import os.path
import shutil
import time
-import tempfile
import stat
import errno
import re
import subprocess
+import random
+import logging
+import tempfile
-from ganeti import logger
from ganeti import errors
from ganeti import utils
from ganeti import ssh
from ganeti import ssconf
-def StartMaster():
- """Activate local node as master node.
+def _GetSshRunner():
+ return ssh.SshRunner()
- There are two needed steps for this:
- - run the master script
- - register the cron script
+
+def _CleanDirectory(path, exclude=[]):
+ """Removes all regular files in a directory.
+
+ @param exclude: List of files to be excluded.
+ @type exclude: list
"""
- result = utils.RunCmd([constants.MASTER_SCRIPT, "-d", "start"])
+ if not os.path.isdir(path):
+ return
- if result.failed:
- logger.Error("could not activate cluster interface with command %s,"
- " error: '%s'" % (result.cmd, result.output))
- return False
+ # Normalize excluded paths
+ exclude = [os.path.normpath(i) for i in exclude]
- return True
+ for rel_name in utils.ListVisibleFiles(path):
+ full_name = os.path.normpath(os.path.join(path, rel_name))
+ if full_name in exclude:
+ continue
+ if os.path.isfile(full_name) and not os.path.islink(full_name):
+ utils.RemoveFile(full_name)
+
+
+def _JobQueuePurge(keep_lock):
+ """Removes job queue files and archived jobs
+
+ """
+ if keep_lock:
+ exclude = [constants.JOB_QUEUE_LOCK_FILE]
+ else:
+ exclude = []
+
+ _CleanDirectory(constants.QUEUE_DIR, exclude=exclude)
+ _CleanDirectory(constants.JOB_QUEUE_ARCHIVE_DIR)
+
+
+def _GetMasterInfo():
+ """Return the master ip and netdev.
+
+ """
+ try:
+ ss = ssconf.SimpleStore()
+ master_netdev = ss.GetMasterNetdev()
+ master_ip = ss.GetMasterIP()
+ except errors.ConfigurationError, err:
+ logging.exception("Cluster configuration incomplete")
+ return (None, None)
+ return (master_netdev, master_ip)
+
+
+def StartMaster(start_daemons):
+ """Activate local node as master node.
+ The function will always try activate the IP address of the master
+ (if someone else has it, then it won't). Then, if the start_daemons
+ parameter is True, it will also start the master daemons
+ (ganet-masterd and ganeti-rapi).
-def StopMaster():
+ """
+ ok = True
+ master_netdev, master_ip = _GetMasterInfo()
+ if not master_netdev:
+ return False
+
+ if utils.TcpPing(master_ip, constants.DEFAULT_NODED_PORT):
+ if utils.TcpPing(master_ip, constants.DEFAULT_NODED_PORT,
+ source=constants.LOCALHOST_IP_ADDRESS):
+ # we already have the ip:
+ logging.debug("Already started")
+ else:
+ logging.error("Someone else has the master ip, not activating")
+ ok = False
+ else:
+ result = utils.RunCmd(["ip", "address", "add", "%s/32" % master_ip,
+ "dev", master_netdev, "label",
+ "%s:0" % master_netdev])
+ if result.failed:
+ logging.error("Can't activate master IP: %s", result.output)
+ ok = False
+
+ result = utils.RunCmd(["arping", "-q", "-U", "-c 3", "-I", master_netdev,
+ "-s", master_ip, master_ip])
+ # we'll ignore the exit code of arping
+
+ # and now start the master and rapi daemons
+ if start_daemons:
+ for daemon in 'ganeti-masterd', 'ganeti-rapi':
+ result = utils.RunCmd([daemon])
+ if result.failed:
+ logging.error("Can't start daemon %s: %s", daemon, result.output)
+ ok = False
+ return ok
+
+
+def StopMaster(stop_daemons):
"""Deactivate this node as master.
- This does two things:
- - run the master stop script
- - remove link to master cron script.
+ The function will always try to deactivate the IP address of the
+ master. Then, if the stop_daemons parameter is True, it will also
+ stop the master daemons (ganet-masterd and ganeti-rapi).
