Package ganeti :: Module backend

[frames] | no frames]

Module backend

source code

Functions used by the node daemon

Classes

[hide private]

RPCFail
Class denoting RPC failure.

HooksRunner
Hook runner.

IAllocatorRunner
IAllocator runner.

DevCacheManager
Simple class for managing a cache of block device information.

Functions

[hide private]

_Fail(msg, *args, **kwargs)
Log an error and the raise an RPCFail exception.

source code

ssconf.SimpleStore

_GetConfig()
Simple wrapper to return a SimpleStore.

source code

ssh.SshRunner

_GetSshRunner(cluster_name)
Simple wrapper to return an SshRunner.

source code

str

_Decompress(data)
Unpacks data compressed by the RPC client.

source code

_CleanDirectory(path, exclude=None)
Removes all regular files in a directory.

source code

_BuildUploadFileList()
Build the list of allowed upload files.

source code

tuple

JobQueuePurge()
Removes job queue files and archived jobs.

source code

tuple

GetMasterInfo()
Returns master information.

source code

None

StartMaster(start_daemons, no_voting)
Activate local node as master node.

source code

None

StopMaster(stop_daemons)
Deactivate this node as master.

source code

EtcHostsModify(mode, host, ip)
Modify a host entry in /etc/hosts.

source code

LeaveCluster(modify_ssh_setup)
Cleans up and remove the current node.

source code

dict

GetNodeInfo(vgname, hypervisor_type)
Gives back a hash with different information about the node.

source code

dict

VerifyNode(what, cluster_name)
Verify the status of the local node.

source code

dict

GetBlockDevSizes(devices)
Return the size of the given block devices

source code

dict

GetVolumeList(vg_names)
Compute list of logical volumes and their size.

source code

dict

ListVolumeGroups()
List the volume groups and their size.

source code

list

NodeVolumes()
List all volumes on this node.

source code

boolean

BridgesExist(bridges_list)
Check if a list of bridges exist on the current node.

source code

list

GetInstanceList(hypervisor_list)
Provides a list of instances.

source code

dict

GetInstanceInfo(instance, hname)
Gives back the information about an instance as a dictionary.

source code

tuple

GetInstanceMigratable(instance)
Gives whether an instance can be migrated.

source code

dict

GetAllInstancesInfo(hypervisor_list)
Gather data about all instances.

source code

_InstanceLogName(kind, os_name, instance, component)
Compute the OS log filename for a given instance and operation.

source code

None

InstanceOsAdd(instance, reinstall, debug)
Add an OS to an instance.

source code

boolean

RunRenameInstance(instance, old_name, debug)
Run the OS rename script for an instance.

source code

_GetBlockDevSymlinkPath(instance_name, idx)

source code

_SymlinkBlockDev(instance_name, device_path, idx)
Set up symlinks to a instance's block device.

source code

_RemoveBlockDevLinks(instance_name, disks)
Remove the block device symlinks belonging to the given instance.

source code

list

_GatherAndLinkBlockDevs(instance)
Set up an instance's block device(s).

source code

None

StartInstance(instance, startup_paused)
Start an instance.

source code

None

InstanceShutdown(instance, timeout)
Shut an instance down.

source code

None

InstanceReboot(instance, reboot_type, shutdown_timeout)
Reboot an instance.

source code

MigrationInfo(instance)
Gather information about an instance to be migrated.

source code

AcceptInstance(instance, info, target)
Prepare the node to accept an instance.

source code

FinalizeMigration(instance, info, success)
Finalize any preparation to accept an instance.

source code

tuple

MigrateInstance(instance, target, live)
Migrates an instance to another node.

source code

BlockdevCreate(disk, size, owner, on_primary, info)
Creates a block device for an instance.

source code

_WipeDevice(path, offset, size)
This function actually wipes the device.

source code

BlockdevWipe(disk, offset, size)
Wipes a block device.

source code

BlockdevPauseResumeSync(disks, pause)
Pause or resume the sync of the block device.

source code

boolean

BlockdevRemove(disk)
Remove a block device.

source code

_RecursiveAssembleBD(disk, owner, as_primary)
Activate a block device for an instance.

source code

str or boolean

BlockdevAssemble(disk, owner, as_primary, idx)
Activate a block device for an instance.

source code

None

BlockdevShutdown(disk)
Shut down a block device.

source code

None

BlockdevAddchildren(parent_cdev, new_cdevs)
Extend a mirrored block device.

source code

None

BlockdevRemovechildren(parent_cdev, new_cdevs)
Shrink a mirrored block device.

source code

disk

BlockdevGetmirrorstatus(disks)
Get the mirroring status of a list of devices.

