Merge branch 'devel-2.7'

[ganeti-local] / lib / bootstrap.py

diff --git a/lib/bootstrap.py b/lib/bootstrap.py
index a162897..69ba1ef 100644 (file)
--- a/lib/bootstrap.py
+++ b/lib/bootstrap.py
@@ -1,7 +1,7 @@
 #
 #
 
 #
 #
 

-# Copyright (C) 2006, 2007, 2008 Google Inc.
+# Copyright (C) 2006, 2007, 2008, 2010, 2011, 2012 Google Inc.

 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by
 #
 # This program is free software; you can redistribute it and/or modify
 # it under the terms of the GNU General Public License as published by

@@ -27,8 +27,8 @@ import os
 import os.path
 import re
 import logging
 import os.path
 import re
 import logging

-import tempfile

 import time
 import time

+import tempfile

 
 from ganeti import rpc
 from ganeti import ssh
 
 from ganeti import rpc
 from ganeti import ssh

@@ -40,6 +40,18 @@ from ganeti import objects
 from ganeti import ssconf
 from ganeti import serializer
 from ganeti import hypervisor
 from ganeti import ssconf
 from ganeti import serializer
 from ganeti import hypervisor

+from ganeti import bdev
+from ganeti import netutils
+from ganeti import luxi
+from ganeti import jstore
+from ganeti import pathutils
+
+
+# ec_id for InitConfig's temporary reservation manager
+_INITCONF_ECID = "initconfig-ecid"
+
+#: After how many seconds daemon must be responsive
+_DAEMON_READY_TIMEOUT = 10.0

 
 
 def _InitSSHSetup():
 
 
 def _InitSSHSetup():

@@ -49,7 +61,7 @@ def _InitSSHSetup():
   permitted hosts and adds the hostkey to its own known hosts.
 
   """
   permitted hosts and adds the hostkey to its own known hosts.
 
   """

-  priv_key, pub_key, auth_keys = ssh.GetUserFiles(constants.GANETI_RUNAS)
+  priv_key, pub_key, auth_keys = ssh.GetUserFiles(constants.SSH_LOGIN_USER)

 
   for name in priv_key, pub_key:
     if os.path.exists(name):
 
   for name in priv_key, pub_key:
     if os.path.exists(name):

@@ -66,40 +78,6 @@ def _InitSSHSetup():
   utils.AddAuthorizedKey(auth_keys, utils.ReadFile(pub_key))
 
 
   utils.AddAuthorizedKey(auth_keys, utils.ReadFile(pub_key))
 
 

-def GenerateSelfSignedSslCert(file_name, validity=(365 * 5)):
-  """Generates a self-signed SSL certificate.
-
-  @type file_name: str
-  @param file_name: Path to output file
-  @type validity: int
-  @param validity: Validity for certificate in days
-
-  """
-  (fd, tmp_file_name) = tempfile.mkstemp(dir=os.path.dirname(file_name))
-  try:
-    try:
-      # Set permissions before writing key
-      os.chmod(tmp_file_name, 0600)
-
-      result = utils.RunCmd(["openssl", "req", "-new", "-newkey", "rsa:1024",
-                             "-days", str(validity), "-nodes", "-x509",
-                             "-keyout", tmp_file_name, "-out", tmp_file_name,
-                             "-batch"])
-      if result.failed:
-        raise errors.OpExecError("Could not generate SSL certificate, command"
-                                 " %s had exitcode %s and error message %s" %
-                                 (result.cmd, result.exit_code, result.output))
-
-      # Make read-only
-      os.chmod(tmp_file_name, 0400)
-
-      os.rename(tmp_file_name, file_name)
-    finally:
-      utils.RemoveFile(tmp_file_name)
-  finally:
-    os.close(fd)
-
-

 def GenerateHmacKey(file_name):
   """Writes a new HMAC key.
 
 def GenerateHmacKey(file_name):
   """Writes a new HMAC key.
 

@@ -107,48 +85,271 @@ def GenerateHmacKey(file_name):
   @param file_name: Path to output file
 
   """
   @param file_name: Path to output file
 
   """

