#: Attribute holding field definitions
FIELDS = None
- def __init__(self, names, fields, use_locking):
+ def __init__(self, filter_, fields, use_locking):
"""Initializes this class.
"""
- self.names = names
self.use_locking = use_locking
- self.query = query.Query(self.FIELDS, fields)
+ self.query = query.Query(self.FIELDS, fields, filter_=filter_,
+ namefield="name")
self.requested_data = self.query.RequestedData()
+ self.names = self.query.RequestedNames()
+
+ # Sort only if no names were requested
+ self.sort_by_name = not self.names
self.do_locking = None
self.wanted = None
"""Collect data and execute query.
"""
- return query.GetQueryResponse(self.query, self._GetQueryData(lu))
+ return query.GetQueryResponse(self.query, self._GetQueryData(lu),
+ sort_by_name=self.sort_by_name)
def OldStyleQuery(self, lu):
"""Collect data and execute query.
"""
- return self.query.OldStyleQuery(self._GetQueryData(lu))
+ return self.query.OldStyleQuery(self._GetQueryData(lu),
+ sort_by_name=self.sort_by_name)
def _GetWantedNodes(lu, nodes):
"""Build a list of nic information tuples.
This list is suitable to be passed to _BuildInstanceHookEnv or as a return
- value in LUQueryInstanceData.
+ value in LUInstanceQueryData.
@type lu: L{LogicalUnit}
@param lu: the logical unit on whose behalf we execute
# Special case for file storage
if storage_type == constants.ST_FILE:
# storage.FileStorage wants a list of storage directories
- return [[cfg.GetFileStorageDir()]]
+ return [[cfg.GetFileStorageDir(), cfg.GetSharedFileStorageDir()]]
return []
@ivar instances: a list of running instances (runtime)
@ivar pinst: list of configured primary instances (config)
@ivar sinst: list of configured secondary instances (config)
- @ivar sbp: diction of {secondary-node: list of instances} of all peers
- of this node (config)
+ @ivar sbp: dictionary of {primary-node: list of instances} for all
+ instances for which this node is secondary (config)
@ivar mfree: free memory, as reported by hypervisor (runtime)
@ivar dfree: free disk, as reported by the node (runtime)
@ivar offline: the offline status (config)
_ErrorIf(test, self.ENODEHV, node,
"hypervisor %s verify failure: '%s'", hv_name, hv_result)
+ hvp_result = nresult.get(constants.NV_HVPARAMS, None)
+ if ninfo.vm_capable and isinstance(hvp_result, list):
+ for item, hv_name, hv_result in hvp_result:
+ _ErrorIf(True, self.ENODEHV, node,
+ "hypervisor %s parameter verify failure (source %s): %s",
+ hv_name, item, hv_result)
+
test = nresult.get(constants.NV_NODESETUP,
["Missing NODESETUP results"])
_ErrorIf(test, self.ENODESETUP, node, "node setup error: %s",
node_current)
for node, n_img in node_image.items():
- if (not node == node_current):
+ if node != node_current:
test = instance in n_img.instances
_ErrorIf(test, self.EINSTANCEWRONGNODE, instance,
"instance should not run on node %s", node)
for idx, (success, status) in enumerate(disks)]
for nname, success, bdev_status, idx in diskdata:
- _ErrorIf(instanceconfig.admin_up and not success,
+ # the 'ghost node' construction in Exec() ensures that we have a
+ # node here
+ snode = node_image[nname]
+ bad_snode = snode.ghost or snode.offline
+ _ErrorIf(instanceconfig.admin_up and not success and not bad_snode,
self.EINSTANCEFAULTYDISK, instance,
"couldn't retrieve status for disk/%s on %s: %s",
idx, nname, bdev_status)
instances it was primary for.
"""
+ cluster_info = self.cfg.GetClusterInfo()
for node, n_img in node_image.items():
# This code checks that every node which is now listed as
# secondary has enough memory to host all instances it is
# WARNING: we currently take into account down instances as well
# as up ones, considering that even if they're down someone
# might want to start them even in the event of a node failure.
+ if n_img.offline:
+ # we're skipping offline nodes from the N+1 warning, since
+ # most likely we don't have good memory infromation from them;
+ # we already list instances living on such nodes, and that's
+ # enough warning
+ continue
for prinode, instances in n_img.sbp.items():
needed_mem = 0
for instance in instances:
- bep = self.cfg.GetClusterInfo().FillBE(instance_cfg[instance])
+ bep = cluster_info.FillBE(instance_cfg[instance])
if bep[constants.BE_AUTO_BALANCE]:
needed_mem += bep[constants.BE_MEMORY]
test = n_img.mfree < needed_mem
return instdisk
+ def _VerifyHVP(self, hvp_data):
+ """Verifies locally the syntax of the hypervisor parameters.
+
+ """
+ for item, hv_name, hv_params in hvp_data:
+ msg = ("hypervisor %s parameters syntax check (source %s): %%s" %
+ (item, hv_name))
+ try:
+ hv_class = hypervisor.GetHypervisor(hv_name)
+ utils.ForceDictType(hv_params, constants.HVS_PARAMETER_TYPES)
+ hv_class.CheckParameterSyntax(hv_params)
+ except errors.GenericError, err:
+ self._ErrorIf(True, self.ECLUSTERCFG, None, msg % str(err))
+
+
def BuildHooksEnv(self):
"""Build hooks env.
