemon, result.output) + ok = False + return ok + + +def StopMaster(stop_daemons): """Deactivate this node as master. - This runs the master stop script. + The function will always try to deactivate the IP address of the + master. Then, if the stop_daemons parameter is True, it will also + stop the master daemons (ganet-masterd and ganeti-rapi). """ - result = utils.RunCmd([constants.MASTER_SCRIPT, "-d", "stop"]) + master_netdev, master_ip, _ = GetMasterInfo() + if not master_netdev: + return False + result = utils.RunCmd(["ip", "address", "del", "%s/32" % master_ip, + "dev", master_netdev]) if result.failed: - logger.Error("could not deactivate cluster interface with command %s," - " error: '%s'" % (result.cmd, result.output)) - return False + logging.error("Can't remove the master IP, error: %s", result.output) + # but otherwise ignore the failure + + if stop_daemons: + # stop/kill the rapi and the master daemon + for daemon in constants.RAPI_PID, constants.MASTERD_PID: + utils.KillProcess(utils.ReadPidFile(utils.DaemonPidFileName(daemon))) return True @@ -100,7 +193,7 @@ def AddNode(dsa, dsapub, rsa, rsapub, sshkey, sshpub): priv_key, pub_key, auth_keys = ssh.GetUserFiles(constants.GANETI_RUNAS, mkdir=True) except errors.OpExecError, err: - logger.Error("Error while processing user ssh files: %s" % err) + logging.exception("Error while processing user ssh files") return False for name, content in [(priv_key, sshkey), (pub_key, sshpub)]: @@ -117,16 +210,15 @@ def LeaveCluster(): """Cleans up the current node and prepares it to be removed from the cluster. """ - if os.path.isdir(constants.DATA_DIR): - for rel_name in utils.ListVisibleFiles(constants.DATA_DIR): - full_name = os.path.join(constants.DATA_DIR, rel_name) - if os.path.isfile(full_name) and not os.path.islink(full_name): - utils.RemoveFile(full_name) + _CleanDirectory(constants.DATA_DIR) + + # The lock can be removed because we're going to quit anyway. + _JobQueuePurge(keep_lock=False) try: priv_key, pub_key, auth_keys = ssh.GetUserFiles(constants.GANETI_RUNAS) - except errors.OpExecError, err: - logger.Error("Error while processing ssh files: %s" % err) + except errors.OpExecError: + logging.exception("Error while processing ssh files") return f = open(pub_key, 'r') @@ -138,6 +230,9 @@ def LeaveCluster(): utils.RemoveFile(priv_key) utils.RemoveFile(pub_key) + # Return a reassuring string to the caller, and quit + raise errors.QuitGanetiException(False, 'Shutdown scheduled') + def GetNodeInfo(vgname): """Gives back a hash with different informations about the node. @@ -200,10 +295,38 @@ def VerifyNode(what): if 'nodelist' in what: result['nodelist'] = {} + random.shuffle(what['nodelist']) for node in what['nodelist']: success, message = _GetSshRunner().VerifyNodeHostname(node) if not success: result['nodelist'][node] = message + if 'node-net-test' in what: + result['node-net-test'] = {} + my_name = utils.HostInfo().name + my_pip = my_sip = None + for name, pip, sip in what['node-net-test']: + if name == my_name: + my_pip = pip + my_sip = sip + break + if not my_pip: + result['node-net-test'][my_name] = ("Can't find my own" + " primary/secondary IP" + " in the node list") + else: + port = ssconf.SimpleStore().GetNodeDaemonPort() + for name, pip, sip in what['node-net-test']: + fail = [] + if not utils.TcpPing(pip, port, source=my_pip): + fail.append("primary") + if sip != pip: + if not utils.TcpPing(sip, port, source=my_sip): + fail.append("secondary") + if fail: + result['node-net-test'][name] = ("failure using the %s" + " interface(s)" % + " and ".join(fail)) + return result @@ -222,15 +345,18 @@ def GetVolumeList(vg_name): "--separator=%s" % sep, "-olv_name,lv_size,lv_attr", vg_name]) if result.failed: - logger.Error("Failed to list logical volumes, lvs output: %s" % - result.output) + logging.error("Failed to list logical volumes, lvs output: %s", + result.output) return result.output + valid_line_re = re.compile("^ *([^|]+)\|([0-9.]