-  utils.WriteFile(file_name, data="%s\n" % utils.GenerateSecret(), mode=0400)
+  utils.WriteFile(file_name, data="%s\n" % utils.GenerateSecret(), mode=0400,
+                  backup=True)
+
+
+def GenerateClusterCrypto(new_cluster_cert, new_rapi_cert, new_spice_cert,
+                          new_confd_hmac_key, new_cds,
+                          rapi_cert_pem=None, spice_cert_pem=None,
+                          spice_cacert_pem=None, cds=None,
+                          nodecert_file=pathutils.NODED_CERT_FILE,
+                          rapicert_file=pathutils.RAPI_CERT_FILE,
+                          spicecert_file=pathutils.SPICE_CERT_FILE,
+                          spicecacert_file=pathutils.SPICE_CACERT_FILE,
+                          hmackey_file=pathutils.CONFD_HMAC_KEY,
+                          cds_file=pathutils.CLUSTER_DOMAIN_SECRET_FILE):
+  """Updates the cluster certificates, keys and secrets.
+
+  @type new_cluster_cert: bool
+  @param new_cluster_cert: Whether to generate a new cluster certificate
+  @type new_rapi_cert: bool
+  @param new_rapi_cert: Whether to generate a new RAPI certificate
+  @type new_spice_cert: bool
+  @param new_spice_cert: Whether to generate a new SPICE certificate
+  @type new_confd_hmac_key: bool
+  @param new_confd_hmac_key: Whether to generate a new HMAC key
+  @type new_cds: bool
+  @param new_cds: Whether to generate a new cluster domain secret
+  @type rapi_cert_pem: string
+  @param rapi_cert_pem: New RAPI certificate in PEM format
+  @type spice_cert_pem: string
+  @param spice_cert_pem: New SPICE certificate in PEM format
+  @type spice_cacert_pem: string
+  @param spice_cacert_pem: Certificate of the CA that signed the SPICE
+                           certificate, in PEM format
+  @type cds: string
+  @param cds: New cluster domain secret
+  @type nodecert_file: string
+  @param nodecert_file: optional override of the node cert file path
+  @type rapicert_file: string
+  @param rapicert_file: optional override of the rapi cert file path
+  @type spicecert_file: string
+  @param spicecert_file: optional override of the spice cert file path
+  @type spicecacert_file: string
+  @param spicecacert_file: optional override of the spice CA cert file path
+  @type hmackey_file: string
+  @param hmackey_file: optional override of the hmac key file path
+
+  """
+  # noded SSL certificate
+  cluster_cert_exists = os.path.exists(nodecert_file)
+  if new_cluster_cert or not cluster_cert_exists:
+    if cluster_cert_exists:
+      utils.CreateBackup(nodecert_file)
+
+    logging.debug("Generating new cluster certificate at %s", nodecert_file)
+    utils.GenerateSelfSignedSslCert(nodecert_file)
+
+  # confd HMAC key
+  if new_confd_hmac_key or not os.path.exists(hmackey_file):
+    logging.debug("Writing new confd HMAC key to %s", hmackey_file)
+    GenerateHmacKey(hmackey_file)
+
+  # RAPI
+  rapi_cert_exists = os.path.exists(rapicert_file)
+
+  if rapi_cert_pem:
+    # Assume rapi_pem contains a valid PEM-formatted certificate and key
+    logging.debug("Writing RAPI certificate at %s", rapicert_file)
+    utils.WriteFile(rapicert_file, data=rapi_cert_pem, backup=True)
+
+  elif new_rapi_cert or not rapi_cert_exists:
+    if rapi_cert_exists:
+      utils.CreateBackup(rapicert_file)
+
+    logging.debug("Generating new RAPI certificate at %s", rapicert_file)
+    utils.GenerateSelfSignedSslCert(rapicert_file)
+
+  # SPICE
+  spice_cert_exists = os.path.exists(spicecert_file)
+  spice_cacert_exists = os.path.exists(spicecacert_file)
+  if spice_cert_pem:
+    # spice_cert_pem implies also spice_cacert_pem
+    logging.debug("Writing SPICE certificate at %s", spicecert_file)
+    utils.WriteFile(spicecert_file, data=spice_cert_pem, backup=True)
+    logging.debug("Writing SPICE CA certificate at %s", spicecacert_file)
+    utils.WriteFile(spicecacert_file, data=spice_cacert_pem, backup=True)
+  elif new_spice_cert or not spice_cert_exists:
+    if spice_cert_exists:
+      utils.CreateBackup(spicecert_file)
+    if spice_cacert_exists:
+      utils.CreateBackup(spicecacert_file)
+
+    logging.debug("Generating new self-signed SPICE certificate at %s",
+                  spicecert_file)
+    (_, cert_pem) = utils.GenerateSelfSignedSslCert(spicecert_file)
+
+    # Self-signed certificate -> the public certificate is also the CA public
+    # certificate
+    logging.debug("Writing the public certificate to %s",
+                  spicecert_file)
+    utils.io.WriteFile(spicecacert_file, mode=0400, data=cert_pem)
+
+  # Cluster domain secret
+  if cds:
+    logging.debug("Writing cluster domain secret to %s", cds_file)
+    utils.WriteFile(cds_file, data=cds, backup=True)
+
+  elif new_cds or not os.path.exists(cds_file):
+    logging.debug("Generating new cluster domain secret at %s", cds_file)
+    GenerateHmacKey(cds_file)

 
 
 def _InitGanetiServerSetup(master_name):
   """Setup the necessary configuration for the initial node daemon.
 
   This creates the nodepass file containing the shared password for
 
 
 def _InitGanetiServerSetup(master_name):
   """Setup the necessary configuration for the initial node daemon.
 
   This creates the nodepass file containing the shared password for

-  the cluster and also generates the SSL certificate.
+  the cluster, generates the SSL certificate and starts the node daemon.

 
 

-  """
-  GenerateSelfSignedSslCert(constants.SSL_CERT_FILE)
+  @type master_name: str
+  @param master_name: Name of the master node

 
 

-  # Don't overwrite existing file
-  if not os.path.exists(constants.RAPI_CERT_FILE):
-    GenerateSelfSignedSslCert(constants.RAPI_CERT_FILE)
-
-  if not os.path.exists(constants.HMAC_CLUSTER_KEY):
-    GenerateHmacKey(constants.HMAC_CLUSTER_KEY)
+  """
+  # Generate cluster secrets
+  GenerateClusterCrypto(True, False, False, False, False)

 
 

-  result = utils.RunCmd([constants.DAEMON_UTIL, "start", constants.NODED])
+  result = utils.RunCmd([pathutils.DAEMON_UTIL, "start", constants.NODED])

   if result.failed:
     raise errors.OpExecError("Could not start the node daemon, command %s"
                              " had exitcode %s and error %s" %
                              (result.cmd, result.exit_code, result.output))
 
   if result.failed:
     raise errors.OpExecError("Could not start the node daemon, command %s"
                              " had exitcode %s and error %s" %
                              (result.cmd, result.exit_code, result.output))
 

-  # Wait for node daemon to become responsive
+  _WaitForNodeDaemon(master_name)
+
+
+def _WaitForNodeDaemon(node_name):
+  """Wait for node daemon to become responsive.
+
+  """

   def _CheckNodeDaemon():
   def _CheckNodeDaemon():

-    result = rpc.RpcRunner.call_version([master_name])[master_name]
+    # Pylint bug <http://www.logilab.org/ticket/35642>
+    # pylint: disable=E1101
+    result = rpc.BootstrapRunner().call_version([node_name])[node_name]

     if result.fail_msg:
       raise utils.RetryAgain()
 
   try:
     if result.fail_msg:
       raise utils.RetryAgain()
 
   try:

-    utils.Retry(_CheckNodeDaemon, 1.0, 10.0)
+    utils.Retry(_CheckNodeDaemon, 1.0, _DAEMON_READY_TIMEOUT)

   except utils.RetryTimeout:
   except utils.RetryTimeout:

-    raise errors.OpExecError("Node daemon didn't answer queries within"
-                             " 10 seconds")
-
-def InitCluster(cluster_name, mac_prefix,
-                master_netdev, file_storage_dir, candidate_pool_size,
-                secondary_ip=None, vg_name=None, beparams=None,
-                nicparams=None, hvparams=None, enabled_hypervisors=None,
-                modify_etc_hosts=True, modify_ssh_setup=True):
+    raise errors.OpExecError("Node daemon on %s didn't answer queries within"
+                             " %s seconds" % (node_name, _DAEMON_READY_TIMEOUT))
+
+
+def _WaitForMasterDaemon():
+  """Wait for master daemon to become responsive.
+
+  """
+  def _CheckMasterDaemon():
+    try:
+      cl = luxi.Client()
+      (cluster_name, ) = cl.QueryConfigValues(["cluster_name"])
+    except Exception:
+      raise utils.RetryAgain()
+
+    logging.debug("Received cluster name %s from master", cluster_name)
+
+  try:
+    utils.Retry(_CheckMasterDaemon, 1.0, _DAEMON_READY_TIMEOUT)
+  except utils.RetryTimeout:
+    raise errors.OpExecError("Master daemon didn't answer queries within"
+                             " %s seconds" % _DAEMON_READY_TIMEOUT)
+
+
+def RunNodeSetupCmd(cluster_name, node, basecmd, debug, verbose,
+                    use_cluster_key, ask_key, strict_host_check, data):
+  """Runs a command to configure something on a remote machine.
+
+  @type cluster_name: string
+  @param cluster_name: Cluster name
+  @type node: string
+  @param node: Node name
+  @type basecmd: string
+  @param basecmd: Base command (path on the remote machine)
+  @type debug: bool
+  @param debug: Enable debug output
+  @type verbose: bool
+  @param verbose: Enable verbose output
+  @type use_cluster_key: bool
+  @param use_cluster_key: See L{ssh.SshRunner.BuildCmd}
+  @type ask_key: bool
+  @param ask_key: See L{ssh.SshRunner.BuildCmd}
+  @type strict_host_check: bool
+  @param strict_host_check: See L{ssh.SshRunner.BuildCmd}
+  @param data: JSON-serializable input data for script (passed to stdin)
+
+  """
+  cmd = [basecmd]
+
+  # Pass --debug/--verbose to the external script if set on our invocation
+  if debug:
+    cmd.append("--debug")
+
+  if verbose:
+    cmd.append("--verbose")
+
+  family = ssconf.SimpleStore().GetPrimaryIPFamily()
+  srun = ssh.SshRunner(cluster_name,
+                       ipv6=(family == netutils.IP6Address.family))
+  scmd = srun.BuildCmd(node, constants.SSH_LOGIN_USER,
+                       utils.ShellQuoteArgs(cmd),
+                       batch=False, ask_key=ask_key, quiet=False,
+                       strict_host_check=strict_host_check,
+                       use_cluster_key=use_cluster_key)
+
+  tempfh = tempfile.TemporaryFile()
+  try:
+    tempfh.write(serializer.DumpJson(data))
+    tempfh.seek(0)
+
+    result = utils.RunCmd(scmd, interactive=True, input_fd=tempfh)
+  finally:
+    tempfh.close()
+
+  if result.failed:
+    raise errors.OpExecError("Command '%s' failed: %s" %
+                             (result.cmd, result.fail_reason))
+
+
+def _InitFileStorage(file_storage_dir):
+  """Initialize if needed the file storage.
+
+  @param file_storage_dir: the user-supplied value
+  @return: either empty string (if file storage was disabled at build
+      time) or the normalized path to the storage directory
+
+  """
+  file_storage_dir = os.path.normpath(file_storage_dir)
+
+  if not os.path.isabs(file_storage_dir):
+    raise errors.OpPrereqError("File storage directory '%s' is not an absolute"
+                               " path" % file_storage_dir, errors.ECODE_INVAL)
+
+  if not os.path.exists(file_storage_dir):
+    try:
+      os.makedirs(file_storage_dir, 0750)
+    except OSError, err:
+      raise errors.OpPrereqError("Cannot create file storage directory"
+                                 " '%s': %s" % (file_storage_dir, err),
+                                 errors.ECODE_ENVIRON)
+
+  if not os.path.isdir(file_storage_dir):
+    raise errors.OpPrereqError("The file storage directory '%s' is not"
+                               " a directory." % file_storage_dir,
+                               errors.ECODE_ENVIRON)
+  return file_storage_dir
+
+
+def InitCluster(cluster_name, mac_prefix, # pylint: disable=R0913, R0914
+                master_netmask, master_netdev, file_storage_dir,
+                shared_file_storage_dir, candidate_pool_size, secondary_ip=None,
+                vg_name=None, beparams=None, nicparams=None, ndparams=None,
+                hvparams=None, diskparams=None, enabled_hypervisors=None,
+                modify_etc_hosts=True, modify_ssh_setup=True,
+                maintain_node_health=False, drbd_helper=None, uid_pool=None,
+                default_iallocator=None, primary_ip_version=None, ipolicy=None,
+                prealloc_wipe_disks=False, use_external_mip_script=False,
+                hv_state=None, disk_state=None):

   """Initialise the cluster.
 
   @type candidate_pool_size: int
   """Initialise the cluster.
 