source code

disk

BlockdevGetmirrorstatusMulti(disks)
Get the mirroring status of a list of devices.

source code

_RecursiveFindBD(disk)
Check if a device is activated.

source code

_OpenRealBD(disk)
Opens the underlying block device of a disk.

source code

None or objects.BlockDevStatus

BlockdevFind(disk)
Check if a device is activated.

source code

list

BlockdevGetsize(disks)
Computes the size of the given disks.

source code

None

BlockdevExport(disk, dest_node, dest_path, cluster_name)
Export a block device to a remote node.

source code

None

UploadFile(file_name, data, mode, uid, gid, atime, mtime)
Write a file to the filesystem.

source code

RunOob(oob_program, command, node, timeout)
Executes oob_program with given command on given node.

source code

WriteSsconfFiles(values)
Update all ssconf files.

source code

_ErrnoOrStr(err)
Format an EnvironmentError exception.

source code

tuple

_OSOndiskAPIVersion(os_dir)
Compute and return the API version of a given OS.

source code

list of objects.OS

DiagnoseOS(top_dirs=None)
Compute the validity for all OSes.

source code

tuple

_TryOSFromDisk(name, base_dir=None)
Create an OS instance from disk.

source code

objects.OS

OSFromDisk(name, base_dir=None)
Create an OS instance from disk.

source code

dict

OSCoreEnv(os_name, inst_os, os_params, debug=0)
Calculate the basic environment for an os script.

source code

dict

OSEnvironment(instance, inst_os, debug=0)
Calculate the environment for an os script.

source code

(status, result)

BlockdevGrow(disk, amount, dryrun)
Grow a stack of block devices.

source code

string

BlockdevSnapshot(disk)
Create a snapshot copy of a block device.

source code

None

FinalizeExport(instance, snap_disks)
Write out the export configuration information.

source code

objects.SerializableConfigParser

ExportInfo(dest)
Get export configuration information.

source code

list

ListExports()
Return a list of exports currently available on this machine.

source code

None

RemoveExport(export)
Remove an existing export from the node.

source code

boolean

BlockdevRename(devlist)
Rename a list of block devices.

source code

_TransformFileStorageDir(fs_dir)
Checks whether given file_storage_dir is valid.

source code

tuple

CreateFileStorageDir(file_storage_dir)
Create file storage directory.

source code

tuple (success,)

RemoveFileStorageDir(file_storage_dir)
Remove file storage directory.

source code

tuple (success,)

RenameFileStorageDir(old_file_storage_dir, new_file_storage_dir)
Rename the file storage directory.

source code

None

_EnsureJobQueueFile(file_name)
Checks whether the given filename is in the queue directory.

source code

boolean

JobQueueUpdate(file_name, content)
Updates a file in the queue directory.

source code

tuple

JobQueueRename(old, new)
Renames a job queue file.

source code

tuple (success, message)

BlockdevClose(instance_name, disks)
Closes the given block devices.

source code

None

ValidateHVParams(hvname, hvparams)
Validates the given hypervisor parameters.

source code

_CheckOSPList(os_obj, parameters)
Check whether a list of parameters is supported by the OS.

source code

boolean

ValidateOS(required, osname, checks, osparams)
Validate the given OS' parameters.

source code

DemoteFromMC()
Demotes the current node from master candidate role.

source code

_GetX509Filenames(cryptodir, name)
Returns the full paths for the private key and certificate.

source code

tuple; (string, string)

CreateX509Certificate(validity, cryptodir=constants.CRYPTO_KEYS_DIR)
Creates a new X509 certificate for SSL/TLS.

source code

RemoveX509Certificate(name, cryptodir=constants.CRYPTO_KEYS_DIR)
Removes a X509 certificate.

source code

_GetImportExportIoCommand(instance, mode, ieio, ieargs)
Returns the command for the requested input/output.

source code

_CreateImportExportStatusDir(prefix)
Creates status directory for import/export.

source code

StartImportExportDaemon(mode, opts, host, port, instance, component, ieio, ieioargs)
Starts an import or export daemon.

source code

List of dicts

GetImportExportStatus(names)
Returns import/export daemon status.

source code

AbortImportExport(name)
Sends SIGTERM to a running import/export daemon.

source code

CleanupImportExport(name)
Cleanup after an import or export.

source code

_FindDisks(nodes_ip, disks)
Sets the physical ID on disks and returns the block devices.

source code

DrbdDisconnectNet(nodes_ip, disks)
Disconnects the network on a list of drbd devices.