+)\|([^|]{6})\|?$") for line in result.stdout.splitlines(): - line = line.strip().rstrip(sep) - name, size, attr = line.split(sep) - if len(attr) != 6: - attr = '------' + line = line.strip() + match = valid_line_re.match(line) + if not match: + logging.error("Invalid line returned from lvs output: '%s'", line) + continue + name, size, attr = match.groups() inactive = attr[4] == '-' online = attr[5] == 'o' lvs[name] = (size, inactive, online) @@ -256,8 +382,8 @@ def NodeVolumes(): "--separator=|", "--options=lv_name,lv_size,devices,vg_name"]) if result.failed: - logger.Error("Failed to list logical volumes, lvs output: %s" % - result.output) + logging.error("Failed to list logical volumes, lvs output: %s", + result.output) return {} def parse_dev(dev): @@ -274,7 +400,8 @@ def NodeVolumes(): 'vg': line[3].strip(), } - return [map_line(line.split('|')) for line in result.stdout.splitlines()] + return [map_line(line.split('|')) for line in result.stdout.splitlines() + if line.count('|') >= 3] def BridgesExist(bridges_list): @@ -303,7 +430,7 @@ def GetInstanceList(): try: names = hypervisor.GetHypervisor().ListInstances() except errors.HypervisorError, err: - logger.Error("error enumerating instances: %s" % str(err)) + logging.exception("Error enumerating instances") raise return names @@ -381,12 +508,12 @@ def AddOSToInstance(instance, os_disk, swap_disk): os_device = instance.FindDisk(os_disk) if os_device is None: - logger.Error("Can't find this device-visible name '%s'" % os_disk) + logging.error("Can't find this device-visible name '%s'", os_disk) return False swap_device = instance.FindDisk(swap_disk) if swap_device is None: - logger.Error("Can't find this device-visible name '%s'" % swap_disk) + logging.error("Can't find this device-visible name '%s'", swap_disk) return False real_os_dev = _RecursiveFindBD(os_device) @@ -410,12 +537,13 @@ def AddOSToInstance(instance, os_disk, swap_disk): inst_os.path, create_script, instance.name, real_os_dev.dev_path, real_swap_dev.dev_path, logfile) + env = {'HYPERVISOR': ssconf.SimpleStore().GetHypervisorType()} - result = utils.RunCmd(command) + result = utils.RunCmd(command, env=env) if result.failed: - logger.Error("os create command '%s' returned error: %s, logfile: %s," - " output: %s" % - (command, result.fail_reason, logfile, result.output)) + logging.error("os create command '%s' returned error: %s, logfile: %s," + " output: %s", command, result.fail_reason, logfile, + result.output) return False return True @@ -437,12 +565,12 @@ def RunRenameInstance(instance, old_name, os_disk, swap_disk): os_device = instance.FindDisk(os_disk) if os_device is None: - logger.Error("Can't find this device-visible name '%s'" % os_disk) + logging.error("Can't find this device-visible name '%s'", os_disk) return False swap_device = instance.FindDisk(swap_disk) if swap_device is None: - logger.Error("Can't find this device-visible name '%s'" % swap_disk) + logging.error("Can't find this device-visible name '%s'", swap_disk) return False real_os_dev = _RecursiveFindBD(os_device) @@ -471,9 +599,8 @@ def RunRenameInstance(instance, old_name, os_disk, swap_disk): result = utils.RunCmd(command) if result.failed: - logger.Error("os create command '%s' returned error: %s" - " output: %s" % - (command, result.fail_reason, result.output)) + logging.error("os