   @type candidate_pool_size: int

@@ -169,39 +370,61 @@ def InitCluster(cluster_name, mac_prefix,
                                " entries: %s" % invalid_hvs,
                                errors.ECODE_INVAL)
 
                                " entries: %s" % invalid_hvs,
                                errors.ECODE_INVAL)
 

-  hostname = utils.GetHostInfo()
+  try:
+    ipcls = netutils.IPAddress.GetClassFromIpVersion(primary_ip_version)
+  except errors.ProgrammerError:
+    raise errors.OpPrereqError("Invalid primary ip version: %d." %
+                               primary_ip_version, errors.ECODE_INVAL)
+
+  hostname = netutils.GetHostname(family=ipcls.family)
+  if not ipcls.IsValid(hostname.ip):
+    raise errors.OpPrereqError("This host's IP (%s) is not a valid IPv%d"
+                               " address." % (hostname.ip, primary_ip_version),
+                               errors.ECODE_INVAL)

 
 

-  if hostname.ip.startswith("127."):
-    raise errors.OpPrereqError("This host's IP resolves to the private"
-                               " range (%s). Please fix DNS or %s." %
-                               (hostname.ip, constants.ETC_HOSTS),
+  if ipcls.IsLoopback(hostname.ip):
+    raise errors.OpPrereqError("This host's IP (%s) resolves to a loopback"
+                               " address. Please fix DNS or %s." %
+                               (hostname.ip, pathutils.ETC_HOSTS),

                                errors.ECODE_ENVIRON)
 
                                errors.ECODE_ENVIRON)
 

-  if not utils.OwnIpAddress(hostname.ip):
+  if not ipcls.Own(hostname.ip):

     raise errors.OpPrereqError("Inconsistency: this host's name resolves"
                                " to %s,\nbut this ip address does not"
     raise errors.OpPrereqError("Inconsistency: this host's name resolves"
                                " to %s,\nbut this ip address does not"

-                               " belong to this host. Aborting." %
+                               " belong to this host" %

                                hostname.ip, errors.ECODE_ENVIRON)
 
                                hostname.ip, errors.ECODE_ENVIRON)
 

-  clustername = utils.GetHostInfo(cluster_name)
+  clustername = netutils.GetHostname(name=cluster_name, family=ipcls.family)

 
 

-  if utils.TcpPing(clustername.ip, constants.DEFAULT_NODED_PORT,
-                   timeout=5):
-    raise errors.OpPrereqError("Cluster IP already active. Aborting.",
+  if netutils.TcpPing(clustername.ip, constants.DEFAULT_NODED_PORT, timeout=5):
+    raise errors.OpPrereqError("Cluster IP already active",

                                errors.ECODE_NOTUNIQUE)
 
                                errors.ECODE_NOTUNIQUE)
 

-  if secondary_ip:
-    if not utils.IsValidIP(secondary_ip):
-      raise errors.OpPrereqError("Invalid secondary ip given",
+  if not secondary_ip:
+    if primary_ip_version == constants.IP6_VERSION:
+      raise errors.OpPrereqError("When using a IPv6 primary address, a valid"
+                                 " IPv4 address must be given as secondary",

                                  errors.ECODE_INVAL)
                                  errors.ECODE_INVAL)

-    if (secondary_ip != hostname.ip and
-        not utils.OwnIpAddress(secondary_ip)):
-      raise errors.OpPrereqError("You gave %s as secondary IP,"
-                                 " but it does not belong to this host." %
-                                 secondary_ip, errors.ECODE_ENVIRON)
-  else:

     secondary_ip = hostname.ip
 
     secondary_ip = hostname.ip
 

+  if not netutils.IP4Address.IsValid(secondary_ip):
+    raise errors.OpPrereqError("Secondary IP address (%s) has to be a valid"
+                               " IPv4 address." % secondary_ip,
+                               errors.ECODE_INVAL)
+
+  if not netutils.IP4Address.Own(secondary_ip):
+    raise errors.OpPrereqError("You gave %s as secondary IP,"
+                               " but it does not belong to this host." %
+                               secondary_ip, errors.ECODE_ENVIRON)
+
+  if master_netmask is not None:
+    if not ipcls.ValidateNetmask(master_netmask):
+      raise errors.OpPrereqError("CIDR netmask (%s) not valid for IPv%s " %
+                                  (master_netmask, primary_ip_version),
+                                 errors.ECODE_INVAL)
+  else:
+    master_netmask = ipcls.iplen
+

   if vg_name is not None:
     # Check if volume group is valid
     vgstatus = utils.CheckVolumeGroupSize(utils.ListVolumeGroups(), vg_name,
   if vg_name is not None:
     # Check if volume group is valid
     vgstatus = utils.CheckVolumeGroupSize(utils.ListVolumeGroups(), vg_name,

@@ -211,24 +434,36 @@ def InitCluster(cluster_name, mac_prefix,
                                  " you are not using lvm" % vgstatus,
                                  errors.ECODE_INVAL)
 
                                  " you are not using lvm" % vgstatus,
                                  errors.ECODE_INVAL)
 

-  file_storage_dir = os.path.normpath(file_storage_dir)
-
-  if not os.path.isabs(file_storage_dir):
-    raise errors.OpPrereqError("The file storage directory you passed is"
-                               " not an absolute path.", errors.ECODE_INVAL)
-
-  if not os.path.exists(file_storage_dir):
+  if drbd_helper is not None:

     try:
     try:

-      os.makedirs(file_storage_dir, 0750)
-    except OSError, err:
-      raise errors.OpPrereqError("Cannot create file storage directory"
-                                 " '%s': %s" % (file_storage_dir, err),
+      curr_helper = bdev.BaseDRBD.GetUsermodeHelper()
+    except errors.BlockDeviceError, err:
+      raise errors.OpPrereqError("Error while checking drbd helper"
+                                 " (specify --no-drbd-storage if you are not"
+                                 " using drbd): %s" % str(err),

                                  errors.ECODE_ENVIRON)
                                  errors.ECODE_ENVIRON)

+    if drbd_helper != curr_helper:
+      raise errors.OpPrereqError("Error: requiring %s as drbd helper but %s"
+                                 " is the current helper" % (drbd_helper,
+                                                             curr_helper),
+                                 errors.ECODE_INVAL)

 
 