"""
- result = utils.RunCmd([constants.MASTER_SCRIPT, "-d", "stop"])
+ master_netdev, master_ip = _GetMasterInfo()
+ if not master_netdev:
+ return False
+ result = utils.RunCmd(["ip", "address", "del", "%s/32" % master_ip,
+ "dev", master_netdev])
if result.failed:
- logger.Error("could not deactivate cluster interface with command %s,"
- " error: '%s'" % (result.cmd, result.output))
- return False
+ logging.error("Can't remove the master IP, error: %s", result.output)
+ # but otherwise ignore the failure
+
+ if stop_daemons:
+ # stop/kill the rapi and the master daemon
+ for daemon in constants.RAPI_PID, constants.MASTERD_PID:
+ utils.KillProcess(utils.ReadPidFile(utils.DaemonPidFileName(daemon)))
return True
priv_key, pub_key, auth_keys = ssh.GetUserFiles(constants.GANETI_RUNAS,
mkdir=True)
except errors.OpExecError, err:
- logger.Error("Error while processing user ssh files: %s" % err)
+ logging.exception("Error while processing user ssh files")
return False
for name, content in [(priv_key, sshkey), (pub_key, sshpub)]:
"""Cleans up the current node and prepares it to be removed from the cluster.
"""
- if os.path.isdir(constants.DATA_DIR):
- for rel_name in utils.ListVisibleFiles(constants.DATA_DIR):
- full_name = os.path.join(constants.DATA_DIR, rel_name)
- if os.path.isfile(full_name) and not os.path.islink(full_name):
- utils.RemoveFile(full_name)
+ _CleanDirectory(constants.DATA_DIR)
+
+ # The lock can be removed because we're going to quit anyway.
+ _JobQueuePurge(keep_lock=False)
try:
priv_key, pub_key, auth_keys = ssh.GetUserFiles(constants.GANETI_RUNAS)
- except errors.OpExecError, err:
- logger.Error("Error while processing ssh files: %s" % err)
+ except errors.OpExecError:
+ logging.exception("Error while processing ssh files")
return
f = open(pub_key, 'r')
utils.RemoveFile(priv_key)
utils.RemoveFile(pub_key)
+ # Return a reassuring string to the caller, and quit
+ raise errors.QuitGanetiException(False, 'Shutdown scheduled')
+
def GetNodeInfo(vgname):
"""Gives back a hash with different informations about the node.
if 'nodelist' in what:
result['nodelist'] = {}
+ random.shuffle(what['nodelist'])
for node in what['nodelist']:
- success, message = ssh.VerifyNodeHostname(node)
+ success, message = _GetSshRunner().VerifyNodeHostname(node)
if not success:
result['nodelist'][node] = message
+ if 'node-net-test' in what:
+ result['node-net-test'] = {}
+ my_name = utils.HostInfo().name
+ my_pip = my_sip = None
+ for name, pip, sip in what['node-net-test']:
+ if name == my_name:
+ my_pip = pip
+ my_sip = sip
+ break
+ if not my_pip:
+ result['node-net-test'][my_name] = ("Can't find my own"
+ " primary/secondary IP"
+ " in the node list")
+ else:
+ port = ssconf.SimpleStore().GetNodeDaemonPort()
+ for name, pip, sip in what['node-net-test']:
+ fail = []
+ if not utils.TcpPing(pip, port, source=my_pip):
+ fail.append("primary")
+ if sip != pip:
+ if not utils.TcpPing(sip, port, source=my_sip):
+ fail.append("secondary")
+ if fail:
+ result['node-net-test'][name] = ("failure using the %s"
+ " interface(s)" %
+ " and ".join(fail))
+
return result
"--separator=%s" % sep,
"-olv_name,lv_size,lv_attr", vg_name])
if result.failed:
- logger.Error("Failed to list logical volumes, lvs output: %s" %
- result.output)
- return lvs
+ logging.error("Failed to list logical volumes, lvs output: %s",
+ result.output)
+ return result.output
+ valid_line_re = re.compile("^ *([^|]+)\|([0-9.]+)\|([^|]{6})\|?$")
for line in result.stdout.splitlines():
- line = line.strip().rstrip(sep)
- name, size, attr = line.split(sep)
- if len(attr) != 6:
- attr = '------'
+ line = line.strip()
+ match = valid_line_re.match(line)
+ if not match:
+ logging.error("Invalid line returned from lvs output: '%s'", line)
+ continue
+ name, size, attr = match.groups()
inactive = attr[4] == '-'
online = attr[5] == 'o'
lvs[name] = (size, inactive, online)
"--separator=|",
"--options=lv_name,lv_size,devices,vg_name"])