create command '%s' returned error: %s output: %s", + command, result.fail_reason, result.output) return False return True @@ -502,8 +629,7 @@ def _GetVGInfo(vg_name): "--nosuffix", "--units=m", "--separator=:", vg_name]) if retval.failed: - errmsg = "volume group %s not present" % vg_name - logger.Error(errmsg) + logging.error("volume group %s not present", vg_name) return retdic valarr = retval.stdout.strip().rstrip(':').split(':') if len(valarr) == 3: @@ -514,10 +640,10 @@ def _GetVGInfo(vg_name): "pv_count": int(valarr[2]), } except ValueError, err: - logger.Error("Fail to parse vgs output: %s" % str(err)) + logging.exception("Fail to parse vgs output") else: - logger.Error("vgs output has the wrong number of fields (expected" - " three): %s" % str(valarr)) + logging.error("vgs output has the wrong number of fields (expected" + " three): %s", str(valarr)) return retdic @@ -557,7 +683,7 @@ def StartInstance(instance, extra_args): try: hyper.StartInstance(instance, block_devices, extra_args) except errors.HypervisorError, err: - logger.Error("Failed to start instance: %s" % err) + logging.exception("Failed to start instance") return False return True @@ -579,7 +705,7 @@ def ShutdownInstance(instance): try: hyper.StopInstance(instance) except errors.HypervisorError, err: - logger.Error("Failed to stop instance: %s" % err) + logging.error("Failed to stop instance") return False # test every 10secs for 2min @@ -592,17 +718,18 @@ def ShutdownInstance(instance): time.sleep(10) else: # the shutdown did not succeed - logger.Error("shutdown of '%s' unsuccessful, using destroy" % instance) + logging.error("shutdown of '%s' unsuccessful, using destroy", instance) try: hyper.StopInstance(instance, force=True) except errors.HypervisorError, err: - logger.Error("Failed to stop instance: %s" % err) + logging.exception("Failed to stop instance") return False time.sleep(1) if instance.name in GetInstanceList(): - logger.Error("could not shutdown instance '%s' even by destroy") + logging.error("could not shutdown instance '%s' even by destroy", + instance.name) return False return True @@ -619,7 +746,7 @@ def RebootInstance(instance, reboot_type, extra_args): running_instances = GetInstanceList() if instance.name not in running_instances: - logger.Error("Cannot reboot instance that is not running") + logging.error("Cannot reboot instance that is not running") return False hyper = hypervisor.GetHypervisor() @@ -627,14 +754,14 @@ def RebootInstance(instance, reboot_type, extra_args): try: hyper.RebootInstance(instance) except errors.HypervisorError, err: - logger.Error("Failed to soft reboot instance: %s" % err) + logging.exception("Failed to soft reboot instance") return False elif reboot_type == constants.INSTANCE_REBOOT_HARD: try: ShutdownInstance(instance) StartInstance(instance, extra_args) except errors.HypervisorError, err: - logger.Error("Failed to hard reboot instance: %s" % err) + logging.exception("Failed to hard reboot instance") return False else: raise errors.ParameterError("reboot_type invalid") @@ -643,6 +770,21 @@ def RebootInstance(instance, reboot_type, extra_args): return True +def MigrateInstance(instance, target, live): + """Migrates an instance to another node. + + """ + hyper = hypervisor.GetHypervisor() + + try: + hyper.MigrateInstance(instance, target, live) + except errors.HypervisorError, err: + msg = "Failed to migrate instance: %s" % str(err) + logging.error(msg) + return (False, msg) + return (True, "Migration successfull") + + def CreateBlockDevice(disk, size, owner, on_primary, info): """Creates a block device for an instance. @@ -671,7 +813,7 @@ def