-  if not os.path.isdir(file_storage_dir):
-    raise errors.OpPrereqError("The file storage directory '%s' is not"
-                               " a directory." % file_storage_dir,
-                               errors.ECODE_ENVIRON)
+  logging.debug("Stopping daemons (if any are running)")
+  result = utils.RunCmd([pathutils.DAEMON_UTIL, "stop-all"])
+  if result.failed:
+    raise errors.OpExecError("Could not stop daemons, command %s"
+                             " had exitcode %s and error '%s'" %
+                             (result.cmd, result.exit_code, result.output))
+
+  if constants.ENABLE_FILE_STORAGE:
+    file_storage_dir = _InitFileStorage(file_storage_dir)
+  else:
+    file_storage_dir = ""
+
+  if constants.ENABLE_SHARED_FILE_STORAGE:
+    shared_file_storage_dir = _InitFileStorage(shared_file_storage_dir)
+  else:
+    shared_file_storage_dir = ""

 
   if not re.match("^[0-9a-z]{2}:[0-9a-z]{2}:[0-9a-z]{2}$", mac_prefix):
     raise errors.OpPrereqError("Invalid mac prefix given '%s'" % mac_prefix,
 
   if not re.match("^[0-9a-z]{2}:[0-9a-z]{2}:[0-9a-z]{2}$", mac_prefix):
     raise errors.OpPrereqError("Invalid mac prefix given '%s'" % mac_prefix,

@@ -240,32 +475,91 @@ def InitCluster(cluster_name, mac_prefix,
                                (master_netdev,
                                 result.output.strip()), errors.ECODE_INVAL)
 
                                (master_netdev,
                                 result.output.strip()), errors.ECODE_INVAL)
 

-  dirs = [(constants.RUN_GANETI_DIR, constants.RUN_DIRS_MODE)]
+  dirs = [(pathutils.RUN_DIR, constants.RUN_DIRS_MODE)]

   utils.EnsureDirs(dirs)
 
   utils.EnsureDirs(dirs)
 

+  objects.UpgradeBeParams(beparams)

   utils.ForceDictType(beparams, constants.BES_PARAMETER_TYPES)
   utils.ForceDictType(nicparams, constants.NICS_PARAMETER_TYPES)
   utils.ForceDictType(beparams, constants.BES_PARAMETER_TYPES)
   utils.ForceDictType(nicparams, constants.NICS_PARAMETER_TYPES)

+

   objects.NIC.CheckParameterSyntax(nicparams)
 
   objects.NIC.CheckParameterSyntax(nicparams)
 

+  full_ipolicy = objects.FillIPolicy(constants.IPOLICY_DEFAULTS, ipolicy)
+
+  if ndparams is not None:
+    utils.ForceDictType(ndparams, constants.NDS_PARAMETER_TYPES)
+  else:
+    ndparams = dict(constants.NDC_DEFAULTS)
+
+  # This is ugly, as we modify the dict itself
+  # FIXME: Make utils.ForceDictType pure functional or write a wrapper
+  # around it
+  if hv_state:
+    for hvname, hvs_data in hv_state.items():
+      utils.ForceDictType(hvs_data, constants.HVSTS_PARAMETER_TYPES)
+      hv_state[hvname] = objects.Cluster.SimpleFillHvState(hvs_data)
+  else:
+    hv_state = dict((hvname, constants.HVST_DEFAULTS)
+                    for hvname in enabled_hypervisors)
+
+  # FIXME: disk_state has no default values yet
+  if disk_state:
+    for storage, ds_data in disk_state.items():
+      if storage not in constants.DS_VALID_TYPES:
+        raise errors.OpPrereqError("Invalid storage type in disk state: %s" %
+                                   storage, errors.ECODE_INVAL)
+      for ds_name, state in ds_data.items():
+        utils.ForceDictType(state, constants.DSS_PARAMETER_TYPES)
+        ds_data[ds_name] = objects.Cluster.SimpleFillDiskState(state)
+

   # hvparams is a mapping of hypervisor->hvparams dict
   for hv_name, hv_params in hvparams.iteritems():
     utils.ForceDictType(hv_params, constants.HVS_PARAMETER_TYPES)
     hv_class = hypervisor.GetHypervisor(hv_name)
     hv_class.CheckParameterSyntax(hv_params)
 
   # hvparams is a mapping of hypervisor->hvparams dict
   for hv_name, hv_params in hvparams.iteritems():
     utils.ForceDictType(hv_params, constants.HVS_PARAMETER_TYPES)
     hv_class = hypervisor.GetHypervisor(hv_name)
     hv_class.CheckParameterSyntax(hv_params)
 

-  # set up the inter-node password and certificate
-  _InitGanetiServerSetup(hostname.name)
+  # diskparams is a mapping of disk-template->diskparams dict
+  for template, dt_params in diskparams.items():
+    param_keys = set(dt_params.keys())
+    default_param_keys = set(constants.DISK_DT_DEFAULTS[template].keys())
+    if not (param_keys <= default_param_keys):
+      unknown_params = param_keys - default_param_keys
+      raise errors.OpPrereqError("Invalid parameters for disk template %s:"
+                                 " %s" % (template,
+                                          utils.CommaJoin(unknown_params)),
+                                 errors.ECODE_INVAL)
+    utils.ForceDictType(dt_params, constants.DISK_DT_TYPES)
+  try:
+    utils.VerifyDictOptions(diskparams, constants.DISK_DT_DEFAULTS)
+  except errors.OpPrereqError, err:
+    raise errors.OpPrereqError("While verify diskparam options: %s" % err,
+                               errors.ECODE_INVAL)

 
   # set up ssh config and /etc/hosts
 
   # set up ssh config and /etc/hosts

-  sshline = utils.ReadFile(constants.SSH_HOST_RSA_PUB)
+  sshline = utils.ReadFile(pathutils.SSH_HOST_RSA_PUB)

   sshkey = sshline.split(" ")[1]
 
   if modify_etc_hosts:
   sshkey = sshline.split(" ")[1]
 
   if modify_etc_hosts:

-    utils.AddHostToEtcHosts(hostname.name)
+    utils.AddHostToEtcHosts(hostname.name, hostname.ip)

 
   if modify_ssh_setup:
     _InitSSHSetup()
 
 
   if modify_ssh_setup:
     _InitSSHSetup()
 

