X-Git-Url: https://code.grnet.gr/git/ganeti-local/blobdiff_plain/b954f097b3f164f083301b12941f14ccf47c2a07..b82d71825cbace5d42639ebae3860e4a71ac93dd:/lib/backend.py diff --git a/lib/backend.py b/lib/backend.py index c902718..a89c55a 100644 --- a/lib/backend.py +++ b/lib/backend.py @@ -1,7 +1,7 @@ # # -# Copyright (C) 2006, 2007, 2008, 2009, 2010, 2011, 2012 Google Inc. +# Copyright (C) 2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013 Google Inc. # # This program is free software; you can redistribute it and/or modify # it under the terms of the GNU General Public License as published by @@ -54,17 +54,21 @@ from ganeti import utils from ganeti import ssh from ganeti import hypervisor from ganeti import constants -from ganeti import bdev +from ganeti.storage import bdev +from ganeti.storage import drbd +from ganeti.storage import filestorage from ganeti import objects from ganeti import ssconf from ganeti import serializer from ganeti import netutils from ganeti import runtime -from ganeti import mcpu from ganeti import compat from ganeti import pathutils from ganeti import vcluster from ganeti import ht +from ganeti.storage.base import BlockDev +from ganeti.storage.drbd import DRBD8 +from ganeti import hooksmaster _BOOT_ID_PATH = "/proc/sys/kernel/random/boot_id" @@ -88,15 +92,15 @@ _LVSLINE_REGEX = re.compile("^ *([^|]+)\|([^|]+)\|([0-9.]+)\|([^|]{6,})\|?$") _MASTER_START = "start" _MASTER_STOP = "stop" -#: Maximum file permissions for remote command directory and executables +#: Maximum file permissions for restricted command directory and executables _RCMD_MAX_MODE = (stat.S_IRWXU | stat.S_IRGRP | stat.S_IXGRP | stat.S_IROTH | stat.S_IXOTH) -#: Delay before returning an error for remote commands +#: Delay before returning an error for restricted commands _RCMD_INVALID_DELAY = 10 -#: How long to wait to acquire lock for remote commands (shorter than +#: How long to wait to acquire lock for restricted commands (shorter than #: L{_RCMD_INVALID_DELAY}) to reduce blockage of noded forks when many #: command requests arrive _RCMD_LOCK_TIMEOUT = _RCMD_INVALID_DELAY * 0.8 @@ -110,6 +114,34 @@ class RPCFail(Exception): """ +def _GetInstReasonFilename(instance_name): + """Path of the file containing the reason of the instance status change. + + @type instance_name: string + @param instance_name: The name of the instance + @rtype: string + @return: The path of the file + + """ + return utils.PathJoin(pathutils.INSTANCE_REASON_DIR, instance_name) + + +def _StoreInstReasonTrail(instance_name, trail): + """Serialize a reason trail related to an instance change of state to file. + + The exact location of the file depends on the name of the instance and on + the configuration of the Ganeti cluster defined at deploy time. + + @type instance_name: string + @param instance_name: The name of the instance + @rtype: None + + """ + json = serializer.DumpJson(trail) + filename = _GetInstReasonFilename(instance_name) + utils.WriteFile(filename, data=json) + + def _Fail(msg, *args, **kwargs): """Log an error and the raise an RPCFail exception. @@ -297,10 +329,10 @@ def RunLocalHooks(hook_opcode, hooks_path, env_builder_fn): cfg = _GetConfig() hr = HooksRunner() - hm = mcpu.HooksMaster(hook_opcode, hooks_path, nodes, hr.RunLocalHooks, - None, env_fn, logging.warning, cfg.GetClusterName(), - cfg.GetMasterNode()) - + hm = hooksmaster.HooksMaster(hook_opcode, hooks_path, nodes, + hr.RunLocalHooks, None, env_fn, + logging.warning, cfg.GetClusterName(), + cfg.GetMasterNode()) hm.RunPhase(constants.HOOKS_PHASE_PRE) result = fn(*args, **kwargs) hm.RunPhase(constants.HOOKS_PHASE_POST) @@ -358,8 +390,8 @@ def _RunMasterSetupScript(master_params, action, use_external_mip_script): result = utils.RunCmd([setup_script, action], env=env, reset_env=True) if result.failed: - _Fail("Failed to %s the master IP. Script return value: %s" % - (action, result.exit_code), log=True) + _Fail("Failed to %s the master IP. Script return value: %s, output: '%s'" % + (action, result.exit_code, result.output), log=True) @RunLocalHooks(constants.FAKE_OP_MASTER_TURNUP, "master-ip-turnup", @@ -555,13 +587,39 @@ def _GetVgInfo(name, excl_stor): vg_size = None return { + "type": constants.ST_LVM_VG, + "name": name, + "storage_free": vg_free, + "storage_size": vg_size, + } + + +def _GetVgSpindlesInfo(name, excl_stor): + """Retrieves information about spindles in an LVM volume group. + + @type name: string + @param name: VG name + @type excl_stor: bool + @param excl_stor: exclusive storage + @rtype: dict + @return: dictionary