local_checksums = utils.FingerprintFiles(file_names)
+ # Compute the set of hypervisor parameters
+ hvp_data = []
+ for hv_name in hypervisors:
+ hvp_data.append(("cluster", hv_name, cluster.GetHVDefaults(hv_name)))
+ for os_name, os_hvp in cluster.os_hvp.items():
+ for hv_name, hv_params in os_hvp.items():
+ if not hv_params:
+ continue
+ full_params = cluster.GetHVDefaults(hv_name, os_name=os_name)
+ hvp_data.append(("os %s" % os_name, hv_name, full_params))
+ # TODO: collapse identical parameter values in a single one
+ for instance in instanceinfo.values():
+ if not instance.hvparams:
+ continue
+ hvp_data.append(("instance %s" % instance.name, instance.hypervisor,
+ cluster.FillHV(instance)))
+ # and verify them locally
+ self._VerifyHVP(hvp_data)
+
feedback_fn("* Gathering data (%d nodes)" % len(nodelist))
node_verify_param = {
constants.NV_FILELIST: file_names,
constants.NV_NODELIST: [node.name for node in nodeinfo
if not node.offline],
constants.NV_HYPERVISOR: hypervisors,
+ constants.NV_HVPARAMS: hvp_data,
constants.NV_NODENETTEST: [(node.name, node.primary_ip,
node.secondary_ip) for node in nodeinfo
if not node.offline],
self.ENODERPC, pnode, "instance %s, connection to"
" primary node failed", instance)
- if pnode_img.offline:
- inst_nodes_offline.append(pnode)
+ _ErrorIf(pnode_img.offline, self.EINSTANCEBADNODE, instance,
+ "instance lives on offline node %s", inst_config.primary_node)
# If the instance is non-redundant we cannot survive losing its primary
# node, so we are not N+1 compliant. On the other hand we have no disk
# warn that the instance lives on offline nodes
_ErrorIf(inst_nodes_offline, self.EINSTANCEBADNODE, instance,
- "instance lives on offline node(s) %s",
+ "instance has offline secondary node(s) %s",
utils.CommaJoin(inst_nodes_offline))
# ... or ghost/non-vm_capable nodes
for node in inst_config.all_nodes:
"""
result = res_nodes, res_instances, res_missing = {}, [], {}
- nodes = utils.NiceSort(self.cfg.GetNodeList())
- instances = [self.cfg.GetInstanceInfo(name)
- for name in self.cfg.GetInstanceList()]
+ nodes = utils.NiceSort(self.cfg.GetVmCapableNodeList())
+ instances = self.cfg.GetAllInstancesInfo().values()
nv_dict = {}
for inst in instances:
inst_lvs = {}
- if (not inst.admin_up or
- inst.disk_template not in constants.DTS_NET_MIRROR):
+ if not inst.admin_up:
continue
inst.MapLVsByNode(inst_lvs)
# transform { iname: {node: [vol,],},} to {(node, vol): iname}
if not nv_dict:
return result
- vg_names = self.rpc.call_vg_list(nodes)
- vg_names.Raise("Cannot get list of VGs")
-
- for node in nodes:
- # node_volume
- node_res = self.rpc.call_lv_list([node],
- vg_names[node].payload.keys())[node]
+ node_lvs = self.rpc.call_lv_list(nodes, [])
+ for node, node_res in node_lvs.items():
if node_res.offline:
continue
msg = node_res.fail_msg
newl = [v[2].Copy() for v in dskl]
for dsk in newl:
self.cfg.SetDiskID(dsk, node)
- result = self.rpc.call_blockdev_getsizes(node, newl)
+ result = self.rpc.call_blockdev_getsize(node, newl)
if result.fail_msg:
- self.LogWarning("Failure in blockdev_getsizes call to node"
+ self.LogWarning("Failure in blockdev_getsize call to node"
" %s, ignoring", node)
continue
- if len(result.data) != len(dskl):
+ if len(result.payload) != len(dskl):
+ logging.warning("Invalid result from node %s: len(dksl)=%d,"
+ " result.payload=%s", node, len(dskl), result.payload)
self.LogWarning("Invalid result from node %s, ignoring node results",
node)
continue
- for ((instance, idx, disk), size) in zip(dskl, result.data):
+ for ((instance, idx, disk), size) in zip(dskl, result.payload):
if size is None:
self.LogWarning("Disk %d of instance %s did not return size"
" information, ignoring", idx, instance.name)
"""
REG_BGL = False
+ _SKIP_MASTER = (constants.OOB_POWER_OFF, constants.OOB_POWER_CYCLE)
def CheckPrereq(self):
"""Check prerequisites.
Any errors are signaled by raising errors.OpPrereqError.
"""
- self.op.node_name = _ExpandNodeName(self.cfg, self.op.node_name)
- node = self.cfg.GetNodeInfo(self.op.node_name)
+ self.nodes = []
+ self.master_node = self.cfg.GetMasterNode()
- if node is None:
- raise errors.OpPrereqError("Node %s not found" % self.op.node_name)
+ if self.op.node_names:
+ if self.op.command in self._SKIP_MASTER:
+ if self.master_node in self.op.node_names:
+ master_node_obj = self.cfg.GetNodeInfo(self.master_node)
+ master_oob_handler = _SupportsOob(self.cfg, master_node_obj)
- self.oob_program = _SupportsOob(self.cfg, node)
+ if master_oob_handler:
+ additional_text = ("Run '%s %s %s' if you want to operate on the"
+ " master regardless") % (master_oob_handler,
+ self.op.command,
+ self.master_node)
+ else:
+ additional_text = "The master node does not support out-of-band"
- if not self.oob_program:
- raise errors.OpPrereqError("OOB is not supported for node %s" %
- self.op.node_name)
+ raise errors.OpPrereqError(("Operating on the master node %s is not"
+ " allowed for %s\n%s") %
+ (self.master_node, self.op.command,
+ additional_text), errors.ECODE_INVAL)
+ else:
+ self.op.node_names = self.cfg.GetNodeList()
+ if self.op.command in self._SKIP_MASTER:
+ self.op.node_names.remove(self.master_node)
- if self.op.command == constants.OOB_POWER_OFF and not node.offline:
- raise errors.OpPrereqError(("Cannot power off node %s because it is"
- " not marked offline") % self.op.node_name)
+ if self.op.command in self._SKIP_MASTER:
+ assert self.master_node not in self.op.node_names
- self.node = node
+ for node_name in self.op.node_names:
+ node = self.cfg.GetNodeInfo(node_name)
+
+ if node is None:
+ raise errors.OpPrereqError("Node %s not found" % node_name,
+ errors.ECODE_NOENT)
+ else:
+ self.nodes.append(node)
+
+ if (not self.op.ignore_status and
+ (self.op.command == constants.OOB_POWER_OFF and not node.offline)):
+ raise errors.OpPrereqError(("Cannot power off node %s because it is"
+ " not marked offline") % node_name,
+ errors.ECODE_STATE)
def ExpandNames(self):
"""Gather locks we need.