source code

DrbdAttachNet(nodes_ip, disks, instance_name, multimaster)
Attaches the network on a list of drbd devices.

source code

DrbdWaitSync(nodes_ip, disks)
Wait until DRBDs have synchronized.

source code

GetDrbdUsermodeHelper()
Returns DRBD usermode helper currently configured.

source code

PowercycleNode(hypervisor_type)
Hard-powercycle the node.

source code

Variables

[hide private]

_BOOT_ID_PATH = "/proc/sys/kernel/random/boot_id"

_ALLOWED_CLEAN_DIRS = frozenset([constants.DATA_DIR, constants...
denotes which directories are accepted in the _CleanDirectory function

_MAX_SSL_CERT_VALIDITY = 7* 24* 60* 60

_X509_KEY_FILE = "key"

_X509_CERT_FILE = "cert"

_IES_STATUS_FILE = "status"

_IES_PID_FILE = "pid"

_IES_CA_FILE = "ca"

_LVSLINE_REGEX = re.compile("^ *([^|]+)\|([^|]+)\|([0-9.]+)\|(...
Valid LVS output line regex

_ALLOWED_UPLOAD_FILES = _BuildUploadFileList()
denotes which files are accepted in the UploadFile function

Imports: os, shutil, time, stat, errno, re, random, logging, tempfile, zlib, base64, signal, errors, utils, ssh, hypervisor, constants, bdev, objects, ssconf, serializer, netutils, runtime

Function Details

[hide private]

_Fail(msg, *args, **kwargs)

source code

Log an error and the raise an RPCFail exception.

This exception is then handled specially in the ganeti daemon and turned into a 'failed' return type. As such, this function is a useful shortcut for logging the error and returning it to the master daemon.

Parameters:

msg (string) - the text of the exception @raise RPCFail

_GetConfig()

source code

Simple wrapper to return a SimpleStore.

Returns: ssconf.SimpleStore: a SimpleStore instance

_GetSshRunner(cluster_name)

source code

Simple wrapper to return an SshRunner.

Parameters:

cluster_name (str) - the cluster name, which is needed by the SshRunner constructor

Returns: ssh.SshRunner

an SshRunner instance

_Decompress(data)

source code

Unpacks data compressed by the RPC client.

Parameters:

data (list or tuple) - Data sent by RPC client

Returns: str

Decompressed data

_CleanDirectory(path, exclude=None)

source code

Removes all regular files in a directory.

Parameters:

path (str) - the directory to clean
exclude (list) - list of files to be excluded, defaults to the empty list

_BuildUploadFileList()

source code

Build the list of allowed upload files.

This is abstracted so that it's built only once at module import time.

JobQueuePurge()

source code

Removes job queue files and archived jobs.

Returns: tuple: True, None

GetMasterInfo()

source code

Returns master information.

This is an utility function to compute master information, either for consumption here or from the node daemon.

Returns: tuple

master_netdev, master_ip, master_name, primary_ip_family

Raises:

RPCFail - in case of errors

StartMaster(start_daemons, no_voting)

source code

Activate local node as master node.

The function will either try activate the IP address of the master (unless someone else has it) or also start the master daemons, based on the start_daemons parameter.

Parameters:

start_daemons (boolean) - whether to start the master daemons (ganeti-masterd and ganeti-rapi), or (if false) activate the master ip
no_voting (boolean) - whether to start ganeti-masterd without a node vote (if start_daemons is True), but still non-interactively

Returns: None

StopMaster(stop_daemons)

source code

Deactivate this node as master.

The function will always try to deactivate the IP address of the master. It will also stop the master daemons depending on the stop_daemons parameter.

Parameters:

stop_daemons (boolean) - whether to also stop the master daemons (ganeti-masterd and ganeti-rapi)

Returns: None

EtcHostsModify(mode, host, ip)

source code

Modify a host entry in /etc/hosts.

Parameters:

mode - The mode to operate. Either add or remove entry
host - The host to operate on
ip - The ip associated with the entry

LeaveCluster(modify_ssh_setup)

source code

Cleans up and remove the current node.

This function cleans up and prepares the current node to be removed from the cluster.

If processing is successful, then it raises an errors.QuitGanetiException which is used as a special case to shutdown the node daemon.

Parameters:

modify_ssh_setup - boolean

GetNodeInfo(vgname, hypervisor_type)

source code

Gives back a hash with different information about the node.

Parameters:

vgname (string) - the name of the volume group to ask for disk space information
hypervisor_type (str) - the name of the hypervisor to ask for memory information

Returns: dict

dictionary with the following keys:

vg_size is the size of the configured volume group in MiB
vg_free is the free size of the volume group in MiB
memory_dom0 is the memory allocated for domain0 in MiB
memory_free is the currently available (free) ram in MiB
memory_total is the total number of ram in MiB

VerifyNode(what, cluster_name)

source code

Verify the status of the local node.