if result.failed:
- logger.Error("Failed to list logical volumes, lvs output: %s" %
- result.output)
+ logging.error("Failed to list logical volumes, lvs output: %s",
+ result.output)
return {}
def parse_dev(dev):
'vg': line[3].strip(),
}
- return [map_line(line.split('|')) for line in result.stdout.splitlines()]
+ return [map_line(line.split('|')) for line in result.stdout.splitlines()
+ if line.count('|') >= 3]
def BridgesExist(bridges_list):
try:
names = hypervisor.GetHypervisor().ListInstances()
except errors.HypervisorError, err:
- logger.Error("error enumerating instances: %s" % str(err))
+ logging.exception("Error enumerating instances")
raise
return names
os_device = instance.FindDisk(os_disk)
if os_device is None:
- logger.Error("Can't find this device-visible name '%s'" % os_disk)
+ logging.error("Can't find this device-visible name '%s'", os_disk)
return False
swap_device = instance.FindDisk(swap_disk)
if swap_device is None:
- logger.Error("Can't find this device-visible name '%s'" % swap_disk)
+ logging.error("Can't find this device-visible name '%s'", swap_disk)
return False
real_os_dev = _RecursiveFindBD(os_device)
inst_os.path, create_script, instance.name,
real_os_dev.dev_path, real_swap_dev.dev_path,
logfile)
+ env = {'HYPERVISOR': ssconf.SimpleStore().GetHypervisorType()}
- result = utils.RunCmd(command)
+ result = utils.RunCmd(command, env=env)
if result.failed:
- logger.Error("os create command '%s' returned error: %s, logfile: %s,"
- " output: %s" %
- (command, result.fail_reason, logfile, result.output))
+ logging.error("os create command '%s' returned error: %s, logfile: %s,"
+ " output: %s", command, result.fail_reason, logfile,
+ result.output)
return False
return True
os_device = instance.FindDisk(os_disk)
if os_device is None:
- logger.Error("Can't find this device-visible name '%s'" % os_disk)
+ logging.error("Can't find this device-visible name '%s'", os_disk)
return False
swap_device = instance.FindDisk(swap_disk)
if swap_device is None:
- logger.Error("Can't find this device-visible name '%s'" % swap_disk)
+ logging.error("Can't find this device-visible name '%s'", swap_disk)
return False
real_os_dev = _RecursiveFindBD(os_device)
result = utils.RunCmd(command)
if result.failed:
- logger.Error("os create command '%s' returned error: %s"
- " output: %s" %
- (command, result.fail_reason, result.output))
+ logging.error("os create command '%s' returned error: %s output: %s",
+ command, result.fail_reason, result.output)
return False
return True
vg_free is the free size of the volume group in MiB
pv_count are the number of physical disks in that vg
+ If an error occurs during gathering of data, we return the same dict
+ with keys all set to None.
+
"""
+ retdic = dict.fromkeys(["vg_size", "vg_free", "pv_count"])
+
retval = utils.RunCmd(["vgs", "-ovg_size,vg_free,pv_count", "--noheadings",
"--nosuffix", "--units=m", "--separator=:", vg_name])
if retval.failed:
- errmsg = "volume group %s not present" % vg_name
- logger.Error(errmsg)
- raise errors.LVMError(errmsg)
- valarr = retval.stdout.strip().split(':')
- retdic = {
- "vg_size": int(round(float(valarr[0]), 0)),
- "vg_free": int(round(float(valarr[1]), 0)),
- "pv_count": int(valarr[2]),
- }
+ logging.error("volume group %s not present", vg_name)
+ return retdic
+ valarr = retval.stdout.strip().rstrip(':').split(':')
+ if len(valarr) == 3:
+ try:
+ retdic = {
+ "vg_size": int(round(float(valarr[0]), 0)),
+ "vg_free": int(round(float(valarr[1]), 0)),
+ "pv_count": int(valarr[2]),
+ }
+ except ValueError, err:
+ logging.exception("Fail to parse vgs output")
+ else:
+ logging.error("vgs output has the wrong number of fields (expected"
+ " three): %s", str(valarr))
return retdic
try:
hyper.StartInstance(instance, block_devices, extra_args)
except errors.HypervisorError, err:
- logger.Error("Failed to start instance: %s" % err)
+ logging.exception("Failed to start instance")
return False
return True
try:
hyper.StopInstance(instance)
except errors.HypervisorError, err:
- logger.Error("Failed to stop instance: %s" % err)
+ logging.error("Failed to stop instance")
return False
# test every 10secs for 2min
time.sleep(10)
else:
# the shutdown did not succeed
- logger.Error("shutdown of '%s' unsuccessful, using destroy" % instance)
+ logging.error("shutdown of '%s' unsuccessful, using destroy", instance)
try:
hyper.StopInstance(instance, force=True)
except errors.HypervisorError, err:
- logger.Error("Failed to stop instance: %s" % err)
+ logging.exception("Failed to stop instance")
return False
time.sleep(1)
if instance.name in GetInstanceList():
- logger.Error("could not shutdown instance '%s' even by destroy")
+ logging.error("could not shutdown instance '%s' even by destroy",
+ instance.name)
return False
return True
running_instances = GetInstanceList()
if instance.name not in running_instances:
- logger.Error("Cannot reboot instance that is not running")
+ logging.error("Cannot reboot instance that is not running")
return False
hyper = hypervisor.GetHypervisor()
try:
hyper.RebootInstance(instance)
except errors.HypervisorError, err:
- logger.Error("Failed to soft reboot instance: %s" % err)
+ logging.exception("Failed to soft reboot instance")
return False
elif reboot_type == constants.INSTANCE_REBOOT_HARD:
try:
ShutdownInstance(instance)
StartInstance(instance, extra_args)
except errors.HypervisorError, err:
- logger.Error("Failed to hard reboot instance: %s" % err)
+ logging.exception("Failed to hard reboot instance")
return False
else:
raise errors.ParameterError("reboot_type invalid")
return True
+def MigrateInstance(instance, target, live):
+ """Migrates an instance to another node.
+
+ """
+ hyper = hypervisor.GetHypervisor()
+
+ try:
+ hyper.MigrateInstance(instance, target, live)
+ except errors.HypervisorError, err:
+ msg = "Failed to migrate instance: %s" % str(err)
+ logging.error(msg)
+ return (False, msg)
+ return (True, "Migration successfull")
+
+
def CreateBlockDevice(disk, size, owner, on_primary, info):
"""Creates a block device for an instance.
Args:
- bdev: a ganeti.objects.Disk object
- size: the size of the physical underlying devices
- do_open: if the device should be `Assemble()`-d and
- `Open()`-ed after creation
+ disk: a ganeti.objects.Disk object
+ size: the size of the physical underlying device
+ owner: a string with the name of the instance
+ on_primary: a boolean indicating if it is the primary node or not
+ info: string that will be sent to the physical device creation
Returns:
the new unique_id of the device (this can sometime be
# we need the children open in case the device itself has to
# be assembled
crdev.Open()
- else:
- crdev.Close()
clist.append(crdev)
try:
device = bdev.FindDevice(disk.dev_type, disk.physical_id, clist)
if device is not None:
- logger.Info("removing existing device %s" % disk)
+ logging.info("removing existing device %s", disk)
device.Remove()
except errors.BlockDeviceError, err:
pass
if on_primary or disk.AssembleOnSecondary():
if not device.Assemble():
errorstring = "Can't assemble device after creation"
- logger.Error(errorstring)
+ logging.error(errorstring)
raise errors.BlockDeviceError("%s, very unusual event - check the node"
" daemon logs" % errorstring)
device.SetSyncSpeed(constants.SYNC_SPEED)
rdev = _RecursiveFindBD(disk, allow_partial=True)
except errors.BlockDeviceError, err:
# probably can't attach
- logger.Info("Can't attach to device %s in remove" % disk)
+ logging.info("Can't attach to device %s in remove", disk)
rdev = None
if rdev is not None:
r_path = rdev.dev_path
if children.count(None) >= mcn:
raise
cdev = None
- logger.Debug("Error in child activation: %s" % str(err))
+ logging.debug("Error in child activation: %s", str(err))
children.append(cdev)
if as_primary or disk.AssembleOnSecondary():
result = r_dev
if as_primary or disk.OpenOnSecondary():
r_dev.Open()
- else:
- r_dev.Close()
