Fix a bug in command line option parsing code
[ganeti-local] / lib / bootstrap.py
index fbf5487..75f5808 100644 (file)
@@ -1,7 +1,7 @@
 #
 #
 
-# Copyright (C) 2006, 2007, 2008, 2010 Google Inc.
+# Copyright (C) 2006, 2007, 2008, 2010, 2011 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
@@ -43,6 +43,7 @@ from ganeti import bdev
 from ganeti import netutils
 from ganeti import backend
 from ganeti import luxi
+from ganeti import jstore
 
 
 # ec_id for InitConfig's temporary reservation manager
@@ -220,14 +221,11 @@ def _InitFileStorage(file_storage_dir):
       time) or the normalized path to the storage directory
 
   """
-  if not constants.ENABLE_FILE_STORAGE:
-    return ""
-
   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)
+    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:
@@ -244,13 +242,13 @@ def _InitFileStorage(file_storage_dir):
   return file_storage_dir
 
 
-def InitCluster(cluster_name, mac_prefix, # pylint: disable-msg=R0913
-                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,
-                maintain_node_health=False, drbd_helper=None,
-                uid_pool=None, default_iallocator=None,
+def InitCluster(cluster_name, mac_prefix, # pylint: disable=R0913
+                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,
+                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, prealloc_wipe_disks=False):
   """Initialise the cluster.
 
@@ -272,7 +270,6 @@ def InitCluster(cluster_name, mac_prefix, # pylint: disable-msg=R0913
                                " entries: %s" % invalid_hvs,
                                errors.ECODE_INVAL)
 
-
   ipcls = None
   if primary_ip_version == constants.IP4_VERSION:
     ipcls = netutils.IP4Address
@@ -345,7 +342,15 @@ def InitCluster(cluster_name, mac_prefix, # pylint: disable-msg=R0913
                                                              curr_helper),
                                  errors.ECODE_INVAL)
 
-  file_storage_dir = _InitFileStorage(file_storage_dir)
+  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,
@@ -364,6 +369,11 @@ def InitCluster(cluster_name, mac_prefix, # pylint: disable-msg=R0913
   utils.ForceDictType(nicparams, constants.NICS_PARAMETER_TYPES)
   objects.NIC.CheckParameterSyntax(nicparams)
 
+  if ndparams is not None:
+    utils.ForceDictType(ndparams, constants.NDS_PARAMETER_TYPES)
+  else:
+    ndparams = dict(constants.NDC_DEFAULTS)
+
   # hvparams is a mapping of hypervisor->hvparams dict
   for hv_name, hv_params in hvparams.iteritems():
     utils.ForceDictType(hv_params, constants.HVS_PARAMETER_TYPES)
@@ -388,6 +398,12 @@ def InitCluster(cluster_name, mac_prefix, # pylint: disable-msg=R0913
       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()
 
@@ -404,9 +420,11 @@ def InitCluster(cluster_name, mac_prefix, # pylint: disable-msg=R0913
     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},
+    ndparams=ndparams,
     hvparams=hvparams,
     candidate_pool_size=candidate_pool_size,
     modify_etc_hosts=modify_etc_hosts,
@@ -426,6 +444,7 @@ def InitCluster(cluster_name, mac_prefix, # pylint: disable-msg=R0913
                                     serial_no=1,
                                     master_candidate=True,
                                     offline=False, drained=False,
+                                    ctime=now, mtime=now,
                                     )
   InitConfig(constants.CONFIG_VERSION, cluster_config, master_node_config)
   cfg = config.ConfigWriter(offline=True)
@@ -473,7 +492,7 @@ def InitConfig(version, cluster_config, master_node_config,
     }
   default_nodegroup = objects.NodeGroup(
     uuid=uuid_generator.Generate([], utils.NewUUID, _INITCONF_ECID),
-    name="default",
+    name=constants.INITIAL_NODE_GROUP_NAME,
     members=[master_node_config.name],
     )
   nodegroups = {
@@ -501,7 +520,7 @@ def FinalizeClusterDestroy(master):
   """
   cfg = config.ConfigWriter()
   modify_ssh_setup = cfg.GetClusterInfo().modify_ssh_setup
-  result = rpc.RpcRunner.call_node_stop_master(master, True)
+  result = rpc.RpcRunner.call_node_stop_master(master)
   msg = result.fail_msg
   if msg:
     logging.warning("Could not disable the master role: %s", msg)
@@ -526,27 +545,7 @@ def SetupNodeDaemon(cluster_name, node, ssh_key_check):
   """
   family = ssconf.SimpleStore().GetPrimaryIPFamily()
   sshrunner = ssh.SshRunner(cluster_name,
-                            ipv6=family==netutils.IP6Address.family)
-
-  noded_cert = utils.ReadFile(constants.NODED_CERT_FILE)
-  rapi_cert = utils.ReadFile(constants.RAPI_CERT_FILE)
-  confd_hmac_key = utils.ReadFile(constants.CONFD_HMAC_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, confd_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 confd_hmac_key.endswith("\n"):
-    confd_hmac_key += "\n"
+                            ipv6=(family == netutils.IP6Address.family))
 
   bind_address = constants.IP4_ADDRESS_ANY
   if family == netutils.IP6Address.family:
@@ -559,10 +558,9 @@ def SetupNodeDaemon(cluster_name, node, ssh_key_check):
   sshrunner.CopyFileToNode(node, constants.NODED_CERT_FILE)
   sshrunner.CopyFileToNode(node, constants.RAPI_CERT_FILE)
   sshrunner.CopyFileToNode(node, constants.CONFD_HMAC_KEY)
-  mycommand = ("%s stop-all; %s start %s -b '%s'" % (constants.DAEMON_UTIL,
-                                                     constants.DAEMON_UTIL,
-                                                     constants.NODED,
-                                                     bind_address))
+  mycommand = ("%s stop-all; %s start %s -b %s" %
+               (constants.DAEMON_UTIL, constants.DAEMON_UTIL, constants.NODED,
+                utils.ShellQuote(bind_address)))
 
   result = sshrunner.Run(node, 'root', mycommand, batch=False,
                          ask_key=ssh_key_check,
@@ -652,7 +650,7 @@ def MasterFailover(no_voting=False):
 
   logging.info("Stopping the master daemon on node %s", old_master)
 
-  result = rpc.RpcRunner.call_node_stop_master(old_master, True)
+  result = rpc.RpcRunner.call_node_stop_master(old_master)
   msg = result.fail_msg
   if msg:
     logging.error("Could not disable the master role on the old master"
@@ -662,6 +660,7 @@ def MasterFailover(no_voting=False):
 
   master_ip = sstore.GetMasterIP()
   total_timeout = 30
+
   # Here we have a phase where no master should be running
   def _check_ip():
     if netutils.TcpPing(master_ip, constants.DEFAULT_NODED_PORT):
@@ -674,9 +673,13 @@ def MasterFailover(no_voting=False):
                     " 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)
+
   logging.info("Starting the master daemons on the new master")
 
-  result = rpc.RpcRunner.call_node_start_master(new_master, True, no_voting)
+  result = rpc.RpcRunner.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"