Based on the input what parameter, various checks are done on the local node.

If the filelist key is present, this list of files is checksummed and the file/checksum pairs are returned.

If the nodelist key is present, we check that we have connectivity via ssh with the target nodes (and check the hostname report).

If the node-net-test key is present, we check that we have connectivity to the given nodes via both primary IP and, if applicable, secondary IPs.

Parameters:

what (dict) - a dictionary of things to check:
- filelist: list of files for which to compute checksums
- nodelist: list of nodes we should check ssh communication with
- node-net-test: list of nodes we should check node daemon port connectivity with
- hypervisor: list with hypervisors to run the verify for

Returns: dict

a dictionary with the same keys as the input dict, and values representing the result of the checks

GetBlockDevSizes(devices)

source code

Return the size of the given block devices

Parameters:

devices (list) - list of block device nodes to query

Returns: dict

dictionary of all block devices under /dev (key). The value is their size in MiB.

{'/dev/disk/by-uuid/123456-12321231-312312-312': 124}

GetVolumeList(vg_names)

source code

Compute list of logical volumes and their size.

Parameters:

vg_names (list) - the volume groups whose LVs we should list, or empty for all volume groups

Returns: dict

dictionary of all partions (key) with value being a tuple of their size (in MiB), inactive and online status:

 {'xenvg/test1': ('20.06', True, True)}

in case of errors, a string is returned with the error details.

ListVolumeGroups()

source code

List the volume groups and their size.

Returns: dict: dictionary with keys volume name and values the size of the volume

NodeVolumes()

source code

List all volumes on this node.

Returns: list

A list of dictionaries, each having four keys:

name: the logical volume name,
size: the size of the logical volume
dev: the physical device on which the LV lives
vg: the volume group to which it belongs

In case of errors, we return an empty list and log the error.

Note that since a logical volume can live on multiple physical volumes, the resulting list might include a logical volume multiple times.

BridgesExist(bridges_list)

source code

Check if a list of bridges exist on the current node.

Returns: boolean: True if all of them exist, False otherwise

GetInstanceList(hypervisor_list)

source code

Provides a list of instances.

Parameters:

hypervisor_list (list) - the list of hypervisors to query information

Returns: list

a list of all running instances on the current node

instance1.example.com
instance2.example.com

GetInstanceInfo(instance, hname)

source code

Gives back the information about an instance as a dictionary.

Parameters:

instance (string) - the instance name
hname (string) - the hypervisor type of the instance

Returns: dict

dictionary with the following keys:

memory: memory size of instance (int)
state: xen state of instance (string)
time: cpu time of instance (float)

GetInstanceMigratable(instance)

source code

Gives whether an instance can be migrated.

Parameters:

instance (objects.Instance) - object representing the instance to be checked.

Returns: tuple

tuple of (result, description) where:

result: whether the instance can be migrated or not
description: a description of the issue, if relevant

GetAllInstancesInfo(hypervisor_list)

source code

Gather data about all instances.

This is the equivalent of GetInstanceInfo, except that it computes data for all instances at once, thus being faster if one needs data about more than one instance.

Parameters:

hypervisor_list (list) - list of hypervisors to query for instance data

Returns: dict

dictionary of instance: data, with data having the following keys:

memory: memory size of instance (int)
state: xen state of instance (string)
time: cpu time of instance (float)
vcpus: the number of vcpus

_InstanceLogName(kind, os_name, instance, component)

source code

Compute the OS log filename for a given instance and operation.

The instance name and os name are passed in as strings since not all operations have these as part of an instance object.

Parameters:

kind (string) - the operation type (e.g. add, import, etc.)
os_name (string) - the os name
instance (string) - the name of the instance being imported/added/etc.
component (string or None) - the name of the component of the instance being transferred

InstanceOsAdd(instance, reinstall, debug)

source code

Add an OS to an instance.

Parameters:

instance (objects.Instance) - Instance whose OS is to be installed
reinstall (boolean) - whether this is an instance reinstall
debug (integer) - debug level, passed to the OS scripts

Returns: None

RunRenameInstance(instance, old_name, debug)

source code

Run the OS rename script for an instance.

Parameters:

instance (objects.Instance) - Instance whose OS is to be installed
old_name (string) - previous instance name
debug (integer) - debug level, passed to the OS scripts

Returns: boolean

the success of the operation

_SymlinkBlockDev(instance_name, device_path, idx)

source code

Set up symlinks to a instance's block device.