+  if default_iallocator is not None:
+    alloc_script = utils.FindFile(default_iallocator,
+                                  constants.IALLOCATOR_SEARCH_PATH,
+                                  os.path.isfile)
+    if alloc_script is None:
+      raise errors.OpPrereqError("Invalid default iallocator script '%s'"
+                                 " specified" % default_iallocator,
+                                 errors.ECODE_INVAL)
+  elif constants.HTOOLS:
+    # htools was enabled at build-time, we default to it
+    if utils.FindFile(constants.IALLOC_HAIL,
+                      constants.IALLOCATOR_SEARCH_PATH,
+                      os.path.isfile):
+      default_iallocator = constants.IALLOC_HAIL
+

   now = time.time()
 
   # init of cluster config file
   now = time.time()
 
   # init of cluster config file

@@ -278,19 +572,32 @@ def InitCluster(cluster_name, mac_prefix,
     tcpudp_port_pool=set(),
     master_node=hostname.name,
     master_ip=clustername.ip,
     tcpudp_port_pool=set(),
     master_node=hostname.name,
     master_ip=clustername.ip,

+    master_netmask=master_netmask,

     master_netdev=master_netdev,
     cluster_name=clustername.name,
     file_storage_dir=file_storage_dir,
     master_netdev=master_netdev,
     cluster_name=clustername.name,
     file_storage_dir=file_storage_dir,

+    shared_file_storage_dir=shared_file_storage_dir,

     enabled_hypervisors=enabled_hypervisors,
     beparams={constants.PP_DEFAULT: beparams},
     nicparams={constants.PP_DEFAULT: nicparams},
     enabled_hypervisors=enabled_hypervisors,
     beparams={constants.PP_DEFAULT: beparams},
     nicparams={constants.PP_DEFAULT: nicparams},

+    ndparams=ndparams,

     hvparams=hvparams,
     hvparams=hvparams,

+    diskparams=diskparams,

     candidate_pool_size=candidate_pool_size,
     modify_etc_hosts=modify_etc_hosts,
     modify_ssh_setup=modify_ssh_setup,
     candidate_pool_size=candidate_pool_size,
     modify_etc_hosts=modify_etc_hosts,
     modify_ssh_setup=modify_ssh_setup,

+    uid_pool=uid_pool,

     ctime=now,
     mtime=now,
     ctime=now,
     mtime=now,

-    uuid=utils.NewUUID(),
+    maintain_node_health=maintain_node_health,
+    drbd_usermode_helper=drbd_helper,
+    default_iallocator=default_iallocator,
+    primary_ip_family=ipcls.family,
+    prealloc_wipe_disks=prealloc_wipe_disks,
+    use_external_mip_script=use_external_mip_script,
+    ipolicy=full_ipolicy,
+    hv_state_static=hv_state,
+    disk_state_static=disk_state,

     )
   master_node_config = objects.Node(name=hostname.name,
                                     primary_ip=hostname.ip,
     )
   master_node_config = objects.Node(name=hostname.name,
                                     primary_ip=hostname.ip,

@@ -298,20 +605,29 @@ def InitCluster(cluster_name, mac_prefix,
                                     serial_no=1,
                                     master_candidate=True,
                                     offline=False, drained=False,
                                     serial_no=1,
                                     master_candidate=True,
                                     offline=False, drained=False,

+                                    ctime=now, mtime=now,

                                     )
   InitConfig(constants.CONFIG_VERSION, cluster_config, master_node_config)
                                     )
   InitConfig(constants.CONFIG_VERSION, cluster_config, master_node_config)

-  cfg = config.ConfigWriter()
-  ssh.WriteKnownHostsFile(cfg, constants.SSH_KNOWN_HOSTS_FILE)
+  cfg = config.ConfigWriter(offline=True)
+  ssh.WriteKnownHostsFile(cfg, pathutils.SSH_KNOWN_HOSTS_FILE)

   cfg.Update(cfg.GetClusterInfo(), logging.error)
   cfg.Update(cfg.GetClusterInfo(), logging.error)

+  ssconf.WriteSsconfFiles(cfg.GetSsconfValues())
+
+  # set up the inter-node password and certificate
+  _InitGanetiServerSetup(hostname.name)
+
+  logging.debug("Starting daemons")
+  result = utils.RunCmd([pathutils.DAEMON_UTIL, "start-all"])
+  if result.failed:
+    raise errors.OpExecError("Could not start daemons, command %s"
+                             " had exitcode %s and error %s" %
+                             (result.cmd, result.exit_code, result.output))

 
 

-  # start the master ip
-  # TODO: Review rpc call from bootstrap
-  # TODO: Warn on failed start master
-  rpc.RpcRunner.call_node_start_master(hostname.name, True, False)
+  _WaitForMasterDaemon()

 
 
 def InitConfig(version, cluster_config, master_node_config,
 
 
 def InitConfig(version, cluster_config, master_node_config,

-               cfg_file=constants.CLUSTER_CONF_FILE):
+               cfg_file=pathutils.CLUSTER_CONF_FILE):

   """Create the initial cluster configuration.
 
   It will contain the current node, which will also be the master
   """Create the initial cluster configuration.
 
   It will contain the current node, which will also be the master

@@ -327,15 +643,30 @@ def InitConfig(version, cluster_config, master_node_config,
   @param cfg_file: configuration file path
 
   """
   @param cfg_file: configuration file path
 
   """

+  uuid_generator = config.TemporaryReservationManager()
+  cluster_config.uuid = uuid_generator.Generate([], utils.NewUUID,
+                                                _INITCONF_ECID)
+  master_node_config.uuid = uuid_generator.Generate([], utils.NewUUID,
+                                                    _INITCONF_ECID)

   nodes = {
     master_node_config.name: master_node_config,
     }
   nodes = {
     master_node_config.name: master_node_config,
     }

-
+  default_nodegroup = objects.NodeGroup(
+    uuid=uuid_generator.Generate([], utils.NewUUID, _INITCONF_ECID),
+    name=constants.INITIAL_NODE_GROUP_NAME,
+    members=[master_node_config.name],
+    diskparams={},
+    )
+  nodegroups = {
+    default_nodegroup.uuid: default_nodegroup,
+    }

   now = time.time()
   config_data = objects.ConfigData(version=version,
                                    cluster=cluster_config,
   now = time.time()
   config_data = objects.ConfigData(version=version,
                                    cluster=cluster_config,