CreateBlockDevice(disk, size, owner, on_primary, info): try: device = bdev.FindDevice(disk.dev_type, disk.physical_id, clist) if device is not None: - logger.Info("removing existing device %s" % disk) + logging.info("removing existing device %s", disk) device.Remove() except errors.BlockDeviceError, err: pass @@ -684,7 +826,7 @@ def CreateBlockDevice(disk, size, owner, on_primary, info): if on_primary or disk.AssembleOnSecondary(): if not device.Assemble(): errorstring = "Can't assemble device after creation" - logger.Error(errorstring) + logging.error(errorstring) raise errors.BlockDeviceError("%s, very unusual event - check the node" " daemon logs" % errorstring) device.SetSyncSpeed(constants.SYNC_SPEED) @@ -711,7 +853,7 @@ def RemoveBlockDevice(disk): rdev = _RecursiveFindBD(disk, allow_partial=True) except errors.BlockDeviceError, err: # probably can't attach - logger.Info("Can't attach to device %s in remove" % disk) + logging.info("Can't attach to device %s in remove", disk) rdev = None if rdev is not None: r_path = rdev.dev_path @@ -757,7 +899,7 @@ def _RecursiveAssembleBD(disk, owner, as_primary): if children.count(None) >= mcn: raise cdev = None - logger.Debug("Error in child activation: %s" % str(err)) + logging.debug("Error in child activation: %s", str(err)) children.append(cdev) if as_primary or disk.AssembleOnSecondary(): @@ -821,12 +963,12 @@ def MirrorAddChildren(parent_cdev, new_cdevs): """ parent_bdev = _RecursiveFindBD(parent_cdev, allow_partial=True) if parent_bdev is None: - logger.Error("Can't find parent device") + logging.error("Can't find parent device") return False new_bdevs = [_RecursiveFindBD(disk) for disk in new_cdevs] if new_bdevs.count(None) > 0: - logger.Error("Can't find new device(s) to add: %s:%s" % - (new_bdevs, new_cdevs)) + logging.error("Can't find new device(s) to add: %s:%s", + new_bdevs, new_cdevs) return False parent_bdev.AddChildren(new_bdevs) return True @@ -838,7 +980,7 @@ def MirrorRemoveChildren(parent_cdev, new_cdevs): """ parent_bdev = _RecursiveFindBD(parent_cdev) if parent_bdev is None: - logger.Error("Can't find parent in remove children: %s" % parent_cdev) + logging.error("Can't find parent in remove children: %s", parent_cdev) return False devs = [] for disk in new_cdevs: @@ -846,8 +988,8 @@ def MirrorRemoveChildren(parent_cdev, new_cdevs): if rpath is None: bd = _RecursiveFindBD(disk) if bd is None: - logger.Error("Can't find dynamic device %s while removing children" % - disk) + logging.error("Can't find dynamic device %s while removing children", + disk) return False else: devs.append(bd.dev_path) @@ -927,19 +1069,21 @@ def UploadFile(file_name, data, mode, uid, gid, atime, mtime): """ if not os.path.isabs(file_name): - logger.Error("Filename passed to UploadFile is not absolute: '%s'" % - file_name) + logging.error("Filename passed to UploadFile is not absolute: '%s'", + file_name) return False allowed_files = [ constants.CLUSTER_CONF_FILE, constants.ETC_HOSTS, constants.SSH_KNOWN_HOSTS_FILE, + constants.VNC_PASSWORD_FILE, ] allowed_files.extend(ssconf.SimpleStore().GetFileList()) + if file_name not in allowed_files: - logger.Error("Filename passed to UploadFile not in allowed" - " upload targets: '%s'" % file_name) + logging.error("Filename passed to UploadFile not in allowed" + " upload targets: '%s'", file_name) return False utils.WriteFile(file_name, data=data, mode=mode, uid=uid, gid=gid, @@ -1023,8 +1167,7 @@ def DiagnoseOS(top_dirs=None): try: f_names = utils.ListVisibleFiles(dir_name) except EnvironmentError, err: - logger.Error("Can't list the OS directory %s: %s" % - (dir_name, str(err))) + logging.exception("Can't list the OS directory %s", dir_name) break for name in f_names: try: @@ -1093,6 +1236,32 @@ def OSFromDisk(name, base_dir=None): api_version=api_version) +def GrowBlockDevice(disk, amount): + """Grow a stack of block devices. + + This function is called recursively, with the childrens being the + first one resize. + + Args: + disk: the disk to be grown + + Returns: a tuple of (status, result), with: + status: the result (true/false) of the operation + result: the error message if the operation failed, otherwise not used + + """ + r_dev = _RecursiveFindBD(disk) + if r_dev is None: + return False, "Cannot find block device %s" % (disk,) + + try: + r_dev.Grow(amount) + except errors.BlockDeviceError, err: + return False, str(err) + + return True, None + + def SnapshotBlockDevice(disk): """Create a snapshot copy of a block device. @@ -1179,9 +1348,8 @@ def ExportSnapshot(disk, dest_node, instance): result = utils.RunCmd(command) if result.failed: - logger.Error("os snapshot export command '%s' returned error: %s" - " output: %s" % - (command, result.fail_reason, result.output)) + logging.error("os snapshot export command '%s' returned error: %s" + " output: %s", command, result.fail_reason, result.output) return False return True @@ -1215,14 +1383,18 @@ def FinalizeExport(instance, snap_disks): config.set(constants.INISECT_INS, 'memory', '%d' % instance.memory) config.set(constants.INISECT_INS, 'vcpus', '%d' % instance.vcpus) config.set(constants.INISECT_INS, 'disk_template', instance.disk_template) + + nic_count = 0 for nic_count, nic in enumerate(instance.nics): config.set(constants.INISECT_INS, 'nic%d_mac' % nic_count, '%s' % nic.mac) config.set(constants.INISECT_INS, 'nic%d_ip' % nic_count, '%s' % nic.ip) - config.set(constants.INISECT_INS, 'nic%d_bridge' % nic_count, '%s' % nic.bridge) + config.set(constants.INISECT_INS, 'nic%d_bridge' % nic_count, + '%s' % nic.bridge) # TODO: redundant: on load can read nics until it doesn't exist config.set(constants.INISECT_INS, 'nic_count' , '%d' % nic_count) + disk_count = 0 for disk_count, disk in enumerate(snap_disks): config.set(constants.INISECT_INS, 'disk%d_ivname' % disk_count, ('%s' % disk.iv_name)) @@ -1286,12 +1458,12 @@ def ImportOSIntoInstance(instance, os_disk, swap_disk, src_node, src_image): os_device = instance.FindDisk(os_disk) if os_device is None: - logger.Error("Can't find this device-visible name '%s'" % os_disk) + logging.error("Can't find this device-visible name '%s'", os_disk) return False swap_device = instance.FindDisk(swap_disk) if swap_device is None: - logger.Error("Can't find this device-visible name '%s'" % swap_disk) + logging.error("Can't find this device-visible name '%s'", swap_disk) return False real_os_dev = _RecursiveFindBD(os_device) @@ -1322,13 +1494,13 @@ def ImportOSIntoInstance(instance, os_disk, swap_disk, src_node, src_image): logfile) command = '|'.join([utils.ShellQuoteArgs(remotecmd), comprcmd, impcmd]) + env = {'HYPERVISOR': ssconf.SimpleStore().GetHypervisorType()} - result = utils.RunCmd(command) + result = utils.RunCmd(command, env=env) if result.failed: - logger.Error("os import command '%s' returned error: %s" - " output: %s" % - (command, result.fail_reason, result.output)) + logging.error("os import command '%s' returned error: %s" + " output: %s", command, result.fail_reason, result.output) return False return True @@ -1389,8 +1561,7 @@ def RenameBlockDevices(devlist): # cache? for now, we only lose lvm data when we rename, which # is less critical than DRBD or MD except errors.BlockDeviceError, err: - logger.Error("Can't rename device '%s' to '%s': %s" % - (dev, unique_id, err)) + logging.exception("Can't rename