whose keys are "name", "vg_free", "vg_size" for VG name, + free spindles, total spindles respectively + + """ + if excl_stor: + (vg_free, vg_size) = bdev.LogicalVolume.GetVgSpindlesInfo(name) + else: + vg_free = 0 + vg_size = 0 + return { + "type": constants.ST_LVM_PV, "name": name, - "vg_free": vg_free, - "vg_size": vg_size, + "storage_free": vg_free, + "storage_size": vg_size, } -def _GetHvInfo(name): +def _GetHvInfo(name, hvparams, get_hv_fn=hypervisor.GetHypervisor): """Retrieves node information from a hypervisor. The information returned depends on the hypervisor. Common items: @@ -573,8 +631,29 @@ def _GetHvInfo(name): - memory_total is the total number of ram in MiB - hv_version: the hypervisor version, if available + @type hvparams: dict of string + @param hvparams: the hypervisor's hvparams + """ - return hypervisor.GetHypervisor(name).GetNodeInfo() + return get_hv_fn(name).GetNodeInfo(hvparams=hvparams) + + +def _GetHvInfoAll(hv_specs, get_hv_fn=hypervisor.GetHypervisor): + """Retrieves node information for all hypervisors. + + See C{_GetHvInfo} for information on the output. + + @type hv_specs: list of pairs (string, dict of strings) + @param hv_specs: list of pairs of a hypervisor's name and its hvparams + + """ + if hv_specs is None: + return None + + result = [] + for hvname, hvparams in hv_specs: + result.append(_GetHvInfo(hvname, hvparams, get_hv_fn)) + return result def _GetNamedNodeInfo(names, fn): @@ -589,13 +668,15 @@ def _GetNamedNodeInfo(names, fn): return map(fn, names) -def GetNodeInfo(vg_names, hv_names, excl_stor): +def GetNodeInfo(storage_units, hv_specs, excl_stor): """Gives back a hash with different information about the node. - @type vg_names: list of string - @param vg_names: Names of the volume groups to ask for disk space information - @type hv_names: list of string - @param hv_names: Names of the hypervisors to ask for node information + @type storage_units: list of pairs (string, string) + @param storage_units: List of pairs (storage unit, identifier) to ask for disk + space information. In case of lvm-vg, the identifier is + the VG name. + @type hv_specs: list of pairs (string, dict of strings) + @param hv_specs: list of pairs of a hypervisor's name and its hvparams @type excl_stor: boolean @param excl_stor: Whether exclusive_storage is active @rtype: tuple; (string, None/dict, None/dict) @@ -604,13 +685,190 @@ def GetNodeInfo(vg_names, hv_names, excl_stor): """ bootid = utils.ReadFile(_BOOT_ID_PATH, size=128).rstrip("\n") - vg_info = _GetNamedNodeInfo(vg_names, (lambda vg: _GetVgInfo(vg, excl_stor))) - hv_info = _GetNamedNodeInfo(hv_names, _GetHvInfo) + storage_info = _GetNamedNodeInfo( + storage_units, + (lambda storage_unit: _ApplyStorageInfoFunction(storage_unit[0], + storage_unit[1], + excl_stor))) + hv_info = _GetHvInfoAll(hv_specs) + return (bootid, storage_info, hv_info) + + +# pylint: disable=W0613 +def _GetFileStorageSpaceInfo(path, *args): + """Wrapper around filestorage.GetSpaceInfo. + + The purpose of this wrapper is to call filestorage.GetFileStorageSpaceInfo + and ignore the *args parameter to not leak it into the filestorage + module's code. + + @see: C{filestorage.GetFileStorageSpaceInfo} for description of the + parameters. + + """ + return filestorage.GetFileStorageSpaceInfo(path) + + +# FIXME: implement storage reporting for all missing storage types. +_STORAGE_TYPE_INFO_FN = { + constants.ST_BLOCK: None, + constants.ST_DISKLESS: None, + constants.ST_EXT: None, + constants.ST_FILE: _GetFileStorageSpaceInfo, + constants.ST_LVM_PV: _GetVgSpindlesInfo, + constants.ST_LVM_VG: _GetVgInfo, + constants.ST_RADOS: None, +} + + +def _ApplyStorageInfoFunction(storage_type, storage_key, *args): + """Looks up and applies the correct function to calculate free and total + storage for the given storage type. + + @type storage_type: string + @param storage_type: the storage type for which the storage shall be reported. + @type storage_key: string + @param storage_key: identifier of a storage unit, e.g. the volume group name + of an LVM storage unit + @type args: any + @param args: various parameters that can be used for storage reporting. These + parameters and their semantics vary from storage type to storage type and + are just propagated in this function. + @return: the results of the application of the storage space function (see + _STORAGE_TYPE_INFO_FN) if storage space reporting is implemented for that + storage type + @raises NotImplementedError: for storage types who don't support space + reporting yet + """ + fn = _STORAGE_TYPE_INFO_FN[storage_type] + if fn is not None: + return fn(storage_key, *args) + else: + raise NotImplementedError + + +def _CheckExclusivePvs(pvi_list): + """Check that PVs are not shared among LVs + + @type pvi_list: list of L{objects.LvmPvInfo} objects + @param pvi_list: information about the PVs + + @rtype: list of tuples (string, list of strings) + @return: offending volumes, as tuples: (pv_name, [lv1_name, lv2_name...]) + + """ + res = [] + for pvi in pvi_list: + if len(pvi.lv_list) > 1: + res.append((pvi.name, pvi.lv_list)) + return res + + +def _VerifyHypervisors(what, vm_capable, result, all_hvparams, + get_hv_fn=hypervisor.GetHypervisor): + """Verifies the hypervisor. Appends the results to the 'results' list. + + @type what: C{dict} + @param what: a dictionary of things to check + @type vm_capable: boolean + @param vm_capable: whether or not this node is vm capable + @type result: dict + @param result: dictionary of verification results; results of the + verifications in this function will be added here + @type all_hvparams: dict of dict of string + @param all_hvparams: dictionary mapping hypervisor names to hvparams + @type get_hv_fn: function + @param get_hv_fn: function to retrieve the hypervisor, to improve testability + + """ + if not vm_capable: + return + + if constants.NV_HYPERVISOR in what: + result[constants.NV_HYPERVISOR] = {} + for hv_name in what[constants.NV_HYPERVISOR]: + hvparams = all_hvparams[hv_name] + try: + val = get_hv_fn(hv_name).Verify(hvparams=hvparams) + except errors.HypervisorError, err: + val = "Error while checking hypervisor: %s" % str(err) + result[constants.NV_HYPERVISOR][hv_name] = val + + +def _VerifyHvparams(what, vm_capable, result, + get_hv_fn=hypervisor.GetHypervisor): + """Verifies the hvparams. Appends the results to the 'results' list. + + @type what: C{dict} + @param what: a dictionary of things to check + @type vm_capable: boolean + @param vm_capable: whether or not this node is vm capable + @type result: dict + @param result: dictionary of verification results; results of the + verifications in this function will be added here + @type get_hv_fn: function + @param get_hv_fn: function to retrieve the hypervisor, to improve testability + + """ + if not vm_capable: + return + + if constants.NV_HVPARAMS in what: + result[constants.NV_HVPARAMS] = [] + for source, hv_name, hvparms in what[constants.NV_HVPARAMS]: + try: + logging.info("Validating hv %s, %s", hv_name, hvparms) + get_hv_fn(hv_name).ValidateParameters(hvparms) + except errors.HypervisorError, err: + result[constants.NV_HVPARAMS].append((source, hv_name, str(err))) + + +def _VerifyInstanceList(what, vm_capable, result, all_hvparams): + """Verifies the instance list. + + @type what: C{dict} + @param what: a dictionary of things to check + @type vm_capable: boolean + @param vm_capable: whether or not this node is vm capable + @type result: dict + @param result: dictionary of verification results; results of the + verifications in this function will be added here + @type all_hvparams: dict of dict of string + @param all_hvparams: dictionary mapping hypervisor names to hvparams + + """ + if constants.NV_INSTANCELIST in what and vm_capable: + # GetInstanceList can fail + try: + val = GetInstanceList(what[constants.NV_INSTANCELIST], + all_hvparams=all_hvparams) + except RPCFail, err: + val = str(err) + result[constants.NV_INSTANCELIST] = val - return (bootid, vg_info, hv_info) +def _VerifyNodeInfo(what, vm_capable, result, all_hvparams): + """Verifies the node info. -def VerifyNode(what, cluster_name): + @type what: C{dict} + @param what: a dictionary of things to check + @type vm_capable: boolean + @param vm_capable: whether or not this node is vm capable + @type result: dict + @param result: dictionary of verification results; results of the + verifications in this function will be added here + @type all_hvparams: dict of dict of string + @param all_hvparams: dictionary mapping hypervisor names to hvparams + + """ + if constants.NV_HVINFO in what and vm_capable: + hvname = what[constants.NV_HVINFO] + hyper = hypervisor.GetHypervisor(hvname) + hvparams = all_hvparams[hvname] + result[constants.NV_HVINFO] = hyper.GetNodeInfo(hvparams=hvparams) + + +def VerifyNode(what, cluster_name, all_hvparams): """Verify the status of the local node. Based on the input L{what} parameter, various checks are done on the @@ -634,6 +892,10 @@ def VerifyNode(what, cluster_name): - node-net-test: list of nodes we should check node daemon port connectivity with - hypervisor: list with hypervisors to run the verify for + @type cluster_name: string + @param cluster_name: the cluster's name + @type all_hvparams: dict of dict of strings + @param all_hvparams: a dictionary mapping hypervisor names to hvparams @rtype: dict @return: a dictionary with the same keys as the input dict, and values representing the result of the checks @@ -644,23 +906,8 @@ def VerifyNode(what, cluster_name): port = netutils.GetDaemonPort(constants.NODED) vm_capable = my_name not in what.get(constants.NV_VMNODES, []) - if constants.NV_HYPERVISOR in what and vm_capable: - result[constants.NV_HYPERVISOR] = tmp = {} - for hv_name in what[constants.NV_HYPERVISOR]: - try: - val = hypervisor.GetHypervisor(hv_name).Verify() - except errors.HypervisorError, err: - val = "Error while checking hypervisor: %s" % str(err) - tmp[hv_name] = val - - if constants.NV_HVPARAMS in what and vm_capable: - result[constants.NV_HVPARAMS] = tmp = [] - for source, hv_name, hvparms in what[constants.NV_HVPARAMS]: - try: - logging.info("Validating hv %s, %s", hv_name, hvparms) - hypervisor.GetHypervisor(hv_name).ValidateParameters(hvparms) - except errors.HypervisorError, err: - tmp.append((source, hv_name, str(err))) + _VerifyHypervisors(what, vm_capable, result, all_hvparams) + _VerifyHvparams(what, vm_capable, result) if constants.NV_FILELIST in what: fingerprints = utils.FingerprintFiles(map(vcluster.LocalizeVirtualPath, @@ -752,33 +999,40 @@ def VerifyNode(what, cluster_name): val = str(err) result[constants.NV_LVLIST] = val - if constants.NV_INSTANCELIST in what and vm_capable: - # GetInstanceList can fail - try: - val = GetInstanceList(what[constants.NV_INSTANCELIST]) - except RPCFail, err: - val = str(err) - result[constants.NV_INSTANCELIST] = val + _VerifyInstanceList(what, vm_capable, result, all_hvparams) if constants.NV_VGLIST in what and vm_capable: result[constants.NV_VGLIST] = utils.ListVolumeGroups() if constants.NV_PVLIST in what and vm_capable: + check_exclusive_pvs = constants.NV_EXCLUSIVEPVS in what val = bdev.LogicalVolume.GetPVInfo(what[constants.NV_PVLIST], - filter_allocatable=False) + filter_allocatable=False, + include_lvs=check_exclusive_pvs) + if check_exclusive_pvs: + result[constants.NV_EXCLUSIVEPVS] = _CheckExclusivePvs(val) + for pvi in val: + # Avoid sending useless data on the wire + pvi.lv_list = [] result[constants.NV_PVLIST] = map(objects.LvmPvInfo.ToDict, val) if constants.NV_VERSION in what: result[constants.NV_VERSION] = (constants.PROTOCOL_VERSION, constants.RELEASE_VERSION) - if constants.NV_HVINFO in what and vm_capable: - hyper = hypervisor.GetHypervisor(what[constants.NV_HVINFO]) - result[constants.NV_HVINFO] = hyper.GetNodeInfo() + _VerifyNodeInfo(what, vm_capable, result, all_hvparams) + + if constants.NV_DRBDVERSION in what and vm_capable: + try: + drbd_version = DRBD8.GetProcInfo().GetVersionString() + except errors.BlockDeviceError, err: + logging.warning("Can't get DRBD version", exc_info=True) + drbd_version = str(err) + result[constants.NV_DRBDVERSION] = drbd_version if constants.NV_DRBDLIST in what and vm_capable: try: - used_minors = bdev.DRBD8.GetUsedDevs().keys() + used_minors = drbd.DRBD8.GetUsedDevs() except errors.BlockDeviceError, err: logging.warning("Can't get used minors list", exc_info=True) used_minors = str(err) @@ -787,7 +1041,7 @@ def VerifyNode(what, cluster_name): if constants.NV_DRBDHELPER in what and vm_capable: status = True try: - payload = bdev.BaseDRBD.GetUsermodeHelper() + payload = drbd.DRBD8.GetUsermodeHelper() except errors.BlockDeviceError, err: logging.error("Can't get DRBD usermode helper: %s", str(err)) status = False @@ -981,11 +1235,47 @@ def BridgesExist(bridges_list): _Fail("Missing bridges %s", utils.CommaJoin(missing)) -def GetInstanceList(hypervisor_list): +def GetInstanceListForHypervisor(hname, hvparams=None, + get_hv_fn=hypervisor.GetHypervisor): + """Provides a list of instances of the given hypervisor. + + @type hname: string + @param hname: name of the hypervisor + @type hvparams: dict of strings + @param hvparams: hypervisor parameters for the given hypervisor + @type get_hv_fn: function + @param get_hv_fn: function that returns a hypervisor for the given hypervisor + name; optional parameter to increase testability + + @rtype: list + @return: a list of all running instances on the current node + - instance1.example.com + - instance2.example.com + + """ + results = [] + try: + hv = get_hv_fn(hname) + names = hv.ListInstances(hvparams=hvparams) + results.extend(names) + except errors.HypervisorError, err: + _Fail("Error enumerating instances (hypervisor %s): %s", + hname, err, exc=True) + return results + + +def GetInstanceList(hypervisor_list, all_hvparams=None, + get_hv_fn=hypervisor.GetHypervisor): """Provides a list of instances. @type hypervisor_list: list @param hypervisor_list: the list of hypervisors to query information + @type all_hvparams: dict of dict of strings + @param all_hvparams: a dictionary mapping hypervisor types to respective + cluster-wide hypervisor parameters + @type get_hv_fn: function + @param get_hv_fn: function that returns a hypervisor for the given hypervisor + name; optional parameter to increase testability @rtype: list @return: a list of all running instances on the current node @@ -995,23 +1285,21 @@ def GetInstanceList(hypervisor_list): """ results = [] for hname in hypervisor_list: - try: - names = hypervisor.GetHypervisor(hname).ListInstances() - results.extend(names) - except errors.HypervisorError, err: - _Fail("Error enumerating instances (hypervisor %s): %s", - hname, err, exc=True) - + hvparams = all_hvparams[hname] + results.extend(GetInstanceListForHypervisor(hname, hvparams=hvparams, + get_hv_fn=get_hv_fn)) return results -def GetInstanceInfo(instance, hname): +def GetInstanceInfo(instance, hname, hvparams=None): """Gives back the information about an instance as a dictionary. @type instance: string @param instance: the instance name @type hname: string @param hname: the hypervisor type of the instance + @type hvparams: dict of strings + @param hvparams: the instance's hvparams @rtype: dict @return: dictionary with the following keys: @@ -1023,7 +1311,8 @@ def GetInstanceInfo(instance, hname): """ output = {} - iinfo = hypervisor.GetHypervisor(hname).GetInstanceInfo(instance) + iinfo = hypervisor.GetHypervisor(hname).GetInstanceInfo(instance, + hvparams=hvparams) if iinfo is not None: output["memory"] = iinfo[2] output["vcpus"] = iinfo[3] @@ -1034,7 +1323,7 @@ def GetInstanceInfo(instance, hname): def GetInstanceMigratable(instance): - """Gives whether an instance can be migrated. + """Computes whether an instance can be migrated. @type instance: L{objects.Instance} @param instance: object representing the instance to be checked. @@ -1047,7 +1336,7 @@ def GetInstanceMigratable(instance): """ hyper = hypervisor.GetHypervisor(instance.hypervisor) iname = instance.name - if iname not in hyper.ListInstances(): + if iname not in hyper.ListInstances(instance.hvparams): _Fail("Instance %s is not running", iname) for idx in range(len(instance.disks)): @@ -1057,7 +1346,7 @@ def GetInstanceMigratable(instance): iname, link_name, idx) -def GetAllInstancesInfo(hypervisor_list): +def GetAllInstancesInfo(hypervisor_list, all_hvparams): """Gather data about all instances. This is the equivalent of L{GetInstanceInfo}, except that it @@ -1066,6 +1355,8 @@ def GetAllInstancesInfo(hypervisor_list): @type hypervisor_list: list @param hypervisor_list: list of hypervisors to query for instance data + @type all_hvparams: dict of dict of strings + @param all_hvparams: mapping of hypervisor names to hvparams @rtype: dict @return: dictionary of instance: data, with data having the following keys: @@ -1078,7 +1369,8 @@ def GetAllInstancesInfo(hypervisor_list): output = {} for hname in hypervisor_list: - iinfo = hypervisor.GetHypervisor(hname).GetAllInstancesInfo() + hvparams = all_hvparams[hname] + iinfo = hypervisor.GetHypervisor(hname).GetAllInstancesInfo(hvparams) if iinfo: for name, _, memory, vcpus, state, times in iinfo: value = { @@ -1193,9 +1485,16 @@ def RunRenameInstance(instance, old_name, debug): " log file:\n%s", result.fail_reason, "\n".join(lines), log=False) -def _GetBlockDevSymlinkPath(instance_name, idx): - return utils.PathJoin(pathutils.DISK_LINKS_DIR, "%s%s%d" % - (instance_name, constants.DISK_SEPARATOR, idx)) +def _GetBlockDevSymlinkPath(instance_name, idx, _dir=None): + """Returns symlink path for block device. + + """ + if _dir is None: + _dir = pathutils.DISK_LINKS_DIR + + return utils.PathJoin(_dir, + ("%s%s%s" % + (instance_name, constants.DISK_SEPARATOR, idx))) def _SymlinkBlockDev(instance_name, device_path, idx): @@ -1269,17 +1568,22 @@ def _GatherAndLinkBlockDevs(instance): return block_devices -def StartInstance(instance, startup_paused): +def StartInstance(instance, startup_paused, reason, store_reason=True): """Start an instance. @type instance: L{objects.Instance} @param instance: the instance object @type startup_paused: bool @param instance: pause instance at startup? + @type reason: list of reasons + @param