This is an auxiliary function run when an instance is start (on the primary node) or when an instance is migrated (on the target node).

Parameters:

instance_name - the name of the target instance
device_path - path of the physical block device, on the node
idx - the disk index

Returns:

absolute path to the disk's symlink

_GatherAndLinkBlockDevs(instance)

source code

Set up an instance's block device(s).

This is run on the primary node at instance startup. The block devices must be already assembled.

Parameters:

instance (objects.Instance) - the instance whose disks we shoul assemble

Returns: list

list of (disk_object, device_path)

StartInstance(instance, startup_paused)

source code

Start an instance.

Parameters:

instance (objects.Instance) - the instance object
instance (objects.Instance) - pause instance at startup?
startup_paused (bool)

Returns: None

InstanceShutdown(instance, timeout)

source code

Shut an instance down.

Parameters:

instance (objects.Instance) - the instance object
timeout (integer) - maximum timeout for soft shutdown

Returns: None

Note: this functions uses polling with a hardcoded timeout.

InstanceReboot(instance, reboot_type, shutdown_timeout)

source code

Reboot an instance.

Parameters:

instance (objects.Instance) - the instance object to reboot
reboot_type (str) - the type of reboot, one the following constants:
- constants.INSTANCE_REBOOT_SOFT: only reboot the instance OS, do not recreate the VM
- constants.INSTANCE_REBOOT_HARD: tear down and restart the VM (at the hypervisor level)
- the other reboot type (constants.INSTANCE_REBOOT_FULL) is not accepted here, since that mode is handled differently, in cmdlib, and translates into full stop and start of the instance (instead of a call_instance_reboot RPC)
shutdown_timeout (integer) - maximum timeout for soft shutdown

Returns: None

MigrationInfo(instance)

source code

Gather information about an instance to be migrated.

Parameters:

instance (objects.Instance) - the instance definition

AcceptInstance(instance, info, target)

source code

Prepare the node to accept an instance.

Parameters:

instance (objects.Instance) - the instance definition
info (string/data (opaque)) - migration information, from the source node
target (string) - target host (usually ip), on this node

FinalizeMigration(instance, info, success)

source code

Finalize any preparation to accept an instance.

Parameters:

instance (objects.Instance) - the instance definition
info (string/data (opaque)) - migration information, from the source node
success (boolean) - whether the migration was a success or a failure

MigrateInstance(instance, target, live)

source code

Migrates an instance to another node.

Parameters:

instance (objects.Instance) - the instance definition
target (string) - the target node name
live (boolean) - whether the migration should be done live or not (the interpretation of this parameter is left to the hypervisor)

Returns: tuple

a tuple of (success, msg) where:

succes is a boolean denoting the success/failure of the operation
msg is a string with details in case of failure

BlockdevCreate(disk, size, owner, on_primary, info)

source code

Creates a block device for an instance.

Parameters:

disk (objects.Disk) - the object describing the disk we should create
size (int) - the size of the physical underlying device, in MiB
owner (str) - the name of the instance for which disk is created, used for device cache data
on_primary (boolean) - indicates if it is the primary node or not
info (string) - string that will be sent to the physical device creation, used for example to set (LVM) tags on LVs

Returns:

the new unique_id of the device (this can sometime be computed only after creation), or None. On secondary nodes, it's not required to return anything.

_WipeDevice(path, offset, size)

source code

This function actually wipes the device.

Parameters:

path - The path to the device to wipe
offset - The offset in MiB in the file
size - The size in MiB to write

BlockdevWipe(disk, offset, size)

source code

Wipes a block device.

Parameters:

disk (objects.Disk) - the disk object we want to wipe
offset (int) - The offset in MiB in the file
size (int) - The size in MiB to write

BlockdevPauseResumeSync(disks, pause)

source code

Pause or resume the sync of the block device.

Parameters:

disks (list of objects.Disk) - the disks object we want to pause/resume
pause (bool) - Wheater to pause or resume

BlockdevRemove(disk)

source code

Remove a block device.

Parameters:

disk (objects.Disk) - the disk object we should remove

Returns: boolean

the success of the operation

Note: This is intended to be called recursively.

_RecursiveAssembleBD(disk, owner, as_primary)

source code

Activate a block device for an instance.

This is run on the primary and secondary nodes for an instance.

Parameters:

disk (objects.Disk) - the disk we try to assemble
owner (str) - the name of the instance which owns the disk
as_primary (boolean) - if we should make the block device read/write

Returns:

the assembled device or None (in case no device was assembled)

Raises:

errors.BlockDeviceError - in case there is an error during the activation of the children or the device itself

Note: this function is called recursively.

