X-Git-Url: https://code.grnet.gr/git/ganeti-local/blobdiff_plain/2a52a064ad9411c5f5b45d69955a6a861a767d4b..29921401f2fb372efac3d5e022a4622b37c8e07e:/lib/backend.py diff --git a/lib/backend.py b/lib/backend.py index 10ddcf5..c8d5c85 100644 --- a/lib/backend.py +++ b/lib/backend.py @@ -68,10 +68,11 @@ def _Fail(msg, *args, **kwargs): """ if args: msg = msg % args - if "exc" in kwargs and kwargs["exc"]: - logging.exception(msg) - else: - logging.error(msg) + if "log" not in kwargs or kwargs["log"]: # if we should log this error + if "exc" in kwargs and kwargs["exc"]: + logging.exception(msg) + else: + logging.error(msg) raise RPCFail(msg) @@ -147,7 +148,8 @@ def _CleanDirectory(path, exclude=None): def JobQueuePurge(): """Removes job queue files and archived jobs. - @rtype: None + @rtype: tuple + @return: True, None """ _CleanDirectory(constants.QUEUE_DIR, exclude=[constants.JOB_QUEUE_LOCK_FILE]) @@ -161,7 +163,7 @@ def GetMasterInfo(): for consumption here or from the node daemon. @rtype: tuple - @return: True, (master_netdev, master_ip, master_name) in case of success + @return: master_netdev, master_ip, master_name @raise RPCFail: in case of errors """ @@ -171,8 +173,8 @@ def GetMasterInfo(): master_ip = cfg.GetMasterIP() master_node = cfg.GetMasterNode() except errors.ConfigurationError, err: - _Fail("Cluster configuration incomplete", exc=True) - return True, (master_netdev, master_ip, master_node) + _Fail("Cluster configuration incomplete: %s", err, exc=True) + return master_netdev, master_ip, master_node def StartMaster(start_daemons): @@ -183,13 +185,13 @@ def StartMaster(start_daemons): based on the start_daemons parameter. @type start_daemons: boolean - @param start_daemons: whther to also start the master + @param start_daemons: whether to also start the master daemons (ganeti-masterd and ganeti-rapi) @rtype: None """ # GetMasterInfo will raise an exception if not able to return data - master_netdev, master_ip, _ = GetMasterInfo()[1] + master_netdev, master_ip, _ = GetMasterInfo() payload = [] if utils.TcpPing(master_ip, constants.DEFAULT_NODED_PORT): @@ -222,7 +224,8 @@ def StartMaster(start_daemons): logging.error(msg) payload.append(msg) - return not payload, "; ".join(payload) + if payload: + _Fail("; ".join(payload)) def StopMaster(stop_daemons): @@ -242,7 +245,7 @@ def StopMaster(stop_daemons): # need to decide in which case we fail the RPC for this # GetMasterInfo will raise an exception if not able to return data - master_netdev, master_ip, _ = GetMasterInfo()[1] + master_netdev, master_ip, _ = GetMasterInfo() result = utils.RunCmd(["ip", "address", "del", "%s/32" % master_ip, "dev", master_netdev]) @@ -255,8 +258,6 @@ def StopMaster(stop_daemons): for daemon in constants.RAPI_PID, constants.MASTERD_PID: utils.KillProcess(utils.ReadPidFile(utils.DaemonPidFileName(daemon))) - return True, None - def AddNode(dsa, dsapub, rsa, rsapub, sshkey, sshpub): """Joins this node to the cluster. @@ -302,8 +303,6 @@ def AddNode(dsa, dsapub, rsa, rsapub, sshkey, sshpub): utils.RunCmd([constants.SSH_INITD_SCRIPT, "restart"]) - return (True, "Node added successfully") - def LeaveCluster(): """Cleans up and remove the current node. @@ -370,7 +369,7 @@ def GetNodeInfo(vgname, hypervisor_type): finally: f.close() - return True, outputarray + return outputarray def VerifyNode(what, cluster_name): @@ -472,7 +471,7 @@ def VerifyNode(what, cluster_name): used_minors = str(err) result[constants.NV_DRBDLIST] = used_minors - return True, result + return result def GetVolumeList(vg_name): @@ -522,7 +521,7 @@ def ListVolumeGroups(): size of the volume """ - return True, utils.ListVolumeGroups() + return utils.ListVolumeGroups() def NodeVolumes(): @@ -565,9 +564,8 @@ def NodeVolumes(): 'vg': line[3].strip(), } - return True, [map_line(line.split('|')) - for line in result.stdout.splitlines() - if line.count('|') >= 3] + return [map_line(line.split('|')) for line in result.stdout.splitlines() + if line.count('|') >= 3] def BridgesExist(bridges_list): @@ -583,9 +581,7 @@ def BridgesExist(bridges_list): missing.append(bridge) if missing: - return False, "Missing bridges %s" % (", ".join(missing),) - - return True, None + _Fail("Missing bridges %s", ", ".join(missing)) def GetInstanceList(hypervisor_list): @@ -635,7 +631,7 @@ def GetInstanceInfo(instance, hname): output['state'] = iinfo[4] output['time'] = iinfo[5] - return True, output + return output def GetInstanceMigratable(instance): @@ -651,15 +647,14 @@ def GetInstanceMigratable(instance): """ hyper = hypervisor.GetHypervisor(instance.hypervisor) - if instance.name not in hyper.ListInstances(): - return (False, 'not running') + iname = instance.name + if iname not in hyper.ListInstances(): + _Fail("Instance %s is not running", iname) for idx in range(len(instance.disks)): - link_name = _GetBlockDevSymlinkPath(instance.name, idx) + link_name = _GetBlockDevSymlinkPath(iname, idx) if not os.path.islink(link_name): - return (False, 'not restarted since ganeti 1.2.5') - - return (True, '') + _Fail("Instance %s was not restarted since ganeti 1.2.5", iname) def GetAllInstancesInfo(hypervisor_list): @@ -685,7 +680,7 @@ def GetAllInstancesInfo(hypervisor_list): for