"""
- node_name = _ExpandNodeName(self.cfg, self.op.node_name)
+ if self.op.node_names:
+ self.op.node_names = [_ExpandNodeName(self.cfg, name)
+ for name in self.op.node_names]
+ lock_names = self.op.node_names
+ else:
+ lock_names = locking.ALL_SET
+
self.needed_locks = {
- locking.LEVEL_NODE: [node_name],
+ locking.LEVEL_NODE: lock_names,
}
def Exec(self, feedback_fn):
"""Execute OOB and return result if we expect any.
"""
- master_node = self.cfg.GetMasterNode()
- node = self.node
-
- logging.info("Executing out-of-band command '%s' using '%s' on %s",
- self.op.command, self.oob_program, self.op.node_name)
- result = self.rpc.call_run_oob(master_node, self.oob_program,
- self.op.command, self.op.node_name,
- self.op.timeout)
+ master_node = self.master_node
+ ret = []
- result.Raise("An error occurred on execution of OOB helper")
+ for node in self.nodes:
+ node_entry = [(constants.RS_NORMAL, node.name)]
+ ret.append(node_entry)
- self._CheckPayload(result)
+ oob_program = _SupportsOob(self.cfg, node)
- if self.op.command == constants.OOB_HEALTH:
- # For health we should log important events
- for item, status in result.payload:
- if status in [constants.OOB_STATUS_WARNING,
- constants.OOB_STATUS_CRITICAL]:
- logging.warning("On node '%s' item '%s' has status '%s'",
- self.op.node_name, item, status)
-
- if self.op.command == constants.OOB_POWER_ON:
- node.powered = True
- elif self.op.command == constants.OOB_POWER_OFF:
- node.powered = False
- elif self.op.command == constants.OOB_POWER_STATUS:
- powered = result.payload[constants.OOB_POWER_STATUS_POWERED]
- if powered != self.node.powered:
- logging.warning(("Recorded power state (%s) of node '%s' does not match"
- " actual power state (%s)"), node.powered,
- self.op.node_name, powered)
+ if not oob_program:
+ node_entry.append((constants.RS_UNAVAIL, None))
+ continue
- self.cfg.Update(node, feedback_fn)
+ logging.info("Executing out-of-band command '%s' using '%s' on %s",
+ self.op.command, oob_program, node.name)
+ result = self.rpc.call_run_oob(master_node, oob_program,
+ self.op.command, node.name,
+ self.op.timeout)
- return result.payload
+ if result.fail_msg:
+ self.LogWarning("On node '%s' out-of-band RPC failed with: %s",
+ node.name, result.fail_msg)
+ node_entry.append((constants.RS_NODATA, None))
+ else:
+ try:
+ self._CheckPayload(result)
+ except errors.OpExecError, err:
+ self.LogWarning("The payload returned by '%s' is not valid: %s",
+ node.name, err)
+ node_entry.append((constants.RS_NODATA, None))
+ else:
+ if self.op.command == constants.OOB_HEALTH:
+ # For health we should log important events
+ for item, status in result.payload:
+ if status in [constants.OOB_STATUS_WARNING,
+ constants.OOB_STATUS_CRITICAL]:
+ self.LogWarning("On node '%s' item '%s' has status '%s'",
+ node.name, item, status)
+
+ if self.op.command == constants.OOB_POWER_ON:
+ node.powered = True
+ elif self.op.command == constants.OOB_POWER_OFF:
+ node.powered = False
+ elif self.op.command == constants.OOB_POWER_STATUS:
+ powered = result.payload[constants.OOB_POWER_STATUS_POWERED]
+ if powered != node.powered:
+ logging.warning(("Recorded power state (%s) of node '%s' does not"
+ " match actual power state (%s)"), node.powered,
+ node.name, powered)
+
+ # For configuration changing commands we should update the node
+ if self.op.command in (constants.OOB_POWER_ON,
+ constants.OOB_POWER_OFF):
+ self.cfg.Update(node, feedback_fn)
+
+ node_entry.append((constants.RS_NORMAL, result.payload))
+
+ return ret
def _CheckPayload(self, result):
"""Checks if the payload is valid.
if not isinstance(result.payload, list):
errs.append("command 'health' is expected to return a list but got %s" %
type(result.payload))
- for item, status in result.payload:
- if status not in constants.OOB_STATUSES:
- errs.append("health item '%s' has invalid status '%s'" %
- (item, status))
+ else:
+ for item, status in result.payload:
+ if status not in constants.OOB_STATUSES:
+ errs.append("health item '%s' has invalid status '%s'" %
+ (item, status))
if self.op.command == constants.OOB_POWER_STATUS:
if not isinstance(result.payload, dict):
-class LUDiagnoseOS(NoHooksLU):
+class LUOsDiagnose(NoHooksLU):
"""Logical unit for OS diagnose/query.
"""
"""Compute the list of OSes.
"""
- valid_nodes = [node for node in self.cfg.GetOnlineNodeList()]
+ valid_nodes = [node.name
+ for node in self.cfg.GetAllNodesInfo().values()
+ if not node.offline and node.vm_capable]
node_data = self.rpc.call_os_diagnose(valid_nodes)
pol = self._DiagnoseByOS(node_data)
output = []
return output
-class LURemoveNode(LogicalUnit):
+class LUNodeRemove(LogicalUnit):
"""Logical unit for removing a node.
"""
# Gather data as requested
if query.NQ_LIVE in self.requested_data:
- node_data = lu.rpc.call_node_info(nodenames, lu.cfg.GetVGName(),
+ # filter out non-vm_capable nodes
+ toquery_nodes = [name for name in nodenames if all_info[name].vm_capable]
+
+ node_data = lu.rpc.call_node_info(toquery_nodes, lu.cfg.GetVGName(),
lu.cfg.GetHypervisorType())
live_data = dict((name, nresult.payload)
for (name, nresult) in node_data.items()
oob_support, lu.cfg.GetClusterInfo())
-class LUQueryNodes(NoHooksLU):
+class LUNodeQuery(NoHooksLU):
"""Logical unit for querying nodes.
"""
REQ_BGL = False
def CheckArguments(self):
- self.nq = _NodeQuery(self.op.names, self.op.output_fields,
- self.op.use_locking)
+ self.nq = _NodeQuery(qlang.MakeSimpleFilter("name", self.op.names),
+ self.op.output_fields, self.op.use_locking)
def ExpandNames(self):
self.nq.ExpandNames(self)
return self.nq.OldStyleQuery(self)
-class LUQueryNodeVolumes(NoHooksLU):
+class LUNodeQueryvols(NoHooksLU):
"""Logical unit for getting volumes on node(s).