device '%s' to '%s'", dev, unique_id) result = False return result @@ -1404,7 +1575,7 @@ def _TransformFileStorageDir(file_storage_dir): Args: file_storage_dir: string with path - + Returns: normalized file_storage_dir (string) if valid, None otherwise @@ -1413,9 +1584,9 @@ def _TransformFileStorageDir(file_storage_dir): base_file_storage_dir = ssconf.SimpleStore().GetFileStorageDir() if (not os.path.commonprefix([file_storage_dir, base_file_storage_dir]) == base_file_storage_dir): - logger.Error("file storage directory '%s' is not under base file" - " storage directory '%s'" % - (file_storage_dir, base_file_storage_dir)) + logging.error("file storage directory '%s' is not under base file" + " storage directory '%s'", + file_storage_dir, base_file_storage_dir) return None return file_storage_dir @@ -1438,14 +1609,14 @@ def CreateFileStorageDir(file_storage_dir): else: if os.path.exists(file_storage_dir): if not os.path.isdir(file_storage_dir): - logger.Error("'%s' is not a directory" % file_storage_dir) + logging.error("'%s' is not a directory", file_storage_dir) result = False, else: try: os.makedirs(file_storage_dir, 0750) except OSError, err: - logger.Error("Cannot create file storage directory '%s': %s" % - (file_storage_dir, err)) + logging.error("Cannot create file storage directory '%s': %s", + file_storage_dir, err) result = False, return result @@ -1470,14 +1641,14 @@ def RemoveFileStorageDir(file_storage_dir): else: if os.path.exists(file_storage_dir): if not os.path.isdir(file_storage_dir): - logger.Error("'%s' is not a directory" % file_storage_dir) + logging.error("'%s' is not a directory", file_storage_dir) result = False, # deletes dir only if empty, otherwise we want to return False try: os.rmdir(file_storage_dir) except OSError, err: - logger.Error("Cannot remove file storage directory '%s': %s" % - (file_storage_dir, err)) + logging.exception("Cannot remove file storage directory '%s'", + file_storage_dir) result = False, return result @@ -1505,20 +1676,93 @@ def RenameFileStorageDir(old_file_storage_dir, new_file_storage_dir): try: os.rename(old_file_storage_dir, new_file_storage_dir) except OSError, err: - logger.Error("Cannot rename '%s' to '%s': %s" - % (old_file_storage_dir, new_file_storage_dir, err)) + logging.exception("Cannot rename '%s' to '%s'", + old_file_storage_dir, new_file_storage_dir) result = False, else: - logger.Error("'%s' is not a directory" % old_file_storage_dir) + logging.error("'%s' is not a directory", old_file_storage_dir) result = False, else: if os.path.exists(old_file_storage_dir): - logger.Error("Cannot rename '%s' to '%s'. Both locations exist." % - old_file_storage_dir, new_file_storage_dir) + logging.error("Cannot rename '%s' to '%s'. Both locations exist.", + old_file_storage_dir, new_file_storage_dir) result = False, return result +def _IsJobQueueFile(file_name): + """Checks whether the given filename is in the queue directory. + + """ + queue_dir = os.path.normpath(constants.QUEUE_DIR) + result = (os.path.commonprefix([queue_dir, file_name]) == queue_dir) + + if not result: + logging.error("'%s' is not a file in the queue directory", + file_name) + + return result + + +def JobQueueUpdate(file_name, content): + """Updates a file in the queue directory. + + """ + if not _IsJobQueueFile(file_name): + return False + + # Write and replace the file atomically + utils.WriteFile(file_name, data=content) + + return True + + +def JobQueuePurge(): + """Removes job queue files and archived jobs + + """ + # The lock must not be removed, otherwise another process could create + # it again. + return _JobQueuePurge(keep_lock=True) + + +def