BlockdevAssemble(disk, owner, as_primary, idx)

source code

Activate a block device for an instance.

This is a wrapper over _RecursiveAssembleBD.

Returns: str or boolean: a /dev/... path for primary nodes, and True for secondary nodes

BlockdevShutdown(disk)

source code

Shut down a block device.

First, if the device is assembled (Attach() is successful), then the device is shutdown. Then the children of the device are shutdown.

This function is called recursively. Note that we don't cache the children or such, as oppossed to assemble, shutdown of different devices doesn't require that the upper device was active.

Parameters:

disk (objects.Disk) - the description of the disk we should shutdown

Returns: None

BlockdevAddchildren(parent_cdev, new_cdevs)

source code

Extend a mirrored block device.

Parameters:

parent_cdev (objects.Disk) - the disk to which we should add children
new_cdevs (list of objects.Disk) - the list of children which we should add

Returns: None

BlockdevRemovechildren(parent_cdev, new_cdevs)

source code

Shrink a mirrored block device.

Parameters:

parent_cdev (objects.Disk) - the disk from which we should remove children
new_cdevs (list of objects.Disk) - the list of children which we should remove

Returns: None

BlockdevGetmirrorstatus(disks)

source code

Get the mirroring status of a list of devices.

Parameters:

disks (list of objects.Disk) - the list of disks which we should query

Returns: disk

List of objects.BlockDevStatus, one for each disk

Raises:

errors.BlockDeviceError - if any of the disks cannot be found

BlockdevGetmirrorstatusMulti(disks)

source code

Get the mirroring status of a list of devices.

Parameters:

disks (list of objects.Disk) - the list of disks which we should query

Returns: disk

List of tuples, (bool, status), one for each disk; bool denotes success/failure, status is objects.BlockDevStatus on success, string otherwise

_RecursiveFindBD(disk)

source code

Check if a device is activated.

If so, return information about the real device.

Parameters:

disk (objects.Disk) - the disk object we need to find

Returns:

None if the device can't be found, otherwise the device instance

_OpenRealBD(disk)

source code

Opens the underlying block device of a disk.

Parameters:

disk (objects.Disk) - the disk object we want to open

BlockdevFind(disk)

source code

Check if a device is activated.

If it is, return information about the real device.

Parameters:

disk (objects.Disk) - the disk to find

Returns: None or objects.BlockDevStatus

None if the disk cannot be found, otherwise a the current information

BlockdevGetsize(disks)

source code

Computes the size of the given disks.

If a disk is not found, returns None instead.

Parameters:

disks (list of objects.Disk) - the list of disk to compute the size for

Returns: list

list with elements None if the disk cannot be found, otherwise the size

BlockdevExport(disk, dest_node, dest_path, cluster_name)

source code

Export a block device to a remote node.

Parameters:

disk (objects.Disk) - the description of the disk to export
dest_node (str) - the destination node to export to
dest_path (str) - the destination path on the target node
cluster_name (str) - the cluster name, needed for SSH hostalias

Returns: None

UploadFile(file_name, data, mode, uid, gid, atime, mtime)

source code

Write a file to the filesystem.

This allows the master to overwrite(!) a file. It will only perform the operation if the file belongs to a list of configuration files.

Parameters:

file_name (str) - the target file name
data (str) - the new contents of the file
mode (int) - the mode to give the file (can be None)
uid (string) - the owner of the file
gid (string) - the group of the file
atime (float) - the atime to set on the file (can be None)
mtime (float) - the mtime to set on the file (can be None)

Returns: None

RunOob(oob_program, command, node, timeout)

source code

Executes oob_program with given command on given node.

Parameters:

oob_program - The path to the executable oob_program
command - The command to invoke on oob_program
node - The node given as an argument to the program
timeout - Timeout after which we kill the oob program

Returns:

stdout

Raises:

RPCFail - If execution fails for some reason

WriteSsconfFiles(values)

source code

Update all ssconf files.

Wrapper around the SimpleStore.WriteFiles.

_ErrnoOrStr(err)

source code

Format an EnvironmentError exception.

If the err argument has an errno attribute, it will be looked up and converted into a textual E... description. Otherwise the string representation of the error will be returned.

Parameters:

err (EnvironmentError) - the exception to format

_OSOndiskAPIVersion(os_dir)

source code

Compute and return the API version of a given OS.

This function will try to read the API version of the OS residing in the 'os_dir' directory.