"""
return output
-class LUQueryNodeStorage(NoHooksLU):
+class LUNodeQueryStorage(NoHooksLU):
"""Logical unit for getting information on storage units on node(s).
"""
"""Computes the list of instances and their attributes.
"""
+ cluster = lu.cfg.GetClusterInfo()
all_info = lu.cfg.GetAllInstancesInfo()
instance_names = self._GetNames(lu, all_info.keys(), locking.LEVEL_INSTANCE)
instance_list = [all_info[name] for name in instance_names]
- nodes = frozenset([inst.primary_node for inst in instance_list])
+ nodes = frozenset(itertools.chain(*(inst.all_nodes
+ for inst in instance_list)))
hv_list = list(set([inst.hypervisor for inst in instance_list]))
bad_nodes = []
offline_nodes = []
+ wrongnode_inst = set()
# Gather data as requested
- if query.IQ_LIVE in self.requested_data:
+ if self.requested_data & set([query.IQ_LIVE, query.IQ_CONSOLE]):
live_data = {}
node_data = lu.rpc.call_all_instances_info(nodes, hv_list)
for name in nodes:
if result.fail_msg:
bad_nodes.append(name)
elif result.payload:
- live_data.update(result.payload)
+ for inst in result.payload:
+ if all_info[inst].primary_node == name:
+ live_data.update(result.payload)
+ else:
+ wrongnode_inst.add(inst)
# else no instance is alive
else:
live_data = {}
else:
disk_usage = None
+ if query.IQ_CONSOLE in self.requested_data:
+ consinfo = {}
+ for inst in instance_list:
+ if inst.name in live_data:
+ # Instance is running
+ consinfo[inst.name] = _GetInstanceConsole(cluster, inst)
+ else:
+ consinfo[inst.name] = None
+ assert set(consinfo.keys()) == set(instance_names)
+ else:
+ consinfo = None
+
return query.InstanceQueryData(instance_list, lu.cfg.GetClusterInfo(),
disk_usage, offline_nodes, bad_nodes,
- live_data)
+ live_data, wrongnode_inst, consinfo)
class LUQuery(NoHooksLU):
def CheckArguments(self):
qcls = _GetQueryImplementation(self.op.what)
- names = qlang.ReadSimpleFilter("name", self.op.filter)
- self.impl = qcls(names, self.op.fields, False)
+ self.impl = qcls(self.op.filter, self.op.fields, False)
def ExpandNames(self):
self.impl.ExpandNames(self)
return self.qcls.FieldsQuery(self.op.fields)
-class LUModifyNodeStorage(NoHooksLU):
+class LUNodeModifyStorage(NoHooksLU):
"""Logical unit for modifying a storage volume on a node.
"""
(self.op.name, self.op.node_name))
-class LUAddNode(LogicalUnit):
+class LUNodeAdd(LogicalUnit):
"""Logical unit for adding node to the cluster.
"""
self.context.AddNode(new_node, self.proc.GetECId())
-class LUSetNodeParams(LogicalUnit):
+class LUNodeSetParams(LogicalUnit):
"""Modifies the parameters of a node.
@cvar _F2R: a dictionary from tuples of flags (mc, drained, offline)
errors.ECODE_STATE)
if node.master_candidate and self.might_demote and not self.lock_all:
- assert not self.op.auto_promote, "auto-promote set but lock_all not"
+ assert not self.op.auto_promote, "auto_promote set but lock_all not"
# check if after removing the current node, we're missing master
# candidates
(mc_remaining, mc_should, _) = \
self.cfg.GetMasterCandidateStats(exceptions=[node.name])
if mc_remaining < mc_should:
raise errors.OpPrereqError("Not enough master candidates, please"
- " pass auto_promote to allow promotion",
- errors.ECODE_STATE)
+ " pass auto promote option to allow"
+ " promotion", errors.ECODE_STATE)
self.old_flags = old_flags = (node.master_candidate,
node.drained, node.offline)
return result
-class LUPowercycleNode(NoHooksLU):
+class LUNodePowercycle(NoHooksLU):
"""Powercycles a node.
"""
"volume_group_name": cluster.volume_group_name,
"drbd_usermode_helper": cluster.drbd_usermode_helper,
"file_storage_dir": cluster.file_storage_dir,
+ "shared_file_storage_dir": cluster.shared_file_storage_dir,
"maintain_node_health": cluster.maintain_node_health,
"ctime": cluster.ctime,
"mtime": cluster.mtime,
"reserved_lvs": cluster.reserved_lvs,
"primary_ip_version": primary_ip_version,
"prealloc_wipe_disks": cluster.prealloc_wipe_disks,
+ "hidden_os": cluster.hidden_os,
+ "blacklisted_os": cluster.blacklisted_os,
}
return result
return values
-class LUActivateInstanceDisks(NoHooksLU):
+class LUInstanceActivateDisks(NoHooksLU):
"""Bring up an instance's disks.
"""
# SyncSource, etc.)
# 1st pass, assemble on all nodes in secondary mode
- for inst_disk in disks:
+ for idx, inst_disk in enumerate(disks):
for node, node_disk in inst_disk.ComputeNodeTree(instance.primary_node):
if ignore_size:
node_disk = node_disk.Copy()
node_disk.UnsetSize()
lu.cfg.SetDiskID(node_disk, node)
- result = lu.rpc.call_blockdev_assemble(node, node_disk, iname, False)
+ result = lu.rpc.call_blockdev_assemble(node, node_disk, iname, False, idx)
msg = result.fail_msg
if msg:
lu.proc.LogWarning("Could not prepare block device %s on node %s"
# FIXME: race condition on drbd migration to primary
# 2nd pass, do only the primary node
- for inst_disk in disks:
+ for idx, inst_disk in enumerate(disks):
dev_path = None
for node, node_disk in inst_disk.ComputeNodeTree(instance.primary_node):
node_disk = node_disk.Copy()
node_disk.UnsetSize()
lu.cfg.SetDiskID(node_disk, node)
- result = lu.rpc.call_blockdev_assemble(node, node_disk, iname, True)
+ result = lu.rpc.call_blockdev_assemble(node, node_disk, iname, True, idx)
msg = result.fail_msg
if msg:
lu.proc.LogWarning("Could not prepare block device %s on node %s"
raise errors.OpExecError("Disk consistency error")
-class LUDeactivateInstanceDisks(NoHooksLU):
+class LUInstanceDeactivateDisks(NoHooksLU):
"""Shutdown an instance's disks.