+                                   nodegroups=nodegroups,

                                    nodes=nodes,
                                    instances={},
                                    nodes=nodes,
                                    instances={},

+                                   networks={},

                                    serial_no=1,
                                    ctime=now, mtime=now)
   utils.WriteFile(cfg_file,
                                    serial_no=1,
                                    ctime=now, mtime=now)
   utils.WriteFile(cfg_file,

@@ -352,18 +683,31 @@ def FinalizeClusterDestroy(master):
   """
   cfg = config.ConfigWriter()
   modify_ssh_setup = cfg.GetClusterInfo().modify_ssh_setup
   """
   cfg = config.ConfigWriter()
   modify_ssh_setup = cfg.GetClusterInfo().modify_ssh_setup

-  result = rpc.RpcRunner.call_node_stop_master(master, True)
+  runner = rpc.BootstrapRunner()
+
+  master_params = cfg.GetMasterNetworkParameters()
+  master_params.name = master
+  ems = cfg.GetUseExternalMipScript()
+  result = runner.call_node_deactivate_master_ip(master_params.name,
+                                                 master_params, ems)
+
+  msg = result.fail_msg
+  if msg:
+    logging.warning("Could not disable the master IP: %s", msg)
+
+  result = runner.call_node_stop_master(master)

   msg = result.fail_msg
   if msg:
     logging.warning("Could not disable the master role: %s", msg)
   msg = result.fail_msg
   if msg:
     logging.warning("Could not disable the master role: %s", msg)

-  result = rpc.RpcRunner.call_node_leave_cluster(master, modify_ssh_setup)
+
+  result = runner.call_node_leave_cluster(master, modify_ssh_setup)

   msg = result.fail_msg
   if msg:
     logging.warning("Could not shutdown the node daemon and cleanup"
                     " the node: %s", msg)
 
 
   msg = result.fail_msg
   if msg:
     logging.warning("Could not shutdown the node daemon and cleanup"
                     " the node: %s", msg)
 
 

-def SetupNodeDaemon(cluster_name, node, ssh_key_check):
+def SetupNodeDaemon(opts, cluster_name, node):

   """Add a node to the cluster.
 
   This function must be called before the actual opcode, and will ssh
   """Add a node to the cluster.
 
   This function must be called before the actual opcode, and will ssh

@@ -372,59 +716,21 @@ def SetupNodeDaemon(cluster_name, node, ssh_key_check):
 
   @param cluster_name: the cluster name
   @param node: the name of the new node
 
   @param cluster_name: the cluster name
   @param node: the name of the new node

-  @param ssh_key_check: whether to do a strict key check

 
   """
 
   """

-  sshrunner = ssh.SshRunner(cluster_name)
-
-  noded_cert = utils.ReadFile(constants.SSL_CERT_FILE)
-  rapi_cert = utils.ReadFile(constants.RAPI_CERT_FILE)
-  hmac_key = utils.ReadFile(constants.HMAC_CLUSTER_KEY)
-
-  # in the base64 pem encoding, neither '!' nor '.' are valid chars,
-  # so we use this to detect an invalid certificate; as long as the
-  # cert doesn't contain this, the here-document will be correctly
-  # parsed by the shell sequence below. HMAC keys are hexadecimal strings,
-  # so the same restrictions apply.
-  for content in (noded_cert, rapi_cert, hmac_key):
-    if re.search('^!EOF\.', content, re.MULTILINE):
-      raise errors.OpExecError("invalid SSL certificate or HMAC key")
-
-  if not noded_cert.endswith("\n"):
-    noded_cert += "\n"
-  if not rapi_cert.endswith("\n"):
-    rapi_cert += "\n"
-  if not hmac_key.endswith("\n"):
-    hmac_key += "\n"
-
-  # set up inter-node password and certificate and restarts the node daemon
-  # and then connect with ssh to set password and start ganeti-noded
-  # note that all the below variables are sanitized at this point,
-  # either by being constants or by the checks above
-  mycommand = ("umask 077 && "
-               "cat > '%s' << '!EOF.' && \n"
-               "%s!EOF.\n"
-               "cat > '%s' << '!EOF.' && \n"
-               "%s!EOF.\n"
-               "cat > '%s' << '!EOF.' && \n"
-               "%s!EOF.\n"
-               "chmod 0400 %s %s %s && "
-               "%s start %s" %
-               (constants.SSL_CERT_FILE, noded_cert,
-                constants.RAPI_CERT_FILE, rapi_cert,
-                constants.HMAC_CLUSTER_KEY, hmac_key,
-                constants.SSL_CERT_FILE, constants.RAPI_CERT_FILE,
-                constants.HMAC_CLUSTER_KEY,
-                constants.DAEMON_UTIL, constants.NODED))
-
-  result = sshrunner.Run(node, 'root', mycommand, batch=False,
-                         ask_key=ssh_key_check,
-                         use_cluster_key=False,
-                         strict_host_check=ssh_key_check)
-  if result.failed:
-    raise errors.OpExecError("Remote command on node %s, error: %s,"
-                             " output: %s" %
-                             (node, result.fail_reason, result.output))
+  data = {
+    constants.NDS_CLUSTER_NAME: cluster_name,
+    constants.NDS_NODE_DAEMON_CERTIFICATE:
+      utils.ReadFile(pathutils.NODED_CERT_FILE),
+    constants.NDS_SSCONF: ssconf.SimpleStore().ReadAll(),
+    constants.NDS_START_NODE_DAEMON: True,
+    }
+
+  RunNodeSetupCmd(cluster_name, node, pathutils.NODE_DAEMON_SETUP,
+                  opts.debug, opts.verbose,
+                  True, opts.ssh_key_check, opts.ssh_key_check, data)
+
+  _WaitForNodeDaemon(node)

 
 
 def MasterFailover(no_voting=False):
 
 
 def MasterFailover(no_voting=False):