Parameters:

os_dir (str) - the directory in which we should look for the OS

Returns: tuple

tuple (status, data) with status denoting the validity and data holding either the vaid versions or an error message

DiagnoseOS(top_dirs=None)

source code

Compute the validity for all OSes.

Parameters:

top_dirs (list) - the list of directories in which to search (if not given defaults to constants.OS_SEARCH_PATH)

Returns: list of objects.OS

a list of tuples (name, path, status, diagnose, variants, parameters, api_version) 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
variants is a list of supported OS variants, if any
parameters is a list of (name, help) parameters, if any
api_version is a list of support OS API versions

_TryOSFromDisk(name, base_dir=None)

source code

Create an OS instance from disk.

This function will return an OS instance if the given name is a valid OS name.

Parameters:

base_dir (string) - Base directory containing OS installations. Defaults to a search in all the OS_SEARCH_PATH dirs.

Returns: tuple

success and either the OS instance if we find a valid one, or error message

OSFromDisk(name, base_dir=None)

source code

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 RPCFail exception, detailing why this is not a valid OS.

This is just a wrapper over _TryOSFromDisk, which doesn't raise an exception but returns true/false status data.

Parameters:

base_dir (string) - Base directory containing OS installations. Defaults to a search in all the OS_SEARCH_PATH dirs.

Returns: objects.OS

the OS instance if we find a valid one

Raises:

RPCFail - if we don't find a valid OS

OSCoreEnv(os_name, inst_os, os_params, debug=0)

source code

Calculate the basic environment for an os script.

Parameters:

os_name (str) - full operating system name (including variant)
inst_os (objects.OS) - operating system for which the environment is being built
os_params (dict) - the OS parameters
debug (integer) - debug level (0 or 1, for OS Api 10)

Returns: dict

dict of environment variables

Raises:

errors.BlockDeviceError - if the block device cannot be found

OSEnvironment(instance, inst_os, debug=0)

source code

Calculate the environment for an os script.

Parameters:

instance (objects.Instance) - target instance for the os script run
inst_os (objects.OS) - operating system for which the environment is being built
debug (integer) - debug level (0 or 1, for OS Api 10)

Returns: dict

dict of environment variables

Raises:

errors.BlockDeviceError - if the block device cannot be found

BlockdevGrow(disk, amount, dryrun)

source code

Grow a stack of block devices.

This function is called recursively, with the childrens being the first ones to resize.

Parameters:

disk (objects.Disk) - the disk to be grown
amount (integer) - the amount (in mebibytes) to grow with
dryrun (boolean) - whether to execute the operation in simulation mode only, without actually increasing the size

Returns: (status, result)

a tuple with the status of the operation (True/False), and the errors message if status is False

BlockdevSnapshot(disk)

source code

Create a snapshot copy of a block device.

This function is called recursively, and the snapshot is actually created just for the leaf lvm backend device.

Parameters:

disk (objects.Disk) - the disk to be snapshotted

Returns: string

snapshot disk ID as (vg, lv)

FinalizeExport(instance, snap_disks)

source code

Write out the export configuration information.

Parameters:

instance (objects.Instance) - the instance which we export, used for saving configuration
snap_disks (list of objects.Disk) - list of snapshot block devices, which will be used to get the actual name of the dump file

Returns: None

ExportInfo(dest)

source code

Get export configuration information.

Parameters:

dest (str) - directory containing the export

Returns: objects.SerializableConfigParser

a serializable config file containing the export info

ListExports()

source code

Return a list of exports currently available on this machine.

Returns: list: list of the exports

RemoveExport(export)

source code

Remove an existing export from the node.

Parameters:

export (str) - the name of the export to remove

Returns: None

BlockdevRename(devlist)

source code

Rename a list of block devices.

Parameters:

devlist (list of tuples) - list of tuples of the form (disk, new_logical_id, new_physical_id); disk is an objects.Disk object describing the current disk, and new logical_id/physical_id is the name we rename it to

Returns: boolean

True if all renames succeeded, False otherwise

_TransformFileStorageDir(fs_dir)

source code

Checks whether given file_storage_dir is valid.

Checks wheter the given fs_dir is within the cluster-wide default file_storage_dir or the shared_file_storage_dir, which are stored in SimpleStore. Only paths under those directories are allowed.

Parameters:

fs_dir (str) - the path to check

Returns:

the normalized path if valid, None otherwise

CreateFileStorageDir(file_storage_dir)

source code

Create file storage directory.

Parameters:

file_storage_dir (str) - directory to create

Returns: tuple

tuple with first element a boolean indicating wheter dir creation was successful or not

RemoveFileStorageDir(file_storage_dir)

source code

Remove file storage directory.

