X-Git-Url: https://code.grnet.gr/git/ganeti-local/blobdiff_plain/99ccf8b915722aaed2029d189b3d522d5a6c8760..4f2f98f170abda39b7075eaf61755bcec663f47a:/lib/bootstrap.py diff --git a/lib/bootstrap.py b/lib/bootstrap.py index 726f7c4..cb2f0f4 100644 --- a/lib/bootstrap.py +++ b/lib/bootstrap.py @@ -28,7 +28,9 @@ import os.path import re import logging import time +import tempfile +from ganeti.cmdlib import cluster from ganeti import rpc from ganeti import ssh from ganeti import utils @@ -39,11 +41,12 @@ from ganeti import objects from ganeti import ssconf from ganeti import serializer from ganeti import hypervisor -from ganeti import bdev +from ganeti.storage import drbd +from ganeti.storage import filestorage from ganeti import netutils -from ganeti import backend from ganeti import luxi from ganeti import jstore +from ganeti import pathutils # ec_id for InitConfig's temporary reservation manager @@ -60,7 +63,7 @@ def _InitSSHSetup(): 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): @@ -92,12 +95,12 @@ 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=constants.NODED_CERT_FILE, - rapicert_file=constants.RAPI_CERT_FILE, - spicecert_file=constants.SPICE_CERT_FILE, - spicecacert_file=constants.SPICE_CACERT_FILE, - hmackey_file=constants.CONFD_HMAC_KEY, - cds_file=constants.CLUSTER_DOMAIN_SECRET_FILE): + 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 @@ -208,7 +211,7 @@ def _InitGanetiServerSetup(master_name): # 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" % @@ -255,7 +258,85 @@ def _WaitForMasterDaemon(): " %s seconds" % _DAEMON_READY_TIMEOUT) -def _InitFileStorage(file_storage_dir): +def _WaitForSshDaemon(hostname, port, family): + """Wait for SSH daemon to become responsive. + + """ + hostip = netutils.GetHostname(name=hostname, family=family).ip + + def _CheckSshDaemon(): + if netutils.TcpPing(hostip, port, timeout=1.0, live_port_needed=True): + logging.debug("SSH daemon on %s:%s (IP address %s) has become" + " responsive", hostname, port, hostip) + else: + raise utils.RetryAgain() + + try: + utils.Retry(_CheckSshDaemon, 1.0, _DAEMON_READY_TIMEOUT) + except utils.RetryTimeout: + raise errors.OpExecError("SSH daemon on %s:%s (IP address %s) didn't" + " become responsive within %s seconds" % + (hostname, port, hostip, _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)) + + _WaitForSshDaemon(node, netutils.GetDaemonPort(constants.SSH), family) + + +def _InitFileStorageDir(file_storage_dir): """Initialize if needed the file storage. @param file_storage_dir: the user-supplied value @@ -281,9 +362,142 @@ def _InitFileStorage(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 _PrepareFileBasedStorage( + enabled_disk_templates, file_storage_dir, + default_dir, file_disk_template, + init_fn=_InitFileStorageDir, acceptance_fn=None): + """Checks if a file-base storage type is enabled and inits the dir. + + @type enabled_disk_templates: list of string + @param enabled_disk_templates: list of enabled disk templates + @type file_storage_dir: string + @param file_storage_dir: the file storage directory + @type default_dir: string + @param default_dir: default file storage directory when C{file_storage_dir} + is 'None' + @type file_disk_template: string + @param file_disk_template: a disk template whose storage type is 'ST_FILE' + @rtype: string + @returns: the name of the actual file storage directory + + """ + assert (file_disk_template in + utils.storage.GetDiskTemplatesOfStorageType(constants.ST_FILE)) + if file_storage_dir is None: + file_storage_dir = default_dir + if not acceptance_fn: + acceptance_fn = \ + lambda path: filestorage.CheckFileStoragePathAcceptance( + path, exact_match_ok=True) + + cluster.CheckFileStoragePathVsEnabledDiskTemplates( + logging.warning, file_storage_dir, enabled_disk_templates) + + file_storage_enabled = file_disk_template in enabled_disk_templates + if file_storage_enabled: + try: + acceptance_fn(file_storage_dir) + except errors.FileStoragePathError as e: + raise errors.OpPrereqError(str(e)) + result_file_storage_dir = init_fn(file_storage_dir) + else: + result_file_storage_dir = file_storage_dir + return result_file_storage_dir + + +def _PrepareFileStorage( + enabled_disk_templates, file_storage_dir, init_fn=_InitFileStorageDir, + acceptance_fn=None): + """Checks if file storage is enabled and inits the