reason: the reason trail for this startup + @type store_reason: boolean + @param store_reason: whether to store the shutdown reason trail on file @rtype: None """ - running_instances = GetInstanceList([instance.hypervisor]) + running_instances = GetInstanceListForHypervisor(instance.hypervisor, + instance.hvparams) if instance.name in running_instances: logging.info("Instance %s already running, not starting", instance.name) @@ -1289,6 +1593,8 @@ def StartInstance(instance, startup_paused): block_devices = _GatherAndLinkBlockDevs(instance) hyper = hypervisor.GetHypervisor(instance.hypervisor) hyper.StartInstance(instance, block_devices, startup_paused) + if store_reason: + _StoreInstReasonTrail(instance.name, reason) except errors.BlockDeviceError, err: _Fail("Block device error: %s", err, exc=True) except errors.HypervisorError, err: @@ -1296,7 +1602,7 @@ def StartInstance(instance, startup_paused): _Fail("Hypervisor error: %s", err, exc=True) -def InstanceShutdown(instance, timeout): +def InstanceShutdown(instance, timeout, reason, store_reason=True): """Shut an instance down. @note: this functions uses polling with a hardcoded timeout. @@ -1305,6 +1611,10 @@ def InstanceShutdown(instance, timeout): @param instance: the instance object @type timeout: integer @param timeout: maximum timeout for soft shutdown + @type reason: list of reasons + @param reason: the reason trail for this shutdown + @type store_reason: boolean + @param store_reason: whether to store the shutdown reason trail on file @rtype: None """ @@ -1312,7 +1622,7 @@ def InstanceShutdown(instance, timeout): hyper = hypervisor.GetHypervisor(hv_name) iname = instance.name - if instance.name not in hyper.ListInstances(): + if instance.name not in hyper.ListInstances(instance.hvparams): logging.info("Instance %s not running, doing nothing", iname) return @@ -1321,13 +1631,15 @@ def InstanceShutdown(instance, timeout): self.tried_once = False def __call__(self): - if iname not in hyper.ListInstances(): + if iname not in hyper.ListInstances(instance.hvparams): return try: hyper.StopInstance(instance, retry=self.tried_once) + if store_reason: + _StoreInstReasonTrail(instance.name, reason) except errors.HypervisorError, err: - if iname not in hyper.ListInstances(): + if iname not in hyper.ListInstances(instance.hvparams): # if the instance is no longer existing, consider this a # success and go to cleanup return @@ -1347,14 +1659,14 @@ def InstanceShutdown(instance, timeout): try: hyper.StopInstance(instance, force=True) except errors.HypervisorError, err: - if iname in hyper.ListInstances(): + if iname in hyper.ListInstances(instance.hvparams): # only raise an error if the instance still exists, otherwise # the error could simply be "instance ... unknown"! _Fail("Failed to force stop instance %s: %s", iname, err) time.sleep(1) - if iname in hyper.ListInstances(): + if iname in hyper.ListInstances(instance.hvparams): _Fail("Could not shutdown instance %s even by destroy", iname) try: @@ -1365,7 +1677,7 @@ def InstanceShutdown(instance, timeout): _RemoveBlockDevLinks(iname, instance.disks) -def InstanceReboot(instance, reboot_type, shutdown_timeout): +def InstanceReboot(instance, reboot_type, shutdown_timeout, reason): """Reboot an instance. @type instance: L{objects.Instance} @@ -1383,10 +1695,13 @@ def InstanceReboot(instance, reboot_type, shutdown_timeout): instance (instead of a call_instance_reboot RPC) @type shutdown_timeout: integer @param shutdown_timeout: maximum timeout for soft shutdown + @type reason: list of reasons + @param reason: the reason trail for this reboot @rtype: None """ - running_instances = GetInstanceList([instance.hypervisor]) + running_instances = GetInstanceListForHypervisor(instance.hypervisor, + instance.hvparams) if instance.name not in running_instances: _Fail("Cannot reboot instance %s that is not running", instance.name) @@ -1399,8 +1714,10 @@ def InstanceReboot(instance, reboot_type, shutdown_timeout): _Fail("Failed to soft reboot instance %s: %s", instance.name, err) elif reboot_type == constants.INSTANCE_REBOOT_HARD: try: - InstanceShutdown(instance, shutdown_timeout) - return StartInstance(instance, False) + InstanceShutdown(instance, shutdown_timeout, reason, store_reason=False) + result = StartInstance(instance, False, reason, store_reason=False) + _StoreInstReasonTrail(instance.name, reason) + return result except errors.HypervisorError, err: _Fail("Failed to hard reboot instance %s: %s", instance.name, err) else: @@ -1418,7 +1735,7 @@ def InstanceBalloonMemory(instance, memory): """ hyper = hypervisor.GetHypervisor(instance.hypervisor) - running = hyper.ListInstances() + running = hyper.ListInstances(instance.hvparams) if instance.name not in running: logging.info("Instance %s is not running, cannot balloon", instance.name) return @@ -1490,9 +1807,11 @@ def FinalizeMigrationDst(instance, info, success): _Fail("Failed to finalize migration on the target node: %s", err, exc=True) -def MigrateInstance(instance, target, live): +def MigrateInstance(cluster_name, instance, target, live): """Migrates an instance to another node. + @type cluster_name: string + @param cluster_name: name of the cluster @type instance: L{objects.Instance} @param instance: the instance definition @type target: string @@ -1506,7 +1825,7 @@ def MigrateInstance(instance, target, live): hyper = hypervisor.GetHypervisor(instance.hypervisor) try: - hyper.MigrateInstance(instance, target, live) + hyper.MigrateInstance(cluster_name, instance, target, live) except errors.HypervisorError, err: _Fail("Failed to migrate instance: %s", err, exc=True) @@ -1809,7 +2128,7 @@ def BlockdevAssemble(disk, owner, as_primary, idx): """ try: result = _RecursiveAssembleBD(disk, owner, as_primary) - if isinstance(result, bdev.BlockDev): + if isinstance(result, BlockDev): # pylint: disable=E1103 result = result.dev_path if as_primary: @@ -2020,7 +2339,7 @@ def BlockdevFind(disk): return rbd.GetSyncStatus() -def BlockdevGetsize(disks): +def BlockdevGetdimensions(disks): """Computes the size of the given disks. If a disk is not found, returns None instead. @@ -2029,7 +2348,8 @@ def BlockdevGetsize(disks): @param disks: the list of disk to compute the size for @rtype: list @return: list with elements None if the disk cannot be found, - otherwise the size + otherwise the pair (size, spindles), where spindles is None if the + device doesn't support that """ result = [] @@ -2042,7 +2362,7 @@ def BlockdevGetsize(disks): if rbd is None: result.append(None) else: - result.append(rbd.GetActualSize()) + result.append(rbd.GetActualDimensions()) return result @@ -2448,6 +2768,9 @@ def OSEnvironment(instance, inst_os, debug=0): real_disk = _OpenRealBD(disk) result["DISK_%d_PATH" % idx] = real_disk.dev_path result["DISK_%d_ACCESS" % idx] = disk.mode + result["DISK_%d_UUID" % idx] = disk.uuid + if disk.name: + result["DISK_%d_NAME" % idx] = disk.name if constants.HV_DISK_TYPE in instance.hvparams: result["DISK_%d_FRONTEND_TYPE" % idx] = \ instance.hvparams[constants.HV_DISK_TYPE] @@ -2460,6 +2783,9 @@ def OSEnvironment(instance, inst_os, debug=0): # NICs for idx, nic in enumerate(instance.nics): result["NIC_%d_MAC" % idx] = nic.mac + result["NIC_%d_UUID" % idx] = nic.uuid + if nic.name: + result["NIC_%d_NAME" % idx] = nic.name if nic.ip: result["NIC_%d_IP" % idx] = nic.ip result["NIC_%d_MODE" % idx] = nic.nicparams[constants.NIC_MODE] @@ -2467,8 +2793,9 @@ def OSEnvironment(instance, inst_os, debug=0): result["NIC_%d_BRIDGE" % idx] = nic.nicparams[constants.NIC_LINK] if nic.nicparams[constants.NIC_LINK]: result["NIC_%d_LINK" % idx] = nic.nicparams[constants.NIC_LINK] - if nic.network: - result["NIC_%d_NETWORK" % idx] = nic.network + if nic.netinfo: + nobj = objects.Network.FromDict(nic.netinfo) + result.update(nobj.HooksDict("NIC_%d_" % idx)) if constants.HV_NIC_TYPE in instance.hvparams: result["NIC_%d_FRONTEND_TYPE" % idx] = \ instance.hvparams[constants.HV_NIC_TYPE] @@ -2930,7 +3257,7 @@ def JobQueueUpdate(file_name, content): # Write and replace the file atomically utils.WriteFile(file_name, data=_Decompress(content), uid=getents.masterd_uid, - gid=getents.masterd_gid) + gid=getents.daemons_gid, mode=constants.JOB_QUEUE_FILES_PERMS) def JobQueueRename(old, new): @@ -2954,8 +3281,8 @@ def JobQueueRename(old, new): getents = runtime.GetEnts() - utils.RenameFile(old, new, mkdir=True, mkdir_mode=0700, - dir_uid=getents.masterd_uid, dir_gid=getents.masterd_gid) + utils.RenameFile(old, new, mkdir=True, mkdir_mode=0750, + dir_uid=getents.masterd_uid, dir_gid=getents.daemons_gid) def BlockdevClose(instance_name, disks): @@ -3474,14 +3801,13 @@ def CleanupImportExport(name): shutil.rmtree(status_dir, ignore_errors=True) -def _FindDisks(nodes_ip, disks): +def _FindDisks(target_node_uuid, nodes_ip, disks): """Sets the physical ID on disks and returns the block devices. """ # set the correct physical ID - my_name = netutils.Hostname.GetSysName() for cf in disks: - cf.SetPhysicalID(my_name, nodes_ip) + cf.SetPhysicalID(target_node_uuid, nodes_ip) bdevs = [] @@ -3493,11 +3819,11 @@ def _FindDisks(nodes_ip, disks): return