@@ -457,7 +763,7 @@ def MasterFailover(no_voting=False):
                                " as master candidates. Only these nodes"
                                " can become masters. Current list of"
                                " master candidates is:\n"
                                " as master candidates. Only these nodes"
                                " can become masters. Current list of"
                                " master candidates is:\n"

-                               "%s" % ('\n'.join(mc_no_master)),
+                               "%s" % ("\n".join(mc_no_master)),

                                errors.ECODE_STATE)
 
   if not no_voting:
                                errors.ECODE_STATE)
 
   if not no_voting:

@@ -481,31 +787,77 @@ def MasterFailover(no_voting=False):
 
   logging.info("Setting master to %s, old master: %s", new_master, old_master)
 
 
   logging.info("Setting master to %s, old master: %s", new_master, old_master)
 

-  result = rpc.RpcRunner.call_node_stop_master(old_master, True)
+  try:
+    # instantiate a real config writer, as we now know we have the
+    # configuration data
+    cfg = config.ConfigWriter(accept_foreign=True)
+
+    cluster_info = cfg.GetClusterInfo()
+    cluster_info.master_node = new_master
+    # this will also regenerate the ssconf files, since we updated the
+    # cluster info
+    cfg.Update(cluster_info, logging.error)
+  except errors.ConfigurationError, err:
+    logging.error("Error while trying to set the new master: %s",
+                  str(err))
+    return 1
+
+  # if cfg.Update worked, then it means the old master daemon won't be
+  # able now to write its own config file (we rely on locking in both
+  # backend.UploadFile() and ConfigWriter._Write(); hence the next
+  # step is to kill the old master
+
+  logging.info("Stopping the master daemon on node %s", old_master)
+
+  runner = rpc.BootstrapRunner()
+  master_params = cfg.GetMasterNetworkParameters()
+  master_params.name = old_master
+  ems = cfg.GetUseExternalMipScript()
+  result = runner.call_node_deactivate_master_ip(master_params.name,
+                                                 master_params, ems)
+
+  msg = result.fail_msg
+  if msg:
+    logging.warning("Could not disable the master IP: %s", msg)
+
+  result = runner.call_node_stop_master(old_master)

   msg = result.fail_msg
   if msg:
     logging.error("Could not disable the master role on the old master"
   msg = result.fail_msg
   if msg:
     logging.error("Could not disable the master role on the old master"

-                 " %s, please disable manually: %s", old_master, msg)
+                  " %s, please disable manually: %s", old_master, msg)
+
+  logging.info("Checking master IP non-reachability...")
+
+  master_ip = sstore.GetMasterIP()
+  total_timeout = 30

 
   # Here we have a phase where no master should be running
 
   # Here we have a phase where no master should be running

+  def _check_ip():
+    if netutils.TcpPing(master_ip, constants.DEFAULT_NODED_PORT):
+      raise utils.RetryAgain()

 
 

-  # instantiate a real config writer, as we now know we have the
-  # configuration data
-  cfg = config.ConfigWriter()
+  try:
+    utils.Retry(_check_ip, (1, 1.5, 5), total_timeout)
+  except utils.RetryTimeout:
+    logging.warning("The master IP is still reachable after %s seconds,"
+                    " continuing but activating the master on the current"
+                    " node will probably fail", total_timeout)
+
+  if jstore.CheckDrainFlag():
+    logging.info("Undraining job queue")
+    jstore.SetDrainFlag(False)

 
 

-  cluster_info = cfg.GetClusterInfo()
-  cluster_info.master_node = new_master
-  # this will also regenerate the ssconf files, since we updated the
-  # cluster info
-  cfg.Update(cluster_info, logging.error)
+  logging.info("Starting the master daemons on the new master")

 
 

-  result = rpc.RpcRunner.call_node_start_master(new_master, True, no_voting)
+  result = rpc.BootstrapRunner().call_node_start_master_daemons(new_master,
+                                                                no_voting)

   msg = result.fail_msg
   if msg:
     logging.error("Could not start the master role on the new master"
                   " %s, please check: %s", new_master, msg)
     rcode = 1
 
   msg = result.fail_msg
   if msg:
     logging.error("Could not start the master role on the new master"
                   " %s, please check: %s", new_master, msg)
     rcode = 1
 

+  logging.info("Master failed over from %s to %s", old_master, new_master)

   return rcode
 
 
   return rcode
 
 

@@ -546,7 +898,7 @@ def GatherMasterVotes(node_list):
   @return: list of (node, votes)
 
   """
   @return: list of (node, votes)
 
   """

-  myself = utils.HostInfo().name
+  myself = netutils.Hostname.GetSysName()

   try:
     node_list.remove(myself)
   except ValueError:
   try:
     node_list.remove(myself)
   except ValueError:

@@ -554,7 +906,7 @@ def GatherMasterVotes(node_list):
   if not node_list:
     # no nodes left (eventually after removing myself)
     return []
   if not node_list:
     # no nodes left (eventually after removing myself)
     return []

-  results = rpc.RpcRunner.call_master_info(node_list)
+  results = rpc.BootstrapRunner().call_master_info(node_list)

   if not isinstance(results, dict):
     # this should not happen (unless internal error in rpc)
     logging.critical("Can't complete rpc call, aborting master startup")
   if not isinstance(results, dict):
     # this should not happen (unless internal error in rpc)
     logging.critical("Can't complete rpc call, aborting master startup")

@@ -568,6 +920,8 @@ def GatherMasterVotes(node_list):
     if msg:
       logging.warning("Error contacting node %s: %s", node, msg)
       fail = True
     if msg:
       logging.warning("Error contacting node %s: %s", node, msg)
       fail = True

+    # for now we accept both length 3, 4 and 5 (data[3] is primary ip version
+    # and data[4] is the master netmask)

     elif not isinstance(data, (tuple, list)) or len(data) < 3:
       logging.warning("Invalid data received from node %s: %s", node, data)
       fail = True
     elif not isinstance(data, (tuple, list)) or len(data) < 3:
       logging.warning("Invalid data received from node %s: %s", node, data)
       fail = True