JobQueueRename(old, new): + """Renames a job queue file. + + """ + if not (_IsJobQueueFile(old) and _IsJobQueueFile(new)): + return False + + os.rename(old, new) + + return True + + +def CloseBlockDevices(disks): + """Closes the given block devices. + + This means they will be switched to secondary mode (in case of DRBD). + + """ + bdevs = [] + for cf in disks: + rd = _RecursiveFindBD(cf) + if rd is None: + return (False, "Can't find device %s" % cf) + bdevs.append(rd) + + msg = [] + for rd in bdevs: + try: + rd.Close() + except errors.BlockDeviceError, err: + msg.append(str(err)) + if msg: + return (False, "Can't make devices secondary: %s" % ",".join(msg)) + else: + return (True, "All devices secondary") + + class HooksRunner(object): """Hook runner. @@ -1534,9 +1778,6 @@ class HooksRunner(object): Args: - hooks_base_dir: if not None, this overrides the constants.HOOKS_BASE_DIR (useful for unittests) - - logs_base_dir: if not None, this overrides the - constants.LOG_HOOKS_DIR (useful for unittests) - - logging: enable or disable logging of script output """ if hooks_base_dir is None: @@ -1548,7 +1789,6 @@ class HooksRunner(object): """Exec one hook script. Args: - - phase: the phase - script: the full path to the script - env: the environment with which to exec the script @@ -1583,7 +1823,7 @@ class HooksRunner(object): fd.close() except EnvironmentError, err: # just log the error - #logger.Error("While closing fd %s: %s" % (fd, err)) + #logging.exception("Error while closing fd %s", fd) pass return result == 0, output @@ -1630,6 +1870,42 @@ class HooksRunner(object): return rr +class IAllocatorRunner(object): + """IAllocator runner. + + This class is instantiated on the node side (ganeti-noded) and not on + the master side. + + """ + def Run(self, name, idata): + """Run an iallocator script. + + Return value: tuple of: + - run status (one of the IARUN_ constants) + - stdout + - stderr + - fail reason (as from utils.RunResult) + + """ + alloc_script = utils.FindFile(name, constants.IALLOCATOR_SEARCH_PATH, + os.path.isfile) + if alloc_script is None: + return (constants.IARUN_NOTFOUND, None, None, None) + + fd, fin_name = tempfile.mkstemp(prefix="ganeti-iallocator.") + try: + os.write(fd, idata) + os.close(fd) + result = utils.RunCmd([alloc_script, fin_name]) + if result.failed: + return (constants.IARUN_FAILURE, result.stdout, result.stderr, + result.fail_reason) + finally: + os.unlink(fin_name) + + return (constants.IARUN_SUCCESS, result.stdout, result.stderr, None) + + class DevCacheManager(object): """Simple class for managing a cache of block device information. @@ -1657,7 +1933,7 @@ class DevCacheManager(object): """ if dev_path is None: - logger.Error("DevCacheManager.UpdateCache got a None dev_path") + logging.error("DevCacheManager.UpdateCache got a None dev_path") return fpath = cls._ConvertPath(dev_path) if on_primary: @@ -1670,8 +1946,7 @@ class DevCacheManager(object): try: utils.WriteFile(fpath, data=fdata) except EnvironmentError, err: - logger.Error("Can't update bdev cache for %s, error %s" % - (dev_path, str(err))) + logging.exception("Can't update bdev cache for %s", dev_path) @classmethod def RemoveCache(cls, dev_path): @@ -1679,11 +1954,10 @@ class DevCacheManager(object): """ if dev_path is None: - logger.Error("DevCacheManager.RemoveCache got a None dev_path") + logging.error("DevCacheManager.RemoveCache got a None dev_path") return fpath = cls._ConvertPath(dev_path) try: utils.RemoveFile(fpath) except EnvironmentError, err: - logger.Error("Can't update bdev cache for %s, error %s" % - (dev_path, str(err))) + logging.exception("Can't update bdev cache for %s", dev_path)