hname in hypervisor_list: iinfo = hypervisor.GetHypervisor(hname).GetAllInstancesInfo() if iinfo: - for name, inst_id, memory, vcpus, state, times in iinfo: + for name, _, memory, vcpus, state, times in iinfo: value = { 'memory': memory, 'vcpus': vcpus, @@ -702,7 +697,7 @@ def GetAllInstancesInfo(hypervisor_list): " with different parameters", name) output[name] = value - return True, output + return output def InstanceOsAdd(instance, reinstall): @@ -712,19 +707,10 @@ def InstanceOsAdd(instance, reinstall): @param instance: Instance whose OS is to be installed @type reinstall: boolean @param reinstall: whether this is an instance reinstall - @rtype: boolean - @return: the success of the operation + @rtype: None """ - try: - inst_os = OSFromDisk(instance.os) - except errors.InvalidOS, err: - os_name, os_dir, os_err = err.args - if os_dir is None: - return (False, "Can't find OS '%s': %s" % (os_name, os_err)) - else: - return (False, "Error parsing OS '%s' in directory %s: %s" % - (os_name, os_dir, os_err)) + inst_os = OSFromDisk(instance.os) create_env = OSEnvironment(instance) if reinstall: @@ -741,10 +727,8 @@ def InstanceOsAdd(instance, reinstall): result.output) lines = [utils.SafeEncode(val) for val in utils.TailFile(logfile, lines=20)] - return (False, "OS create script failed (%s), last lines in the" - " log file:\n%s" % (result.fail_reason, "\n".join(lines))) - - return (True, "Successfully installed") + _Fail("OS create script failed (%s), last lines in the" + " log file:\n%s", result.fail_reason, "\n".join(lines), log=False) def RunRenameInstance(instance, old_name): @@ -775,10 +759,8 @@ def RunRenameInstance(instance, old_name): result.cmd, result.fail_reason, result.output) lines = [utils.SafeEncode(val) for val in utils.TailFile(logfile, lines=20)] - return (False, "OS rename script failed (%s), last lines in the" - " log file:\n%s" % (result.fail_reason, "\n".join(lines))) - - return (True, "Rename successful") + _Fail("OS rename script failed (%s), last lines in the" + " log file:\n%s", result.fail_reason, "\n".join(lines), log=False) def _GetVGInfo(vg_name): @@ -814,7 +796,7 @@ def _GetVGInfo(vg_name): "pv_count": int(valarr[2]), } except ValueError, err: - logging.exception("Fail to parse vgs output") + logging.exception("Fail to parse vgs output: %s", err) else: logging.error("vgs output has the wrong number of fields (expected" " three): %s", str(valarr)) @@ -858,7 +840,7 @@ def _RemoveBlockDevLinks(instance_name, disks): """Remove the block device symlinks belonging to the given instance. """ - for idx, disk in enumerate(disks): + for idx, _ in enumerate(disks): link_name = _GetBlockDevSymlinkPath(instance_name, idx) if os.path.islink(link_name): try: @@ -902,14 +884,14 @@ def StartInstance(instance): @type instance: L{objects.Instance} @param instance: the instance object - @rtype: boolean - @return: whether the startup was successful or not + @rtype: None """ running_instances = GetInstanceList([instance.hypervisor]) if instance.name in running_instances: - return (True, "Already running") + logging.info("Instance %s already running, not starting", instance.name) + return try: block_devices = _GatherAndLinkBlockDevs(instance) @@ -921,8 +903,6 @@ def StartInstance(instance): _RemoveBlockDevLinks(instance.name, instance.disks) _Fail("Hypervisor error: %s", err, exc=True) - return (True, "Instance started successfully") - def InstanceShutdown(instance): """Shut an instance down. @@ -931,21 +911,22 @@ def InstanceShutdown(instance): @type instance: L{objects.Instance} @param instance: the instance object - @rtype: boolean - @return: whether the startup was successful or not + @rtype: None """ hv_name = instance.hypervisor running_instances = GetInstanceList([hv_name]) + iname = instance.name - if instance.name not in running_instances: - return (True, "Instance already stopped") + if iname not in running_instances: + logging.info("Instance %s not running, doing nothing", iname) + return hyper = hypervisor.GetHypervisor(hv_name) try: hyper.StopInstance(instance) except errors.HypervisorError, err: - _Fail("Failed to stop instance %s: %s", instance.name, err) + _Fail("Failed to stop instance %s: %s", iname, err) # test every 10secs for 2min @@ -956,21 +937,18 @@ def InstanceShutdown(instance): time.sleep(10) else: # the shutdown did not succeed - logging.error("Shutdown of '%s' unsuccessful, using destroy", - instance.name) + logging.error("Shutdown of '%s' unsuccessful, using destroy", iname) try: hyper.StopInstance(instance, force=True) except errors.HypervisorError, err: - _Fail("Failed to force stop instance %s: %s", instance.name, err) + _Fail("Failed to force stop instance %s: %s", iname, err) time.sleep(1) if instance.name in GetInstanceList([hv_name]): - _Fail("Could not shutdown instance %s even by destroy", instance.name) + _Fail("Could not shutdown instance %s even by destroy", iname) - _RemoveBlockDevLinks(instance.name, instance.disks) - - return (True, "Instance has been shutdown successfully") + _RemoveBlockDevLinks(iname, instance.disks) def InstanceReboot(instance, reboot_type): @@ -988,8 +966,7 @@ def InstanceReboot(instance, reboot_type): - the other reboot type (L{constants.INSTANCE_REBOOT_HARD}) is not accepted here, since that mode is handled differently - @rtype: boolean - @return: the success of the operation + @rtype: None """ running_instances = GetInstanceList([instance.hypervisor]) @@ -1005,17 +982,13 @@ def InstanceReboot(instance, reboot_type): _Fail("Failed to soft reboot instance %s: %s", instance.name, err) elif reboot_type == constants.INSTANCE_REBOOT_HARD: try: - stop_result = InstanceShutdown(instance) - if not stop_result[0]: - return stop_result + InstanceShutdown(instance) return StartInstance(instance) except errors.HypervisorError, err: _Fail("Failed to hard reboot instance %s: %s", instance.name, err) else: _Fail("Invalid reboot_type received: %s", reboot_type) - return (True, "Reboot successful") - def MigrationInfo(instance): """Gather information about an instance to be migrated. @@ -1029,7 +1002,7 @@ def MigrationInfo(instance): info = hyper.MigrationInfo(instance) except errors.HypervisorError, err: _Fail("Failed to fetch migration information: %s", err, exc=True) - return (True, info) + return info def AcceptInstance(instance, info, target): @@ -1048,7 +1021,6 @@ def AcceptInstance(instance, info, target): hyper.AcceptInstance(instance, info, target) except errors.HypervisorError, err: _Fail("Failed to accept instance: %s", err, exc=True) - return (True, "Accept successfull") def FinalizeMigration(instance, info, success): @@ -1067,7 +1039,6 @@ def FinalizeMigration(instance, info, success): hyper.FinalizeMigration(instance, info, success) except errors.HypervisorError, err: _Fail("Failed to finalize migration: %s", err, exc=True) - return (True, "Migration Finalized") def MigrateInstance(instance, target, live): @@ -1092,7 +1063,6 @@ def MigrateInstance(instance, target, live): hyper.MigrateInstance(instance.name, target, live) except errors.HypervisorError, err: _Fail("Failed to migrate instance: %s", err, exc=True) - return (True, "Migration successfull") def BlockdevCreate(disk, size, owner, on_primary, info): @@ -1133,7 +1103,7 @@ def BlockdevCreate(disk, size, owner, on_primary, info): clist.append(crdev) try: - device = bdev.Create(disk.dev_type, disk.physical_id, clist, size) + device = bdev.Create(disk.dev_type, disk.physical_id, clist, disk.size) except errors.BlockDeviceError, err: _Fail("Can't create block device: %s", err) @@ -1153,8 +1123,7 @@ def BlockdevCreate(disk, size, owner, on_primary, info): device.SetInfo(info) - physical_id = device.unique_id - return True, physical_id + return device.unique_id def BlockdevRemove(disk): @@ -1169,7 +1138,6 @@ def BlockdevRemove(disk): """ msgs = [] - result = True try: rdev = _RecursiveFindBD(disk) except errors.BlockDeviceError, err: @@ -1182,18 +1150,18 @@ def BlockdevRemove(disk): rdev.Remove() except errors.BlockDeviceError, err: msgs.append(str(err)) - result = False - if result: + if not msgs: DevCacheManager.RemoveCache(r_path) if disk.children: for child in disk.children: - c_status, c_msg = BlockdevRemove(child) - result = result and c_status - if c_msg: # not an empty message - msgs.append(c_msg) + try: + BlockdevRemove(child) + except RPCFail, err: + msgs.append(str(err)) - return (result, "; ".join(msgs)) + if msgs: + _Fail("; ".join(msgs)) def _RecursiveAssembleBD(disk, owner, as_primary): @@ -1237,7 +1205,7 @@ def _RecursiveAssembleBD(disk, owner, as_primary): children.append(cdev) if as_primary or disk.AssembleOnSecondary(): - r_dev = bdev.Assemble(disk.dev_type, disk.physical_id, children) + r_dev = bdev.Assemble(disk.dev_type, disk.physical_id, children, disk.size) r_dev.SetSyncSpeed(constants.SYNC_SPEED) result = r_dev if as_primary or disk.OpenOnSecondary(): @@ -1260,16 +1228,14 @@ def BlockdevAssemble(disk, owner, as_primary): C{True} for secondary nodes """ - status = True - result = "no error information" try: result = _RecursiveAssembleBD(disk, owner, as_primary) if isinstance(result, bdev.BlockDev): result = result.dev_path except errors.BlockDeviceError, err: - result = "Error while assembling disk: %s" % str(err) - status = False - return (status, result) + _Fail("Error while assembling disk: %s", err, exc=True) + + return result def BlockdevShutdown(disk): @@ -1286,12 +1252,10 @@ def BlockdevShutdown(disk): @type disk: L{objects.Disk} @param disk: the description of the disk we should shutdown - @rtype: boolean - @return: the success of the operation + @rtype: None """ msgs = [] - result = True r_dev = _RecursiveFindBD(disk) if r_dev is not None: r_path = r_dev.dev_path @@ -1300,16 +1264,16 @@ def BlockdevShutdown(disk): DevCacheManager.RemoveCache(r_path) except errors.BlockDeviceError, err: msgs.append(str(err)) - result = False if disk.children: for child in disk.children: - c_status, c_msg = BlockdevShutdown(child) - result = result and c_status - if c_msg: # not an empty message - msgs.append(c_msg) + try: + BlockdevShutdown(child) + except RPCFail, err: + msgs.append(str(err)) - return (result, "; ".join(msgs)) + if msgs: + _Fail("; ".join(msgs)) def BlockdevAddchildren(parent_cdev, new_cdevs): @@ -1319,8 +1283,7 @@ def BlockdevAddchildren(parent_cdev, new_cdevs): @param parent_cdev: the disk to which we should add children @type new_cdevs: list of L{objects.Disk} @param new_cdevs: the list of children which we should add - @rtype: boolean - @return: the success of the operation + @rtype: None """ parent_bdev = _RecursiveFindBD(parent_cdev) @@ -1330,7 +1293,6 @@ def BlockdevAddchildren(parent_cdev, new_cdevs): if new_bdevs.count(None) > 0: _Fail("Can't find new device(s) to add: %s:%s", new_bdevs, new_cdevs) parent_bdev.AddChildren(new_bdevs) - return (True, None) def BlockdevRemovechildren(parent_cdev, new_cdevs): @@ -1340,8 +1302,7 @@ def BlockdevRemovechildren(parent_cdev, new_cdevs): @param parent_cdev: the disk from which we should remove children @type new_cdevs: list of L{objects.Disk} @param new_cdevs: the list of children which we should remove - @rtype: boolean - @return: the success of the operation + @rtype: None """ parent_bdev = _RecursiveFindBD(parent_cdev) @@ -1359,7 +1320,6 @@ def BlockdevRemovechildren(parent_cdev, new_cdevs): else: devs.append(rpath) parent_bdev.RemoveChildren(devs) - return (True, None) def BlockdevGetmirrorstatus(disks): @@ -1381,7 +1341,7 @@ def BlockdevGetmirrorstatus(disks): if rbd is None: _Fail("Can't find device %s", dsk) stats.append(rbd.CombinedSyncStatus()) - return True, stats + return stats def _RecursiveFindBD(disk): @@ -1401,7 +1361,7 @@ def _RecursiveFindBD(disk): for chdisk in disk.children: children.append(_RecursiveFindBD(chdisk)) - return bdev.FindDevice(disk.dev_type, disk.physical_id, children) + return bdev.FindDevice(disk.dev_type, disk.physical_id, children, disk.size) def BlockdevFind(disk): @@ -1422,8 +1382,8 @@ def BlockdevFind(disk): except errors.BlockDeviceError, err: _Fail("Failed to find device: %s", err, exc=True) if rbd is None: - return (True, None) - return (True, (rbd.dev_path, rbd.major, rbd.minor) + rbd.GetSyncStatus()) + return None + return (rbd.dev_path, rbd.major, rbd.minor) + rbd.GetSyncStatus() def UploadFile(file_name, data, mode, uid, gid, atime, mtime): @@ -1446,9 +1406,7 @@ def UploadFile(file_name, data, mode, uid, gid, atime, mtime): @param atime: the atime to set on the file (can be None) @type mtime: float @param mtime: the mtime to set on the file (can be None) - @rtype: boolean - @return: the success of the operation; errors are logged - in the node daemon log + @rtype: None """ if not os.path.isabs(file_name): @@ -1475,7 +1433,6 @@ def UploadFile(file_name, data, mode, uid, gid, atime, mtime): utils.WriteFile(file_name, data=raw_data, mode=mode, uid=uid, gid=gid, atime=atime, mtime=mtime) - return (True, "success") def WriteSsconfFiles(values): @@ -1515,11 +1472,9 @@ def _OSOndiskVersion(name, os_dir): @param name: the OS name we should look for @type os_dir: str @param os_dir: the directory inwhich we should look for the OS - @rtype: int or None - @return: - Either an integer denoting the version or None in the - case when this is not a valid OS name. - @raise errors.InvalidOS: if the OS cannot be found + @rtype: tuple + @return: tuple (status, data) with status denoting the validity and + data holding either the vaid versions or an error message """ api_file = os.path.sep.join([os_dir, "ganeti_api_version"]) @@ -1527,12 +1482,12 @@ def _OSOndiskVersion(name, os_dir): try: st = os.stat(api_file) except EnvironmentError, err: - raise errors.InvalidOS(name, os_dir, "'ganeti_api_version' file not" - " found (%s)" % _ErrnoOrStr(err)) + return False, ("Required file 'ganeti_api_version' file not" + " found under path %s: %s" % (os_dir, _ErrnoOrStr(err))) if not stat.S_ISREG(stat.S_IFMT(st.st_mode)): - raise errors.InvalidOS(name, os_dir, "'ganeti_api_version' file is not" - " a regular file") + return False, ("File 'ganeti_api_version' file at %s is not" + " a regular file" % os_dir) try: f = open(api_file) @@ -1541,17 +1496,17 @@ def _OSOndiskVersion(name, os_dir): finally: f.close() except EnvironmentError, err: - raise errors.InvalidOS(name, os_dir, "error while reading the" - " API version (%s)" % _ErrnoOrStr(err)) + return False, ("Error while reading the API version file at %s: %s" % + (api_file, _ErrnoOrStr(err))) api_versions = [version.strip() for version in api_versions] try: api_versions = [int(version) for version in api_versions] except (TypeError, ValueError), err: - raise errors.InvalidOS(name, os_dir, - "API version is not integer (%s)" % str(err)) + return False, ("API version(s) can't be converted to integer: %s" % + str(err)) - return api_versions + return True, api_versions def DiagnoseOS(top_dirs=None): @@ -1562,8 +1517,12 @@ def DiagnoseOS(top_dirs=None): search (if not given defaults to L{constants.OS_SEARCH_PATH}) @rtype: list of L{objects.OS} - @return: an OS object for each name in all the given - directories + @return: a list of tuples (name, path, status, diagnose) + for all (potential) OSes under all search paths, where: + - name is the (potential) OS name + - path is the full path to the OS + - status True/False is the validity of the OS + - diagnose is the error message for an invalid OS, otherwise empty """ if top_dirs is None: @@ -1575,46 +1534,49 @@ def DiagnoseOS(top_dirs=None): try: f_names = utils.ListVisibleFiles(dir_name) except EnvironmentError, err: - logging.exception("Can't list the OS directory %s", dir_name) + logging.exception("Can't list the OS directory %s: %s", dir_name, err) break for name in f_names: - try: - os_inst = OSFromDisk(name, base_dir=dir_name) - result.append(os_inst) - except errors.InvalidOS, err: - result.append(objects.OS.FromInvalidOS(err)) + os_path = os.path.sep.join([dir_name, name]) + status, os_inst = _TryOSFromDisk(name, base_dir=dir_name) + if status: + diagnose = "" + else: + diagnose = os_inst + result.append((name, os_path, status, diagnose)) return result -def OSFromDisk(name, base_dir=None): +def _TryOSFromDisk(name, base_dir=None): """Create an OS instance from disk. This function will return an OS instance if the given name is a - valid OS name. Otherwise, it will raise an appropriate - L{errors.InvalidOS} exception, detailing why this is not a valid OS. + valid OS name. @type base_dir: string @keyword base_dir: Base directory containing OS installations. Defaults to a search in all the OS_SEARCH_PATH dirs. - @rtype: L{objects.OS} - @return: the OS instance if we find a valid one - @raise errors.InvalidOS: if we don't find a valid OS + @rtype: tuple + @return: success and either the OS instance if we find a valid one, + or error message """ if base_dir is None: os_dir = utils.FindFile(name, constants.OS_SEARCH_PATH, os.path.isdir) if os_dir is None: - raise errors.InvalidOS(name, None, "OS dir not found in search path") + return False, "Directory for OS %s not found in search path" % name else: os_dir = os.path.sep.join([base_dir, name]) - api_versions = _OSOndiskVersion(name, os_dir) + status, api_versions = _OSOndiskVersion(name, os_dir) + if not status: + # push the error up + return status, api_versions if constants.OS_API_VERSION not in api_versions: - raise errors.InvalidOS(name, os_dir, "API version mismatch" - " (found %s want %s)" - % (api_versions, constants.OS_API_VERSION)) + return False, ("API version mismatch for path '%s': found %s, want %s." % + (os_dir, api_versions, constants.OS_API_VERSION)) # OS Scripts dictionary, we will populate it with the actual script names os_scripts = dict.fromkeys(constants.OS_SCRIPTS) @@ -1625,24 +1587,51 @@ def OSFromDisk(name, base_dir=None): try: st = os.stat(os_scripts[script]) except EnvironmentError, err: - raise errors.InvalidOS(name, os_dir, "'%s' script missing (%s)" % - (script, _ErrnoOrStr(err))) + return False, ("Script '%s' under path '%s' is missing (%s)" % + (script, os_dir, _ErrnoOrStr(err))) if stat.S_IMODE(st.st_mode) & stat.S_IXUSR != stat.S_IXUSR: - raise errors.InvalidOS(name, os_dir, "'%s' script not executable" % - script) + return False, ("Script '%s' under path '%s' is not executable" % + (script, os_dir)) if not stat.S_ISREG(stat.S_IFMT(st.st_mode)): - raise errors.InvalidOS(name, os_dir, "'%s' is not a regular file" % - script) + return False, ("Script '%s' under path '%s' is not a regular file" % + (script, os_dir)) + + os_obj = objects.OS(name=name, path=os_dir, + create_script=os_scripts[constants.OS_SCRIPT_CREATE], + export_script=os_scripts[constants.OS_SCRIPT_EXPORT], + import_script=os_scripts[constants.OS_SCRIPT_IMPORT], + rename_script=os_scripts[constants.OS_SCRIPT_RENAME], + api_versions=api_versions) + return True, os_obj + + +def OSFromDisk(name, base_dir=None): + """Create an OS instance from disk. + + This function will return an OS instance if the given name is a + valid OS name. Otherwise, it will raise an appropriate + L{RPCFail} exception, detailing why this is not a valid OS. + + This is just a wrapper over L{_TryOSFromDisk}, which doesn't raise + an exception but returns true/false status data. + + @type base_dir: string + @keyword base_dir: Base directory containing OS installations. + Defaults to a search in all the OS_SEARCH_PATH dirs. + @rtype: L{objects.OS} + @return: the OS instance if we find a valid one + @raise RPCFail: if we don't find a valid OS + + """ + status, payload = _TryOSFromDisk(name, base_dir) + + if not status: + _Fail(payload) + return payload - return objects.OS(name=name, path=os_dir, status=constants.OS_VALID_STATUS, - create_script=os_scripts[constants.OS_SCRIPT_CREATE], - export_script=os_scripts[constants.OS_SCRIPT_EXPORT], - import_script=os_scripts[constants.OS_SCRIPT_IMPORT], - rename_script=os_scripts[constants.OS_SCRIPT_RENAME], - api_versions=api_versions) def OSEnvironment(instance, debug=0): """Calculate the environment for an os script. @@ -1694,6 +1683,10 @@ def OSEnvironment(instance, debug=0): result['NIC_%d_FRONTEND_TYPE' % idx] = \ instance.hvparams[constants.HV_NIC_TYPE] + for source, kind in [(instance.beparams, "BE"), (instance.hvparams, "HV")]: + for key, value in source.items(): + result["INSTANCE_%s_%s" % (kind, key)] = str(value) + return result def BlockdevGrow(disk, amount): @@ -1712,15 +1705,13 @@ def BlockdevGrow(disk, amount): """ r_dev = _RecursiveFindBD(disk) if r_dev is None: - return False, "Cannot find block device %s" % (disk,) + _Fail("Cannot find block device %s", disk) try: r_dev.Grow(amount) except errors.BlockDeviceError, err: _Fail("Failed to grow block device: %s", err, exc=True) - return True, None - def BlockdevSnapshot(disk): """Create a snapshot copy of a block device. @@ -1748,7 +1739,7 @@ def