DevCacheManager.UpdateCache(r_dev.dev_path, owner,
as_primary, disk.iv_name)
"""
parent_bdev = _RecursiveFindBD(parent_cdev, allow_partial=True)
if parent_bdev is None:
- logger.Error("Can't find parent device")
+ logging.error("Can't find parent device")
return False
new_bdevs = [_RecursiveFindBD(disk) for disk in new_cdevs]
if new_bdevs.count(None) > 0:
- logger.Error("Can't find new device(s) to add: %s:%s" %
- (new_bdevs, new_cdevs))
+ logging.error("Can't find new device(s) to add: %s:%s",
+ new_bdevs, new_cdevs)
return False
parent_bdev.AddChildren(new_bdevs)
return True
"""
parent_bdev = _RecursiveFindBD(parent_cdev)
if parent_bdev is None:
- logger.Error("Can't find parent in remove children: %s" % parent_cdev)
+ logging.error("Can't find parent in remove children: %s", parent_cdev)
return False
devs = []
for disk in new_cdevs:
if rpath is None:
bd = _RecursiveFindBD(disk)
if bd is None:
- logger.Error("Can't find dynamic device %s while removing children" %
- disk)
+ logging.error("Can't find dynamic device %s while removing children",
+ disk)
return False
else:
devs.append(bd.dev_path)
"""
if not os.path.isabs(file_name):
- logger.Error("Filename passed to UploadFile is not absolute: '%s'" %
- file_name)
+ logging.error("Filename passed to UploadFile is not absolute: '%s'",
+ file_name)
return False
- allowed_files = [constants.CLUSTER_CONF_FILE, "/etc/hosts",
- constants.SSH_KNOWN_HOSTS_FILE]
+ allowed_files = [
+ constants.CLUSTER_CONF_FILE,
+ constants.ETC_HOSTS,
+ constants.SSH_KNOWN_HOSTS_FILE,
+ constants.VNC_PASSWORD_FILE,
+ ]
allowed_files.extend(ssconf.SimpleStore().GetFileList())
+
if file_name not in allowed_files:
- logger.Error("Filename passed to UploadFile not in allowed"
- " upload targets: '%s'" % file_name)
+ logging.error("Filename passed to UploadFile not in allowed"
+ " upload targets: '%s'", file_name)
return False
- dir_name, small_name = os.path.split(file_name)
- fd, new_name = tempfile.mkstemp('.new', small_name, dir_name)
- # here we need to make sure we remove the temp file, if any error
- # leaves it in place
- try:
- os.chown(new_name, uid, gid)
- os.chmod(new_name, mode)
- os.write(fd, data)
- os.fsync(fd)
- os.utime(new_name, (atime, mtime))
- os.rename(new_name, file_name)
- finally:
- os.close(fd)
- utils.RemoveFile(new_name)
+ utils.WriteFile(file_name, data=data, mode=mode, uid=uid, gid=gid,
+ atime=atime, mtime=mtime)
return True
return detail
-def _OSSearch(name, search_path=None):
- """Search for OSes with the given name in the search_path.
-
- Args:
- name: The name of the OS to look for
- search_path: List of dirs to search (defaults to constants.OS_SEARCH_PATH)
-
- Returns:
- The base_dir the OS resides in
-
- """
- if search_path is None:
- search_path = constants.OS_SEARCH_PATH
-
- for dir_name in search_path:
- t_os_dir = os.path.sep.join([dir_name, name])
- if os.path.isdir(t_os_dir):
- return dir_name
-
- return None
-
-
def _OSOndiskVersion(name, os_dir):
"""Compute and return the API version of a given OS.
try:
f_names = utils.ListVisibleFiles(dir_name)
except EnvironmentError, err:
- logger.Error("Can't list the OS directory %s: %s" %
- (dir_name, str(err)))
+ logging.exception("Can't list the OS directory %s", dir_name)
break
for name in f_names:
try:
"""
if base_dir is None:
- base_dir = _OSSearch(name)
-
- if base_dir is None:
- raise errors.InvalidOS(name, None, "OS dir not found in search path")
+ os_dir = utils.FindFile(name, constants.OS_SEARCH_PATH, os.path.isdir)
+ if os_dir is None:
+ raise errors.InvalidOS(name, None, "OS dir not found in search path")
+ else:
+ os_dir = os.path.sep.join([base_dir, name])
- os_dir = os.path.sep.join([base_dir, name])
api_version = _OSOndiskVersion(name, os_dir)
if api_version != constants.OS_API_VERSION:
api_version=api_version)
+def GrowBlockDevice(disk, amount):
+ """Grow a stack of block devices.
+
+ This function is called recursively, with the childrens being the
+ first one resize.