"""
"""
instance = self.instance
- _SafeShutdownInstanceDisks(self, instance)
+ if self.op.force:
+ _ShutdownInstanceDisks(self, instance)
+ else:
+ _SafeShutdownInstanceDisks(self, instance)
def _SafeShutdownInstanceDisks(lu, instance, disks=None):
errors.ECODE_NORES)
-class LUStartupInstance(LogicalUnit):
+class LUInstanceStartup(LogicalUnit):
"""Starts an instance.
"""
raise errors.OpExecError("Could not start instance: %s" % msg)
-class LURebootInstance(LogicalUnit):
+class LUInstanceReboot(LogicalUnit):
"""Reboot an instance.
"""
ignore_secondaries = self.op.ignore_secondaries
reboot_type = self.op.reboot_type
+ remote_info = self.rpc.call_instance_info(instance.primary_node,
+ instance.name,
+ instance.hypervisor)
+ remote_info.Raise("Error checking node %s" % instance.primary_node)
+ instance_running = bool(remote_info.payload)
+
node_current = instance.primary_node
- if reboot_type in [constants.INSTANCE_REBOOT_SOFT,
- constants.INSTANCE_REBOOT_HARD]:
+ if instance_running and reboot_type in [constants.INSTANCE_REBOOT_SOFT,
+ constants.INSTANCE_REBOOT_HARD]:
for disk in instance.disks:
self.cfg.SetDiskID(disk, node_current)
result = self.rpc.call_instance_reboot(node_current, instance,
self.op.shutdown_timeout)
result.Raise("Could not reboot instance")
else:
- result = self.rpc.call_instance_shutdown(node_current, instance,
- self.op.shutdown_timeout)
- result.Raise("Could not shutdown instance for full reboot")
- _ShutdownInstanceDisks(self, instance)
+ if instance_running:
+ result = self.rpc.call_instance_shutdown(node_current, instance,
+ self.op.shutdown_timeout)
+ result.Raise("Could not shutdown instance for full reboot")
+ _ShutdownInstanceDisks(self, instance)
+ else:
+ self.LogInfo("Instance %s was already stopped, starting now",
+ instance.name)
_StartInstanceDisks(self, instance, ignore_secondaries)
result = self.rpc.call_instance_start(node_current, instance, None, None)
msg = result.fail_msg
self.cfg.MarkInstanceUp(instance.name)
-class LUShutdownInstance(LogicalUnit):
+class LUInstanceShutdown(LogicalUnit):
"""Shutdown an instance.
"""
_ShutdownInstanceDisks(self, instance)
-class LUReinstallInstance(LogicalUnit):
+class LUInstanceReinstall(LogicalUnit):
"""Reinstall an instance.
"""
_ShutdownInstanceDisks(self, inst)
-class LURecreateInstanceDisks(LogicalUnit):
+class LUInstanceRecreateDisks(LogicalUnit):
"""Recreate an instance's missing disks.
"""
_CreateDisks(self, self.instance, to_skip=to_skip)
-class LURenameInstance(LogicalUnit):
+class LUInstanceRename(LogicalUnit):
"""Rename an instance.
"""
hostname = netutils.GetHostname(name=new_name)
self.LogInfo("Resolved given name '%s' to '%s'", new_name,
hostname.name)
+ if not utils.MatchNameComponent(self.op.new_name, [hostname.name]):
+ raise errors.OpPrereqError(("Resolved hostname '%s' does not look the"
+ " same as given hostname '%s'") %
+ (hostname.name, self.op.new_name),
+ errors.ECODE_INVAL)
new_name = self.op.new_name = hostname.name
if (self.op.ip_check and
netutils.TcpPing(hostname.ip, constants.DEFAULT_NODED_PORT)):
old_name = inst.name
rename_file_storage = False
- if (inst.disk_template == constants.DT_FILE and
+ if (inst.disk_template in (constants.DT_FILE, constants.DT_SHARED_FILE) and
self.op.new_name != inst.name):
old_file_storage_dir = os.path.dirname(inst.disks[0].logical_id[1])
rename_file_storage = True
return inst.name
-class LURemoveInstance(LogicalUnit):
+class LUInstanceRemove(LogicalUnit):
"""Remove an instance.
"""
lu.remove_locks[locking.LEVEL_INSTANCE] = instance.name
-class LUQueryInstances(NoHooksLU):
+class LUInstanceQuery(NoHooksLU):
"""Logical unit for querying instances.
"""
REQ_BGL = False
def CheckArguments(self):
- self.iq = _InstanceQuery(self.op.names, self.op.output_fields,
- self.op.use_locking)
+ self.iq = _InstanceQuery(qlang.MakeSimpleFilter("name", self.op.names),
+ self.op.output_fields, self.op.use_locking)
def ExpandNames(self):
self.iq.ExpandNames(self)
return self.iq.OldStyleQuery(self)
-class LUFailoverInstance(LogicalUnit):
+class LUInstanceFailover(LogicalUnit):
"""Failover an instance.
"""
(instance.name, target_node, msg))
-class LUMigrateInstance(LogicalUnit):
+class LUInstanceMigrate(LogicalUnit):
"""Migrate an instance.
This is migration without shutting down, compared to the failover,
return env, nl, nl_post
-class LUMoveInstance(LogicalUnit):
+class LUInstanceMove(LogicalUnit):
"""Move an instance by data-copying.
"""
for idx, disk in enumerate(instance.disks):
self.LogInfo("Copying data for disk %d", idx)
result = self.rpc.call_blockdev_assemble(target_node, disk,
- instance.name, True)
+ instance.name, True, idx)
if result.fail_msg:
self.LogWarning("Can't assemble newly created disk %d: %s",
idx, result.fail_msg)
(instance.name, target_node, msg))
-class LUMigrateNode(LogicalUnit):
+class LUNodeMigrate(LogicalUnit):
"""Migrate all instances from a node.