BlockdevSnapshot(disk): r_dev = _RecursiveFindBD(disk) if r_dev is not None: # let's stay on the safe side and ask for the full size, for now - return True, r_dev.Snapshot(disk.size) + return r_dev.Snapshot(disk.size) else: _Fail("Cannot find block device %s", disk) else: @@ -1770,8 +1761,7 @@ def ExportSnapshot(disk, dest_node, instance, cluster_name, idx): @type idx: int @param idx: the index of the disk in the instance's disk list, used to export to the OS scripts environment - @rtype: boolean - @return: the success of the operation + @rtype: None """ export_env = OSEnvironment(instance) @@ -1818,8 +1808,6 @@ def ExportSnapshot(disk, dest_node, instance, cluster_name, idx): _Fail("OS snapshot export command '%s' returned error: %s" " output: %s", command, result.fail_reason, result.output) - return (True, None) - def FinalizeExport(instance, snap_disks): """Write out the export configuration information. @@ -1831,8 +1819,7 @@ def FinalizeExport(instance, snap_disks): @param snap_disks: list of snapshot block devices, which will be used to get the actual name of the dump file - @rtype: boolean - @return: the success of the operation + @rtype: None """ destdir = os.path.join(constants.EXPORT_DIR, instance.name + ".new") @@ -1884,8 +1871,6 @@ def FinalizeExport(instance, snap_disks): shutil.rmtree(finaldestdir, True) shutil.move(destdir, finaldestdir) - return True, None - def ExportInfo(dest): """Get export configuration information. @@ -1907,7 +1892,7 @@ def ExportInfo(dest): not config.has_section(constants.INISECT_INS)): _Fail("Export info file doesn't have the required fields") - return True, config.Dumps() + return config.Dumps() def ImportOSIntoInstance(instance, src_node, src_images, cluster_name): @@ -1955,8 +1940,7 @@ def ImportOSIntoInstance(instance, src_node, src_images, cluster_name): (idx, result.fail_reason, result.output[-100])) if final_result: - return False, "; ".join(final_result) - return True, None + _Fail("; ".join(final_result), log=False) def ListExports(): @@ -1967,9 +1951,9 @@ def ListExports(): """ if os.path.isdir(constants.EXPORT_DIR): - return True, utils.ListVisibleFiles(constants.EXPORT_DIR) + return utils.ListVisibleFiles(constants.EXPORT_DIR) else: - return False, "No exports directory" + _Fail("No exports directory") def RemoveExport(export): @@ -1977,8 +1961,7 @@ def RemoveExport(export): @type export: str @param export: the name of the export to remove - @rtype: boolean - @return: the success of the operation + @rtype: None """ target = os.path.join(constants.EXPORT_DIR, export) @@ -1988,8 +1971,6 @@ def RemoveExport(export): except EnvironmentError, err: _Fail("Error while removing the export: %s", err, exc=True) - return True, None - def BlockdevRename(devlist): """Rename a list of block devices. @@ -2028,7 +2009,8 @@ def BlockdevRename(devlist): (dev, unique_id, err)) logging.exception("Can't rename device '%s' to '%s'", dev, unique_id) result = False - return (result, "; ".join(msgs)) + if not result: + _Fail("; ".join(msgs)) def _TransformFileStorageDir(file_storage_dir): @@ -2049,10 +2031,8 @@ def _TransformFileStorageDir(file_storage_dir): base_file_storage_dir = cfg.GetFileStorageDir() if (not os.path.commonprefix([file_storage_dir, base_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 + _Fail("File storage directory '%s' is not under base file" + " storage directory '%s'", file_storage_dir, base_file_storage_dir) return file_storage_dir @@ -2068,22 +2048,16 @@ def CreateFileStorageDir(file_storage_dir): """ file_storage_dir = _TransformFileStorageDir(file_storage_dir) - result = True, - if not file_storage_dir: - result = False, + if os.path.exists(file_storage_dir): + if not os.path.isdir(file_storage_dir): + _Fail("Specified storage dir '%s' is not a directory", + file_storage_dir) else: - if os.path.exists(file_storage_dir): - if not os.path.isdir(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: - logging.error("Cannot create file storage directory '%s': %s", - file_storage_dir, err) - result = False, - return result + try: + os.makedirs(file_storage_dir, 0750) + except OSError, err: + _Fail("Cannot create file storage directory '%s': %s", + file_storage_dir, err, exc=True) def RemoveFileStorageDir(file_storage_dir): @@ -2099,22 +2073,16 @@ def RemoveFileStorageDir(file_storage_dir): """ file_storage_dir = _TransformFileStorageDir(file_storage_dir) - result = True, - if not file_storage_dir: - result = False, - else: - if os.path.exists(file_storage_dir): - if not os.path.isdir(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: - logging.exception("Cannot remove file storage directory '%s'", - file_storage_dir) - result = False, - return result + if os.path.exists(file_storage_dir): + if not os.path.isdir(file_storage_dir): + _Fail("Specified Storage directory '%s' is not a directory", + file_storage_dir) + # deletes dir only if empty, otherwise we want to fail the rpc call + try: + os.rmdir(file_storage_dir) + except OSError, err: + _Fail("Cannot remove file storage directory '%s': %s", + file_storage_dir, err) def RenameFileStorageDir(old_file_storage_dir, new_file_storage_dir): @@ -2131,46 +2099,37 @@ def