+
+ Args:
+ disk: the disk to be grown
+
+ Returns: a tuple of (status, result), with:
+ status: the result (true/false) of the operation
+ result: the error message if the operation failed, otherwise not used
+
+ """
+ r_dev = _RecursiveFindBD(disk)
+ if r_dev is None:
+ return False, "Cannot find block device %s" % (disk,)
+
+ try:
+ r_dev.Grow(amount)
+ except errors.BlockDeviceError, err:
+ return False, str(err)
+
+ return True, None
+
+
def SnapshotBlockDevice(disk):
"""Create a snapshot copy of a block device.
destcmd = utils.BuildShellCmd("mkdir -p %s && cat > %s/%s",
destdir, destdir, destfile)
- remotecmd = ssh.BuildSSHCmd(dest_node, constants.GANETI_RUNAS, destcmd)
-
-
+ remotecmd = _GetSshRunner().BuildCmd(dest_node, constants.GANETI_RUNAS,
+ destcmd)
# all commands have been checked, so we're safe to combine them
command = '|'.join([expcmd, comprcmd, utils.ShellQuoteArgs(remotecmd)])
result = utils.RunCmd(command)
if result.failed:
- logger.Error("os snapshot export command '%s' returned error: %s"
- " output: %s" %
- (command, result.fail_reason, result.output))
+ logging.error("os snapshot export command '%s' returned error: %s"
+ " output: %s", command, result.fail_reason, result.output)
return False
return True
config.set(constants.INISECT_INS, 'memory', '%d' % instance.memory)
config.set(constants.INISECT_INS, 'vcpus', '%d' % instance.vcpus)
config.set(constants.INISECT_INS, 'disk_template', instance.disk_template)
+
+ nic_count = 0
for nic_count, nic in enumerate(instance.nics):
config.set(constants.INISECT_INS, 'nic%d_mac' %
nic_count, '%s' % nic.mac)
config.set(constants.INISECT_INS, 'nic%d_ip' % nic_count, '%s' % nic.ip)
+ config.set(constants.INISECT_INS, 'nic%d_bridge' % nic_count,
+ '%s' % nic.bridge)
# TODO: redundant: on load can read nics until it doesn't exist
config.set(constants.INISECT_INS, 'nic_count' , '%d' % nic_count)
+ disk_count = 0
for disk_count, disk in enumerate(snap_disks):
config.set(constants.INISECT_INS, 'disk%d_ivname' % disk_count,
('%s' % disk.iv_name))
os_device = instance.FindDisk(os_disk)
if os_device is None:
- logger.Error("Can't find this device-visible name '%s'" % os_disk)
+ logging.error("Can't find this device-visible name '%s'", os_disk)
return False
swap_device = instance.FindDisk(swap_disk)
if swap_device is None:
- logger.Error("Can't find this device-visible name '%s'" % swap_disk)
+ logging.error("Can't find this device-visible name '%s'", swap_disk)
return False
real_os_dev = _RecursiveFindBD(os_device)
os.mkdir(constants.LOG_OS_DIR, 0750)
destcmd = utils.BuildShellCmd('cat %s', src_image)
- remotecmd = ssh.BuildSSHCmd(src_node, constants.GANETI_RUNAS, destcmd)
+ remotecmd = _GetSshRunner().BuildCmd(src_node, constants.GANETI_RUNAS,
+ destcmd)
comprcmd = "gunzip"
impcmd = utils.BuildShellCmd("(cd %s; %s -i %s -b %s -s %s &>%s)",
logfile)
command = '|'.join([utils.ShellQuoteArgs(remotecmd), comprcmd, impcmd])
+ env = {'HYPERVISOR': ssconf.SimpleStore().GetHypervisorType()}
- result = utils.RunCmd(command)
+ result = utils.RunCmd(command, env=env)
if result.failed:
- logger.Error("os import command '%s' returned error: %s"
- " output: %s" %
- (command, result.fail_reason, result.output))
+ logging.error("os import command '%s' returned error: %s"
+ " output: %s", command, result.fail_reason, result.output)
return False
return True
# cache? for now, we only lose lvm data when we rename, which
# is less critical than DRBD or MD
except errors.BlockDeviceError, err:
- logger.Error("Can't rename device '%s' to '%s': %s" %
- (dev, unique_id, err))
+ logging.exception("Can't rename device '%s' to '%s'", dev, unique_id)
result = False
return result
+def _TransformFileStorageDir(file_storage_dir):
+ """Checks whether given file_storage_dir is valid.
+
+ Checks wheter the given file_storage_dir is within the cluster-wide
+ default file_storage_dir stored in SimpleStore. Only paths under that
+ directory are allowed.