"""
disk_index)),
mode=disk["mode"])
disks.append(disk_dev)
+ elif template_name == constants.DT_SHARED_FILE:
+ if len(secondary_nodes) != 0:
+ raise errors.ProgrammerError("Wrong template configuration")
+
+ opcodes.RequireSharedFileStorage()
+
+ for idx, disk in enumerate(disk_info):
+ disk_index = idx + base_index
+ disk_dev = objects.Disk(dev_type=constants.LD_FILE, size=disk["size"],
+ iv_name="disk/%d" % disk_index,
+ logical_id=(file_driver,
+ "%s/disk%d" % (file_storage_dir,
+ disk_index)),
+ mode=disk["mode"])
+ disks.append(disk_dev)
+ elif template_name == constants.DT_BLOCK:
+ if len(secondary_nodes) != 0:
+ raise errors.ProgrammerError("Wrong template configuration")
+
+ for idx, disk in enumerate(disk_info):
+ disk_index = idx + base_index
+ disk_dev = objects.Disk(dev_type=constants.LD_BLOCKDEV, size=disk["size"],
+ logical_id=(constants.BLOCKDEV_DRIVER_MANUAL,
+ disk["adopt"]),
+ iv_name="disk/%d" % disk_index,
+ mode=disk["mode"])
+ disks.append(disk_dev)
+
else:
raise errors.ProgrammerError("Invalid disk template '%s'" % template_name)
return disks
"""
node = instance.primary_node
+
+ for device in instance.disks:
+ lu.cfg.SetDiskID(device, node)
+
logging.info("Pause sync of instance %s disks", instance.name)
result = lu.rpc.call_blockdev_pause_resume_sync(node, instance.disks, True)
try:
for idx, device in enumerate(instance.disks):
lu.LogInfo("* Wiping disk %d", idx)
- logging.info("Wiping disk %d for instance %s", idx, instance.name)
+ logging.info("Wiping disk %d for instance %s, node %s",
+ idx, instance.name, node)
# The wipe size is MIN_WIPE_CHUNK_PERCENT % of the instance disk but
# MAX_WIPE_CHUNK at max
pnode = target_node
all_nodes = [pnode]
- if instance.disk_template == constants.DT_FILE:
+ if instance.disk_template in (constants.DT_FILE, constants.DT_SHARED_FILE):
file_storage_dir = os.path.dirname(instance.disks[0].logical_id[1])
result = lu.rpc.call_file_storage_dir_create(pnode, file_storage_dir)
" node %s" % (file_storage_dir, pnode))
# Note: this needs to be kept in sync with adding of disks in
- # LUSetInstanceParams
+ # LUInstanceSetParams
for idx, device in enumerate(instance.disks):
if to_skip and idx in to_skip:
continue
# 128 MB are added for drbd metadata for each disk
constants.DT_DRBD8: _compute(disks, 128),
constants.DT_FILE: {},
+ constants.DT_SHARED_FILE: {},
}
if disk_template not in req_size_dict:
# 128 MB are added for drbd metadata for each disk
constants.DT_DRBD8: sum(d["size"] + 128 for d in disks),
constants.DT_FILE: None,
+ constants.DT_SHARED_FILE: 0,
+ constants.DT_BLOCK: 0,
}
if disk_template not in req_size_dict:
return req_size_dict[disk_template]
+def _FilterVmNodes(lu, nodenames):
+ """Filters out non-vm_capable nodes from a list.
+
+ @type lu: L{LogicalUnit}
+ @param lu: the logical unit for which we check
+ @type nodenames: list
+ @param nodenames: the list of nodes on which we should check
+ @rtype: list
+ @return: the list of vm-capable nodes
+
+ """
+ vm_nodes = frozenset(lu.cfg.GetNonVmCapableNodeList())
+ return [name for name in nodenames if name not in vm_nodes]
+
+
def _CheckHVParams(lu, nodenames, hvname, hvparams):
"""Hypervisor parameter validation.
@raise errors.OpPrereqError: if the parameters are not valid
"""
+ nodenames = _FilterVmNodes(lu, nodenames)
hvinfo = lu.rpc.call_hypervisor_validate_params(nodenames,
hvname,
hvparams)
@raise errors.OpPrereqError: if the parameters are not valid
"""
+ nodenames = _FilterVmNodes(lu, nodenames)
result = lu.rpc.call_os_validate(required, nodenames, osname,
[constants.OS_VALIDATE_PARAMETERS],
osparams)
osname, node)
-class LUCreateInstance(LogicalUnit):
+class LUInstanceCreate(LogicalUnit):
"""Create an instance.
"""
if self.op.mode == constants.INSTANCE_IMPORT:
raise errors.OpPrereqError("Disk adoption not allowed for"
" instance import", errors.ECODE_INVAL)
+ else:
+ if self.op.disk_template in constants.DTS_MUST_ADOPT:
+ raise errors.OpPrereqError("Disk template %s requires disk adoption,"
+ " but no 'adopt' parameter given" %
+ self.op.disk_template,
+ errors.ECODE_INVAL)
self.adopt_disks = has_adopt
" in cluster" % mac,
errors.ECODE_NOTUNIQUE)
- # bridge verification
- bridge = nic.get("bridge", None)
- link = nic.get("link", None)
- if bridge and link:
- raise errors.OpPrereqError("Cannot pass 'bridge' and 'link'"
- " at the same time", errors.ECODE_INVAL)
- elif bridge and nic_mode == constants.NIC_MODE_ROUTED:
- raise errors.OpPrereqError("Cannot pass 'bridge' on a routed nic",
- errors.ECODE_INVAL)
- elif bridge:
- link = bridge
-
+ # Build nic parameters
+ link = nic.get(constants.INIC_LINK, None)
nicparams = {}
if nic_mode_req:
nicparams[constants.NIC_MODE] = nic_mode_req
req_sizes = _ComputeDiskSizePerVG(self.op.disk_template, self.disks)
_CheckNodesFreeDiskPerVG(self, nodenames, req_sizes)
- else: # instead, we must check the adoption data
+ elif self.op.disk_template == constants.DT_PLAIN: # Check the adoption data
all_lvs = set([i["vg"] + "/" + i["adopt"] for i in self.disks])
if len(all_lvs) != len(self.disks):
raise errors.OpPrereqError("Duplicate volume names given for adoption",