bdevs -def DrbdDisconnectNet(nodes_ip, disks): +def DrbdDisconnectNet(target_node_uuid, nodes_ip, disks): """Disconnects the network on a list of drbd devices. """ - bdevs = _FindDisks(nodes_ip, disks) + bdevs = _FindDisks(target_node_uuid, nodes_ip, disks) # disconnect disks for rd in bdevs: @@ -3508,11 +3834,12 @@ def DrbdDisconnectNet(nodes_ip, disks): err, exc=True) -def DrbdAttachNet(nodes_ip, disks, instance_name, multimaster): +def DrbdAttachNet(target_node_uuid, nodes_ip, disks, instance_name, + multimaster): """Attaches the network on a list of drbd devices. """ - bdevs = _FindDisks(nodes_ip, disks) + bdevs = _FindDisks(target_node_uuid, nodes_ip, disks) if multimaster: for idx, rd in enumerate(bdevs): @@ -3570,7 +3897,7 @@ def DrbdAttachNet(nodes_ip, disks, instance_name, multimaster): _Fail("Can't change to primary mode: %s", err) -def DrbdWaitSync(nodes_ip, disks): +def DrbdWaitSync(target_node_uuid, nodes_ip, disks): """Wait until DRBDs have synchronized. """ @@ -3580,7 +3907,7 @@ def DrbdWaitSync(nodes_ip, disks): raise utils.RetryAgain() return stats - bdevs = _FindDisks(nodes_ip, disks) + bdevs = _FindDisks(target_node_uuid, nodes_ip, disks) min_resync = 100 alldone = True @@ -3605,12 +3932,12 @@ def GetDrbdUsermodeHelper(): """ try: - return bdev.BaseDRBD.GetUsermodeHelper() + return drbd.DRBD8.GetUsermodeHelper() except errors.BlockDeviceError, err: _Fail(str(err)) -def PowercycleNode(hypervisor_type): +def PowercycleNode(hypervisor_type, hvparams=None): """Hard-powercycle the node. Because we need to return first, and schedule the powercycle in the @@ -3631,11 +3958,11 @@ def PowercycleNode(hypervisor_type): except Exception: # pylint: disable=W0703 pass time.sleep(5) - hyper.PowercycleNode() + hyper.PowercycleNode(hvparams=hvparams) def _VerifyRestrictedCmdName(cmd): - """Verifies a remote command name. + """Verifies a restricted command name. @type cmd: string @param cmd: Command name @@ -3657,7 +3984,7 @@ def _VerifyRestrictedCmdName(cmd): def _CommonRestrictedCmdCheck(path, owner): - """Common checks for remote command file system directories and files. + """Common checks for restricted command file system directories and files. @type path: string @param path: Path to check @@ -3687,7 +4014,7 @@ def _CommonRestrictedCmdCheck(path, owner): def _VerifyRestrictedCmdDirectory(path, _owner=None): - """Verifies remote command directory. + """Verifies restricted command directory. @type path: string @param path: Path to check @@ -3708,10 +4035,10 @@ def _VerifyRestrictedCmdDirectory(path, _owner=None): def _VerifyRestrictedCmd(path, cmd, _owner=None): - """Verifies a whole remote command and returns its executable filename. + """Verifies a whole restricted command and returns its executable filename. @type path: string - @param path: Directory containing remote commands + @param path: Directory containing restricted commands @type cmd: string @param cmd: Command name @rtype: tuple; (boolean, string) @@ -3737,10 +4064,10 @@ def _PrepareRestrictedCmd(path, cmd, _verify_dir=_VerifyRestrictedCmdDirectory, _verify_name=_VerifyRestrictedCmdName, _verify_cmd=_VerifyRestrictedCmd): - """Performs a number of tests on a remote command. + """Performs a number of tests on a restricted command. @type path: string - @param path: Directory containing remote commands + @param path: Directory containing restricted commands @type cmd: string @param cmd: Command name @return: Same as L{_VerifyRestrictedCmd} @@ -3767,7 +4094,7 @@ def RunRestrictedCmd(cmd, _prepare_fn=_PrepareRestrictedCmd, _runcmd_fn=utils.RunCmd, _enabled=constants.ENABLE_RESTRICTED_COMMANDS): - """Executes a remote command after performing strict tests. + """Executes a restricted command after performing strict tests. @type cmd: string @param cmd: Command name @@ -3776,10 +4103,10 @@ def RunRestrictedCmd(cmd, @raise RPCFail: In case of an error """ - logging.info("Preparing to run remote command '%s'", cmd) + logging.info("Preparing to run restricted command '%s'", cmd) if not _enabled: - _Fail("Remote commands disabled at configure time") + _Fail("Restricted commands disabled at configure time") lock = None try: @@ -3807,7 +4134,7 @@ def RunRestrictedCmd(cmd, # Do not include original error message in returned error _Fail("Executing command '%s' failed" % cmd) elif cmdresult.failed or cmdresult.fail_reason: - _Fail("Remote command '%s' failed: %s; output: %s", + _Fail("Restricted command '%s' failed: %s; output: %s", cmd, cmdresult.fail_reason, cmdresult.output) else: return cmdresult.output