+
+ Args:
+ file_storage_dir: string with path
+
+ Returns:
+ normalized file_storage_dir (string) if valid, None otherwise
+
+ """
+ file_storage_dir = os.path.normpath(file_storage_dir)
+ base_file_storage_dir = ssconf.SimpleStore().GetFileStorageDir()
+ if (not os.path.commonprefix([file_storage_dir, base_file_storage_dir]) ==
+ base_file_storage_dir):
+ logging.error("file storage directory '%s' is not under base file"
+ " storage directory '%s'",
+ file_storage_dir, base_file_storage_dir)
+ return None
+ return file_storage_dir
+
+
+def CreateFileStorageDir(file_storage_dir):
+ """Create file storage directory.
+
+ Args:
+ file_storage_dir: string containing the path
+
+ Returns:
+ tuple with first element a boolean indicating wheter dir
+ creation was successful or not
+
+ """
+ file_storage_dir = _TransformFileStorageDir(file_storage_dir)
+ result = True,
+ if not file_storage_dir:
+ result = False,
+ else:
+ if os.path.exists(file_storage_dir):
+ if not os.path.isdir(file_storage_dir):
+ logging.error("'%s' is not a directory", file_storage_dir)
+ result = False,
+ else:
+ try:
+ os.makedirs(file_storage_dir, 0750)
+ except OSError, err:
+ logging.error("Cannot create file storage directory '%s': %s",
+ file_storage_dir, err)
+ result = False,
+ return result
+
+
+def RemoveFileStorageDir(file_storage_dir):
+ """Remove file storage directory.
+
+ Remove it only if it's empty. If not log an error and return.
+
+ Args:
+ file_storage_dir: string containing the path
+
+ Returns:
+ tuple with first element a boolean indicating wheter dir
+ removal was successful or not
+
+ """
+ file_storage_dir = _TransformFileStorageDir(file_storage_dir)
+ result = True,
+ if not file_storage_dir:
+ result = False,
+ else:
+ if os.path.exists(file_storage_dir):
+ if not os.path.isdir(file_storage_dir):
+ logging.error("'%s' is not a directory", file_storage_dir)
+ result = False,
+ # deletes dir only if empty, otherwise we want to return False
+ try:
+ os.rmdir(file_storage_dir)
+ except OSError, err:
+ logging.exception("Cannot remove file storage directory '%s'",
+ file_storage_dir)
+ result = False,
+ return result
+
+
+def RenameFileStorageDir(old_file_storage_dir, new_file_storage_dir):
+ """Rename the file storage directory.
+
+ Args:
+ old_file_storage_dir: string containing the old path
+ new_file_storage_dir: string containing the new path
+
+ Returns:
+ tuple with first element a boolean indicating wheter dir
+ rename was successful or not
+
+ """
+ old_file_storage_dir = _TransformFileStorageDir(old_file_storage_dir)
+ new_file_storage_dir = _TransformFileStorageDir(new_file_storage_dir)
+ result = True,
+ if not old_file_storage_dir or not new_file_storage_dir:
+ result = False,
+ else:
+ if not os.path.exists(new_file_storage_dir):
+ if os.path.isdir(old_file_storage_dir):
+ try:
+ os.rename(old_file_storage_dir, new_file_storage_dir)
+ except OSError, err:
+ logging.exception("Cannot rename '%s' to '%s'",
+ old_file_storage_dir, new_file_storage_dir)
+ result = False,
+ else:
+ logging.error("'%s' is not a directory", old_file_storage_dir)
+ result = False,
+ else:
+ if os.path.exists(old_file_storage_dir):
+ logging.error("Cannot rename '%s' to '%s'. Both locations exist.",
+ old_file_storage_dir, new_file_storage_dir)
+ result = False,
+ return result
+
+
+def _IsJobQueueFile(file_name):
+ """Checks whether the given filename is in the queue directory.
+
+ """
+ queue_dir = os.path.normpath(constants.QUEUE_DIR)
+ result = (os.path.commonprefix([queue_dir, file_name]) == queue_dir)
+
+ if not result:
+ logging.error("'%s' is not a file in the queue directory",
+ file_name)
+
+ return result
+
+
+def JobQueueUpdate(file_name, content):
+ """Updates a file in the queue directory.
+
+ """
+ if not _IsJobQueueFile(file_name):
+ return False
+
+ # Write and replace the file atomically
+ utils.WriteFile(file_name, data=content)
+
+ return True
+
+
+def JobQueuePurge():
+ """Removes job queue files and archived jobs
+
+ """
+ # The lock must not be removed, otherwise another process could create
+ # it again.