RenameFileStorageDir(old_file_storage_dir, new_file_storage_dir): """ old_file_storage_dir = _TransformFileStorageDir(old_file_storage_dir) new_file_storage_dir = _TransformFileStorageDir(new_file_storage_dir) - result = True, - if not old_file_storage_dir or not new_file_storage_dir: - result = False, - else: - if not os.path.exists(new_file_storage_dir): - if os.path.isdir(old_file_storage_dir): - try: - os.rename(old_file_storage_dir, new_file_storage_dir) - except OSError, err: - logging.exception("Cannot rename '%s' to '%s'", - old_file_storage_dir, new_file_storage_dir) - result = False, - else: - logging.error("'%s' is not a directory", old_file_storage_dir) - result = False, + if not os.path.exists(new_file_storage_dir): + if os.path.isdir(old_file_storage_dir): + try: + os.rename(old_file_storage_dir, new_file_storage_dir) + except OSError, err: + _Fail("Cannot rename '%s' to '%s': %s", + old_file_storage_dir, new_file_storage_dir, err) else: - if os.path.exists(old_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 + _Fail("Specified storage dir '%s' is not a directory", + old_file_storage_dir) + else: + if os.path.exists(old_file_storage_dir): + _Fail("Cannot rename '%s' to '%s': both locations exist", + old_file_storage_dir, new_file_storage_dir) -def _IsJobQueueFile(file_name): +def _EnsureJobQueueFile(file_name): """Checks whether the given filename is in the queue directory. @type file_name: str @param file_name: the file name we should check - @rtype: boolean - @return: whether the file is under the queue directory + @rtype: None + @raises RPCFail: if the file is not valid """ 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 + _Fail("Passed job queue file '%s' does not belong to" + " the queue directory '%s'", file_name, queue_dir) def JobQueueUpdate(file_name, content): @@ -2187,14 +2146,11 @@ def JobQueueUpdate(file_name, content): @return: the success of the operation """ - if not _IsJobQueueFile(file_name): - return False + _EnsureJobQueueFile(file_name) # Write and replace the file atomically utils.WriteFile(file_name, data=_Decompress(content)) - return True - def JobQueueRename(old, new): """Renames a job queue file. @@ -2205,17 +2161,15 @@ def JobQueueRename(old, new): @param old: the old (actual) file name @type new: str @param new: the desired file name - @rtype: boolean - @return: the success of the operation + @rtype: tuple + @return: the success of the operation and payload """ - if not (_IsJobQueueFile(old) and _IsJobQueueFile(new)): - return False + _EnsureJobQueueFile(old) + _EnsureJobQueueFile(new) utils.RenameFile(old, new, mkdir=True) - return True - def JobQueueSetDrainFlag(drain_flag): """Set the drain flag for the queue. @@ -2224,8 +2178,8 @@ def JobQueueSetDrainFlag(drain_flag): @type drain_flag: boolean @param drain_flag: if True, will set the drain flag, otherwise reset it. - @rtype: boolean - @return: always True + @rtype: truple + @return: always True, None @warning: the function always returns True """ @@ -2234,8 +2188,6 @@ def JobQueueSetDrainFlag(drain_flag): else: utils.RemoveFile(constants.JOB_QUEUE_DRAIN_FILE) - return True - def BlockdevClose(instance_name, disks): """Closes the given block devices. @@ -2268,11 +2220,10 @@ def BlockdevClose(instance_name, disks): except errors.BlockDeviceError, err: msg.append(str(err)) if msg: - return (False, "Can't make devices secondary: %s" % ",".join(msg)) + _Fail("Can't make devices secondary: %s", ",".join(msg)) else: if instance_name: _RemoveBlockDevLinks(instance_name, disks) - return (True, "All devices secondary") def ValidateHVParams(hvname, hvparams): @@ -2282,19 +2233,14 @@ def ValidateHVParams(hvname, hvparams): @param hvname: the hypervisor name @type hvparams: dict @param hvparams: the hypervisor parameters to be validated - @rtype: tuple (success, message) - @return: a tuple of success and message, where success - indicates the succes of the operation, and message - which will contain the error details in case we - failed + @rtype: None """ try: hv_type = hypervisor.GetHypervisor(hvname) hv_type.ValidateParameters(hvparams) - return (True, "Validation passed") except errors.HypervisorError, err: - return (False, str(err)) + _Fail(str(err), log=False) def DemoteFromMC(): @@ -2304,17 +2250,16 @@ def DemoteFromMC(): # try to ensure we're not the master by mistake master, myself = ssconf.GetMasterAndMyself() if master == myself: - return (False, "ssconf status shows I'm the master node, will not demote") + _Fail("ssconf status shows I'm the master node, will not demote") pid_file = utils.DaemonPidFileName(constants.MASTERD_PID) if utils.IsProcessAlive(utils.ReadPidFile(pid_file)): - return (False, "The master daemon is running, will not demote") + _Fail("The master daemon is running, will not demote") try: utils.CreateBackup(constants.CLUSTER_CONF_FILE) except EnvironmentError, err: if err.errno != errno.ENOENT: - return (False, "Error while backing up cluster file: %s" % str(err)) + _Fail("Error while backing up cluster file: %s", err, exc=True) utils.RemoveFile(constants.CLUSTER_CONF_FILE) - return (True, "Done") def _FindDisks(nodes_ip, disks): @@ -2331,18 +2276,16 @@ def _FindDisks(nodes_ip, disks): for cf in