raise errors.OpPrereqError("LV named %s used by another instance" %
lv_name, errors.ECODE_NOTUNIQUE)
- vg_names = self.rpc.call_vg_list([pnode.name])
+ vg_names = self.rpc.call_vg_list([pnode.name])[pnode.name]
vg_names.Raise("Cannot get VG information from node %s" % pnode.name)
node_lvs = self.rpc.call_lv_list([pnode.name],
- vg_names[pnode.name].payload.keys()
- )[pnode.name]
+ vg_names.payload.keys())[pnode.name]
node_lvs.Raise("Cannot get LV information from node %s" % pnode.name)
node_lvs = node_lvs.payload
for dsk in self.disks:
dsk["size"] = int(float(node_lvs[dsk["vg"] + "/" + dsk["adopt"]][0]))
+ elif self.op.disk_template == constants.DT_BLOCK:
+ # Normalize and de-duplicate device paths
+ all_disks = set([os.path.abspath(i["adopt"]) for i in self.disks])
+ if len(all_disks) != len(self.disks):
+ raise errors.OpPrereqError("Duplicate disk names given for adoption",
+ errors.ECODE_INVAL)
+ baddisks = [d for d in all_disks
+ if not d.startswith(constants.ADOPTABLE_BLOCKDEV_ROOT)]
+ if baddisks:
+ raise errors.OpPrereqError("Device node(s) %s lie outside %s and"
+ " cannot be adopted" %
+ (", ".join(baddisks),
+ constants.ADOPTABLE_BLOCKDEV_ROOT),
+ errors.ECODE_INVAL)
+
+ node_disks = self.rpc.call_bdev_sizes([pnode.name],
+ list(all_disks))[pnode.name]
+ node_disks.Raise("Cannot get block device information from node %s" %
+ pnode.name)
+ node_disks = node_disks.payload
+ delta = all_disks.difference(node_disks.keys())
+ if delta:
+ raise errors.OpPrereqError("Missing block device(s): %s" %
+ utils.CommaJoin(delta),
+ errors.ECODE_INVAL)
+ for dsk in self.disks:
+ dsk["size"] = int(float(node_disks[dsk["adopt"]]))
+
_CheckHVParams(self, nodenames, self.op.hypervisor, self.op.hvparams)
_CheckNodeHasOS(self, pnode.name, self.op.os_type, self.op.force_variant)
else:
network_port = None
- if constants.ENABLE_FILE_STORAGE:
+ if constants.ENABLE_FILE_STORAGE or constants.ENABLE_SHARED_FILE_STORAGE:
# this is needed because os.path.join does not accept None arguments
if self.op.file_storage_dir is None:
string_file_storage_dir = ""
string_file_storage_dir = self.op.file_storage_dir
# build the full file storage dir path
- file_storage_dir = utils.PathJoin(self.cfg.GetFileStorageDir(),
+ if self.op.disk_template == constants.DT_SHARED_FILE:
+ get_fsd_fn = self.cfg.GetSharedFileStorageDir
+ else:
+ get_fsd_fn = self.cfg.GetFileStorageDir
+
+ file_storage_dir = utils.PathJoin(get_fsd_fn(),
string_file_storage_dir, instance)
else:
file_storage_dir = ""
)
if self.adopt_disks:
- # rename LVs to the newly-generated names; we need to construct
- # 'fake' LV disks with the old data, plus the new unique_id
- tmp_disks = [objects.Disk.FromDict(v.ToDict()) for v in disks]
- rename_to = []
- for t_dsk, a_dsk in zip (tmp_disks, self.disks):
- rename_to.append(t_dsk.logical_id)
- t_dsk.logical_id = (t_dsk.logical_id[0], a_dsk["adopt"])
- self.cfg.SetDiskID(t_dsk, pnode_name)
- result = self.rpc.call_blockdev_rename(pnode_name,
- zip(tmp_disks, rename_to))
- result.Raise("Failed to rename adoped LVs")
+ if self.op.disk_template == constants.DT_PLAIN:
+ # rename LVs to the newly-generated names; we need to construct
+ # 'fake' LV disks with the old data, plus the new unique_id
+ tmp_disks = [objects.Disk.FromDict(v.ToDict()) for v in disks]
+ rename_to = []
+ for t_dsk, a_dsk in zip (tmp_disks, self.disks):
+ rename_to.append(t_dsk.logical_id)
+ t_dsk.logical_id = (t_dsk.logical_id[0], a_dsk["adopt"])
+ self.cfg.SetDiskID(t_dsk, pnode_name)
+ result = self.rpc.call_blockdev_rename(pnode_name,
+ zip(tmp_disks, rename_to))
+ result.Raise("Failed to rename adoped LVs")
else:
feedback_fn("* creating instance disks...")
try:
return list(iobj.all_nodes)
-class LUConnectConsole(NoHooksLU):
+class LUInstanceConsole(NoHooksLU):
"""Connect to an instance's console.
This is somewhat special in that it returns the command line that
if instance.name not in node_insts.payload:
if instance.admin_up:
- state = "ERROR_down"
+ state = constants.INSTST_ERRORDOWN
else:
- state = "ADMIN_down"
+ state = constants.INSTST_ADMINDOWN
raise errors.OpExecError("Instance %s is not running (state %s)" %
(instance.name, state))
logging.debug("Connecting to console of %s on %s", instance.name, node)
- hyper = hypervisor.GetHypervisor(instance.hypervisor)
- cluster = self.cfg.GetClusterInfo()
- # beparams and hvparams are passed separately, to avoid editing the
- # instance and then saving the defaults in the instance itself.
- hvparams = cluster.FillHV(instance)
- beparams = cluster.FillBE(instance)
- console = hyper.GetInstanceConsole(instance, hvparams, beparams)
+ return _GetInstanceConsole(self.cfg.GetClusterInfo(), instance)
+
+
+def _GetInstanceConsole(cluster, instance):
+ """Returns console information for an instance.
+
+ @type cluster: L{objects.Cluster}
+ @type instance: L{objects.Instance}
+ @rtype: dict
+
+ """
+ hyper = hypervisor.GetHypervisor(instance.hypervisor)
+ # beparams and hvparams are passed separately, to avoid editing the
+ # instance and then saving the defaults in the instance itself.