+ return _JobQueuePurge(keep_lock=True)
+
+
+def JobQueueRename(old, new):
+ """Renames a job queue file.
+
+ """
+ if not (_IsJobQueueFile(old) and _IsJobQueueFile(new)):
+ return False
+
+ os.rename(old, new)
+
+ return True
+
+
+def CloseBlockDevices(disks):
+ """Closes the given block devices.
+
+ This means they will be switched to secondary mode (in case of DRBD).
+
+ """
+ bdevs = []
+ for cf in disks:
+ rd = _RecursiveFindBD(cf)
+ if rd is None:
+ return (False, "Can't find device %s" % cf)
+ bdevs.append(rd)
+
+ msg = []
+ for rd in bdevs:
+ try:
+ rd.Close()
+ except errors.BlockDeviceError, err:
+ msg.append(str(err))
+ if msg:
+ return (False, "Can't make devices secondary: %s" % ",".join(msg))
+ else:
+ return (True, "All devices secondary")
+
+
class HooksRunner(object):
"""Hook runner.
Args:
- hooks_base_dir: if not None, this overrides the
constants.HOOKS_BASE_DIR (useful for unittests)
- - logs_base_dir: if not None, this overrides the
- constants.LOG_HOOKS_DIR (useful for unittests)
- - logging: enable or disable logging of script output
"""
if hooks_base_dir is None:
"""Exec one hook script.
Args:
- - phase: the phase
- script: the full path to the script
- env: the environment with which to exec the script
fdstdin = open("/dev/null", "r")
child = subprocess.Popen([script], stdin=fdstdin, stdout=subprocess.PIPE,
stderr=subprocess.STDOUT, close_fds=True,
- shell=False, cwd="/",env=env)
+ shell=False, cwd="/", env=env)
output = ""
try:
output = child.stdout.read(4096)
fd.close()
except EnvironmentError, err:
# just log the error
- #logger.Error("While closing fd %s: %s" % (fd, err))
+ #logging.exception("Error while closing fd %s", fd)
pass
return result == 0, output
return rr
+class IAllocatorRunner(object):
+ """IAllocator runner.
+
+ This class is instantiated on the node side (ganeti-noded) and not on
+ the master side.
+
+ """
+ def Run(self, name, idata):
+ """Run an iallocator script.
+
+ Return value: tuple of:
+ - run status (one of the IARUN_ constants)
+ - stdout
+ - stderr
+ - fail reason (as from utils.RunResult)
+
+ """
+ alloc_script = utils.FindFile(name, constants.IALLOCATOR_SEARCH_PATH,
+ os.path.isfile)
+ if alloc_script is None:
+ return (constants.IARUN_NOTFOUND, None, None, None)
+
+ fd, fin_name = tempfile.mkstemp(prefix="ganeti-iallocator.")
+ try:
+ os.write(fd, idata)
+ os.close(fd)
+ result = utils.RunCmd([alloc_script, fin_name])
+ if result.failed:
+ return (constants.IARUN_FAILURE, result.stdout, result.stderr,
+ result.fail_reason)
+ finally:
+ os.unlink(fin_name)
+
+ return (constants.IARUN_SUCCESS, result.stdout, result.stderr, None)
+
+
class DevCacheManager(object):
- """Simple class for managing a chache of block device information.
+ """Simple class for managing a cache of block device information.
"""
_DEV_PREFIX = "/dev/"
"""
if dev_path is None:
- logger.Error("DevCacheManager.UpdateCache got a None dev_path")
+ logging.error("DevCacheManager.UpdateCache got a None dev_path")
return
fpath = cls._ConvertPath(dev_path)
if on_primary:
try:
utils.WriteFile(fpath, data=fdata)
except EnvironmentError, err:
- logger.Error("Can't update bdev cache for %s, error %s" %
- (dev_path, str(err)))
+ logging.exception("Can't update bdev cache for %s", dev_path)
@classmethod
def RemoveCache(cls, dev_path):
"""
if dev_path is None:
- logger.Error("DevCacheManager.RemoveCache got a None dev_path")
+ logging.error("DevCacheManager.RemoveCache got a None dev_path")
return
fpath = cls._ConvertPath(dev_path)
try:
utils.RemoveFile(fpath)
except EnvironmentError, err:
- logger.Error("Can't update bdev cache for %s, error %s" %
- (dev_path, str(err)))
+ logging.exception("Can't update bdev cache for %s", dev_path)