Remove it only if it's empty. If not log an error and return.

Parameters:

file_storage_dir (str) - the directory we should cleanup

Returns: tuple (success,)

tuple of one element, success, denoting whether the operation was successful

RenameFileStorageDir(old_file_storage_dir, new_file_storage_dir)

source code

Rename the file storage directory.

Parameters:

old_file_storage_dir (str) - the current path
new_file_storage_dir (str) - the name we should rename to

Returns: tuple (success,)

tuple of one element, success, denoting whether the operation was successful

_EnsureJobQueueFile(file_name)

source code

Checks whether the given filename is in the queue directory.

Parameters:

file_name (str) - the file name we should check

Returns: None

Raises:

RPCFail - if the file is not valid

JobQueueUpdate(file_name, content)

source code

Updates a file in the queue directory.

This is just a wrapper over utils.io.WriteFile, with proper checking.

Parameters:

file_name (str) - the job file name
content (str) - the new job contents

Returns: boolean

the success of the operation

JobQueueRename(old, new)

source code

Renames a job queue file.

This is just a wrapper over os.rename with proper checking.

Parameters:

old (str) - the old (actual) file name
new (str) - the desired file name

Returns: tuple

the success of the operation and payload

BlockdevClose(instance_name, disks)

source code

Closes the given block devices.

This means they will be switched to secondary mode (in case of DRBD).

Parameters:

instance_name - if the argument is not empty, the symlinks of this instance will be removed
disks (list of objects.Disk) - the list of disks to be closed

Returns: tuple (success, message)

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

ValidateHVParams(hvname, hvparams)

source code

Validates the given hypervisor parameters.

Parameters:

hvname (string) - the hypervisor name
hvparams (dict) - the hypervisor parameters to be validated

Returns: None

_CheckOSPList(os_obj, parameters)

source code

Check whether a list of parameters is supported by the OS.

Parameters:

os_obj (objects.OS) - OS object to check
parameters (list) - the list of parameters to check

ValidateOS(required, osname, checks, osparams)

source code

Validate the given OS' parameters.

Parameters:

required (boolean) - whether absence of the OS should translate into failure or not
osname (string) - the OS to be validated
checks (list) - list of the checks to run (currently only 'parameters')
osparams (dict) - dictionary with OS parameters

Returns: boolean

True if the validation passed, or False if the OS was not found and required was false

CreateX509Certificate(validity, cryptodir=constants.CRYPTO_KEYS_DIR)

source code

Creates a new X509 certificate for SSL/TLS.

Parameters:

validity (int) - Validity in seconds

Returns: tuple; (string, string)

Certificate name and public part

RemoveX509Certificate(name, cryptodir=constants.CRYPTO_KEYS_DIR)

source code

Removes a X509 certificate.

Parameters:

name (string) - Certificate name

_GetImportExportIoCommand(instance, mode, ieio, ieargs)

source code

Returns the command for the requested input/output.

Parameters:

instance (objects.Instance) - The instance object
mode - Import/export mode
ieio - Input/output type
ieargs - Input/output arguments

StartImportExportDaemon(mode, opts, host, port, instance, component, ieio, ieioargs)

source code

Starts an import or export daemon.

Parameters:

mode - Import/output mode
opts (objects.ImportExportOptions) - Daemon options
host (string) - Remote host for export (None for import)
port (int) - Remote port for export (None for import)
instance (objects.Instance) - Instance object
component (string) - which part of the instance is transferred now, e.g. 'disk/0'
ieio - Input/output type
ieioargs - Input/output arguments

GetImportExportStatus(names)

source code

Returns import/export daemon status.

Parameters:

names (sequence) - List of names

Returns: List of dicts

Returns a list of the state of each named import/export or None if a status couldn't be read

CleanupImportExport(name)

source code

Cleanup after an import or export.

If the import/export daemon is still running it's killed. Afterwards the whole status directory is removed.

PowercycleNode(hypervisor_type)

source code

Hard-powercycle the node.

Because we need to return first, and schedule the powercycle in the background, we won't be able to report failures nicely.

Variables Details

[hide private]

_ALLOWED_CLEAN_DIRS

denotes which directories are accepted in the _CleanDirectory function

Value:

frozenset([constants.DATA_DIR, constants.JOB_QUEUE_ARCHIVE_DIR, consta
nts.QUEUE_DIR, constants.CRYPTO_KEYS_DIR,])

_LVSLINE_REGEX

Valid LVS output line regex

Value:

re.compile("^ *([^|]+)\|([^|]+)\|([0-9.]+)\|([^|]{6,})\|?$")