+ hvparams = cluster.FillHV(instance)
+ beparams = cluster.FillBE(instance)
+ console = hyper.GetInstanceConsole(instance, hvparams, beparams)
- assert console.instance == instance.name
- assert console.Validate()
+ assert console.instance == instance.name
+ assert console.Validate()
- return console.ToDict()
+ return console.ToDict()
-class LUReplaceDisks(LogicalUnit):
+class LUInstanceReplaceDisks(LogicalUnit):
"""Replace the disks of an instance.
"""
(self.op.name, self.op.node_name))
-class LUNodeEvacuationStrategy(NoHooksLU):
+class LUNodeEvacStrategy(NoHooksLU):
"""Computes the node evacuation strategy.
"""
return result
-class LUGrowDisk(LogicalUnit):
+class LUInstanceGrowDisk(LogicalUnit):
"""Grow a disk of an instance.
"""
self.disk = instance.FindDisk(self.op.disk)
- if instance.disk_template != constants.DT_FILE:
- # TODO: check the free disk space for file, when that feature
- # will be supported
+ if instance.disk_template not in (constants.DT_FILE,
+ constants.DT_SHARED_FILE):
+ # TODO: check the free disk space for file, when that feature will be
+ # supported
_CheckNodesFreeDiskPerVG(self, nodenames,
self.disk.ComputeGrowth(self.op.amount))
" sync mode was requested.")
-class LUQueryInstanceData(NoHooksLU):
+class LUInstanceQueryData(NoHooksLU):
"""Query runtime instance data.
"""
return result
-class LUSetInstanceParams(LogicalUnit):
+class LUInstanceSetParams(LogicalUnit):
"""Modifies an instances's parameters.
"""
_CheckInstanceDown(self, instance, "cannot remove disks")
if (disk_op == constants.DDM_ADD and
- len(instance.nics) >= constants.MAX_DISKS):
+ len(instance.disks) >= constants.MAX_DISKS):
raise errors.OpPrereqError("Instance has too many disks (%d), cannot"
" add more" % constants.MAX_DISKS,
errors.ECODE_STATE)
result.append(("disk/%d" % device_idx, "remove"))
elif disk_op == constants.DDM_ADD:
# add a new disk
- if instance.disk_template == constants.DT_FILE:
+ if instance.disk_template in (constants.DT_FILE,
+ constants.DT_SHARED_FILE):
file_driver, file_path = instance.disks[0].logical_id
file_path = os.path.dirname(file_path)
else:
del self.remove_locks[locking.LEVEL_NODEGROUP]
-class LUAssignGroupNodes(NoHooksLU):
+class LUGroupAssignNodes(NoHooksLU):
"""Logical unit for assigning nodes to groups.
"""
class _GroupQuery(_QueryBase):
-
FIELDS = query.GROUP_FIELDS
def ExpandNames(self, lu):
group_to_nodes, group_to_instances)
-class LUQueryGroups(NoHooksLU):
+class LUGroupQuery(NoHooksLU):
"""Logical unit for querying node groups.
"""
REQ_BGL = False
def CheckArguments(self):
- self.gq = _GroupQuery(self.op.names, self.op.output_fields, False)
+ self.gq = _GroupQuery(qlang.MakeSimpleFilter("name", self.op.names),
+ self.op.output_fields, False)
def ExpandNames(self):
self.gq.ExpandNames(self)
return self.gq.OldStyleQuery(self)
-class LUSetGroupParams(LogicalUnit):
+class LUGroupSetParams(LogicalUnit):
"""Modifies the parameters of a node group.
"""
-class LURemoveGroup(LogicalUnit):
+class LUGroupRemove(LogicalUnit):
HPATH = "group-remove"
HTYPE = constants.HTYPE_GROUP
REQ_BGL = False
# Verify the cluster would not be left group-less.
if len(self.cfg.GetNodeGroupList()) == 1:
- raise errors.OpPrereqError("Group '%s' is the last group in the cluster,"
- " which cannot be left without at least one"
- " group" % self.op.group_name,
+ raise errors.OpPrereqError("Group '%s' is the only group,"
+ " cannot be removed" %
+ self.op.group_name,
errors.ECODE_STATE)
def BuildHooksEnv(self):
self.remove_locks[locking.LEVEL_NODEGROUP] = self.group_uuid
-class LURenameGroup(LogicalUnit):
+class LUGroupRename(LogicalUnit):
HPATH = "group-rename"
HTYPE = constants.HTYPE_GROUP
REQ_BGL = False
def ExpandNames(self):
# This raises errors.OpPrereqError on its own:
- self.group_uuid = self.cfg.LookupNodeGroup(self.op.old_name)
+ self.group_uuid = self.cfg.LookupNodeGroup(self.op.group_name)
self.needed_locks = {
locking.LEVEL_NODEGROUP: [self.group_uuid],
def CheckPrereq(self):
"""Check prerequisites.
- This checks that the given old_name exists as a node group, and that
- new_name doesn't.
+ Ensures requested new name is not yet used.
"""
try:
"""
env = {
- "OLD_NAME": self.op.old_name,
+ "OLD_NAME": self.op.group_name,
"NEW_NAME": self.op.new_name,
}
if group is None:
raise errors.OpExecError("Could not retrieve group '%s' (UUID: %s)" %
- (self.op.old_name, self.group_uuid))
+ (self.op.group_name, self.group_uuid))
group.name = self.op.new_name
self.cfg.Update(group, feedback_fn)
str(self.op.kind), errors.ECODE_INVAL)
-class LUGetTags(TagsLU):
+class LUTagsGet(TagsLU):
"""Returns the tags of a given object.
"""
return list(self.target.GetTags())
-class LUSearchTags(NoHooksLU):
+class LUTagsSearch(NoHooksLU):
"""Searches the tags for a given pattern.
"""
return results
-class LUAddTags(TagsLU):
+class LUTagsSet(TagsLU):
"""Sets a tag on a given object.
"""
self.cfg.Update(self.target, feedback_fn)
-class LUDelTags(TagsLU):
+class LUTagsDel(TagsLU):
"""Delete a list of tags from a given object.
"""
self._TestDelay()
-class LUTestJobqueue(NoHooksLU):
+class LUTestJqueue(NoHooksLU):
"""Utility LU to test some aspects of the job queue.
"""
"i_pri_up_memory": i_p_up_mem,
}
pnr_dyn.update(node_results[nname])
-
- node_results[nname] = pnr_dyn
+ node_results[nname] = pnr_dyn
return node_results
projects / ganeti-local / blobdiff