disks: rd = _RecursiveFindBD(cf) if rd is None: - return (False, "Can't find device %s" % cf) + _Fail("Can't find device %s", cf) bdevs.append(rd) - return (True, bdevs) + return bdevs def DrbdDisconnectNet(nodes_ip, disks): """Disconnects the network on a list of drbd devices. """ - status, bdevs = _FindDisks(nodes_ip, disks) - if not status: - return status, bdevs + bdevs = _FindDisks(nodes_ip, disks) # disconnect disks for rd in bdevs: @@ -2351,16 +2294,13 @@ def DrbdDisconnectNet(nodes_ip, disks): except errors.BlockDeviceError, err: _Fail("Can't change network configuration to standalone mode: %s", err, exc=True) - return (True, "All disks are now disconnected") def DrbdAttachNet(nodes_ip, disks, instance_name, multimaster): """Attaches the network on a list of drbd devices. """ - status, bdevs = _FindDisks(nodes_ip, disks) - if not status: - return status, bdevs + bdevs = _FindDisks(nodes_ip, disks) if multimaster: for idx, rd in enumerate(bdevs): @@ -2394,7 +2334,7 @@ def DrbdAttachNet(nodes_ip, disks, instance_name, multimaster): # standalone, even though this should not happen with the # new staged way of changing disk configs try: - rd.ReAttachNet(multimaster) + rd.AttachNet(multimaster) except errors.BlockDeviceError, err: _Fail("Can't change network configuration: %s", err) if all_connected: @@ -2402,7 +2342,7 @@ def DrbdAttachNet(nodes_ip, disks, instance_name, multimaster): time.sleep(sleep_time) sleep_time = min(5, sleep_time * 1.5) if not all_connected: - return (False, "Timeout in disk reconnecting") + _Fail("Timeout in disk reconnecting") if multimaster: # change to primary mode for rd in bdevs: @@ -2410,33 +2350,25 @@ def DrbdAttachNet(nodes_ip, disks, instance_name, multimaster): rd.Open() except errors.BlockDeviceError, err: _Fail("Can't change to primary mode: %s", err) - if multimaster: - msg = "multi-master and primary" - else: - msg = "single-master" - return (True, "Disks are now configured as %s" % msg) def DrbdWaitSync(nodes_ip, disks): """Wait until DRBDs have synchronized. """ - status, bdevs = _FindDisks(nodes_ip, disks) - if not status: - return status, bdevs + bdevs = _FindDisks(nodes_ip, disks) min_resync = 100 alldone = True - failure = False for rd in bdevs: stats = rd.GetProcStatus() if not (stats.is_connected or stats.is_in_resync): - failure = True - break + _Fail("DRBD device %s is not in sync: stats=%s", rd, stats) alldone = alldone and (not stats.is_in_resync) if stats.sync_percent is not None: min_resync = min(min_resync, stats.sync_percent) - return (not failure, (alldone, min_resync)) + + return (alldone, min_resync) def PowercycleNode(hypervisor_type): @@ -2449,11 +2381,11 @@ def PowercycleNode(hypervisor_type): hyper = hypervisor.GetHypervisor(hypervisor_type) try: pid = os.fork() - except OSError, err: + except OSError: # if we can't fork, we'll pretend that we're in the child process pid = 0 if pid > 0: - return (True, "Reboot scheduled in 5 seconds") + return "Reboot scheduled in 5 seconds" time.sleep(5) hyper.PowercycleNode() @@ -2556,14 +2488,15 @@ class HooksRunner(object): elif phase == constants.HOOKS_PHASE_POST: suffix = "post" else: - raise errors.ProgrammerError("Unknown hooks phase: '%s'" % phase) + _Fail("Unknown hooks phase '%s'", phase) + rr = [] subdir = "%s-%s.d" % (hpath, suffix) dir_name = "%s/%s" % (self._BASE_DIR, subdir) try: dir_contents = utils.ListVisibleFiles(dir_name) - except OSError, err: + except OSError: # FIXME: must log output in case of failures return rr @@ -2603,17 +2536,15 @@ class IAllocatorRunner(object): @param idata: the allocator input data @rtype: tuple - @return: four element tuple of: - - run status (one of the IARUN_ constants) - - stdout - - stderr - - fail reason (as from L{utils.RunResult}) + @return: two element tuple of: + - status + - either error message or stdout of allocator (for success) """ alloc_script = utils.FindFile(name, constants.IALLOCATOR_SEARCH_PATH, os.path.isfile) if alloc_script is None: - return (constants.IARUN_NOTFOUND, None, None, None) + _Fail("iallocator module '%s' not found in the search path", name) fd, fin_name = tempfile.mkstemp(prefix="ganeti-iallocator.") try: @@ -2621,12 +2552,12 @@ class IAllocatorRunner(object): os.close(fd) result = utils.RunCmd([alloc_script, fin_name]) if result.failed: - return (constants.IARUN_FAILURE, result.stdout, result.stderr, - result.fail_reason) + _Fail("iallocator module '%s' failed: %s, output '%s'", + name, result.fail_reason, result.output) finally: os.unlink(fin_name) - return (constants.IARUN_SUCCESS, result.stdout, result.stderr, None) + return result.stdout class DevCacheManager(object): @@ -2687,7 +2618,7 @@ class DevCacheManager(object): try: utils.WriteFile(fpath, data=fdata) except EnvironmentError, err: - logging.exception("Can't update bdev cache for %s", dev_path) + logging.exception("Can't update bdev cache for %s: %s", dev_path, err) @classmethod def RemoveCache(cls, dev_path): @@ -2709,4 +2640,4 @@ class DevCacheManager(object): try: utils.RemoveFile(fpath) except EnvironmentError, err: - logging.exception("Can't update bdev cache for %s", dev_path) + logging.exception("Can't update bdev cache for %s: %s", dev_path, err)