dir. + + @see: C{_PrepareFileBasedStorage} + + """ + return _PrepareFileBasedStorage( + enabled_disk_templates, file_storage_dir, + pathutils.DEFAULT_FILE_STORAGE_DIR, constants.DT_FILE, + init_fn=init_fn, acceptance_fn=acceptance_fn) + + +def _PrepareSharedFileStorage( + enabled_disk_templates, file_storage_dir, init_fn=_InitFileStorageDir, + acceptance_fn=None): + """Checks if shared file storage is enabled and inits the dir. + + @see: C{_PrepareFileBasedStorage} + + """ + return _PrepareFileBasedStorage( + enabled_disk_templates, file_storage_dir, + pathutils.DEFAULT_SHARED_FILE_STORAGE_DIR, constants.DT_SHARED_FILE, + init_fn=init_fn, acceptance_fn=acceptance_fn) + + +def _InitCheckEnabledDiskTemplates(enabled_disk_templates): + """Checks the sanity of the enabled disk templates. + + """ + if not enabled_disk_templates: + raise errors.OpPrereqError("Enabled disk templates list must contain at" + " least one member", errors.ECODE_INVAL) + invalid_disk_templates = \ + set(enabled_disk_templates) - constants.DISK_TEMPLATES + if invalid_disk_templates: + raise errors.OpPrereqError("Enabled disk templates list contains invalid" + " entries: %s" % invalid_disk_templates, + errors.ECODE_INVAL) + + +def _RestrictIpolicyToEnabledDiskTemplates(ipolicy, enabled_disk_templates): + """Restricts the ipolicy's disk templates to the enabled ones. + + This function clears the ipolicy's list of allowed disk templates from the + ones that are not enabled by the cluster. + + @type ipolicy: dict + @param ipolicy: the instance policy + @type enabled_disk_templates: list of string + @param enabled_disk_templates: the list of cluster-wide enabled disk + templates + + """ + assert constants.IPOLICY_DTS in ipolicy + allowed_disk_templates = ipolicy[constants.IPOLICY_DTS] + restricted_disk_templates = list(set(allowed_disk_templates) + .intersection(set(enabled_disk_templates))) + ipolicy[constants.IPOLICY_DTS] = restricted_disk_templates + + +def _InitCheckDrbdHelper(drbd_helper, drbd_enabled): + """Checks the DRBD usermode helper. + + @type drbd_helper: string + @param drbd_helper: name of the DRBD usermode helper that the system should + use + + """ + if not drbd_enabled: + return + + if drbd_helper is not None: + try: + curr_helper = drbd.DRBD8.GetUsermodeHelper() + except errors.BlockDeviceError, err: + raise errors.OpPrereqError("Error while checking drbd helper" + " (disable drbd with --enabled-disk-templates" + " if you are not using drbd): %s" % str(err), + 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) + + 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, @@ -293,11 +507,14 @@ def InitCluster(cluster_name, mac_prefix, # pylint: disable=R0913, R0914 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): + hv_state=None, disk_state=None, enabled_disk_templates=None): """Initialise the cluster. @type candidate_pool_size: int @param candidate_pool_size: master candidate pool size + @type enabled_disk_templates: list of string + @param enabled_disk_templates: list of disk_templates to be used in this + cluster """ # TODO: complete the docstring @@ -314,21 +531,24 @@ def InitCluster(cluster_name, mac_prefix, # pylint: disable=R0913, R0914 " entries: %s" % invalid_hvs, errors.ECODE_INVAL) + _InitCheckEnabledDiskTemplates(enabled_disk_templates) + try: ipcls = netutils.IPAddress.GetClassFromIpVersion(primary_ip_version) except errors.ProgrammerError: raise errors.OpPrereqError("Invalid primary ip version: %d." % - primary_ip_version) + 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)) + " address." % (hostname.ip, primary_ip_version), + errors.ECODE_INVAL) 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, constants.ETC_HOSTS), + (hostname.ip, pathutils.ETC_HOSTS), errors.ECODE_ENVIRON) if not ipcls.Own(hostname.ip): @@ -363,42 +583,32 @@ def InitCluster(cluster_name, mac_prefix, # pylint: disable=R0913, R0914 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)) + (master_netmask, primary_ip_version), + errors.ECODE_INVAL) else: master_netmask = ipcls.iplen - if vg_name is not None: + if vg_name: # Check if volume group is valid vgstatus = utils.CheckVolumeGroupSize(utils.ListVolumeGroups(), vg_name, constants.MIN_VG_SIZE) if vgstatus: - raise errors.OpPrereqError("Error: %s\nspecify --no-lvm-storage if" - " you are not using lvm" % vgstatus, - errors.ECODE_INVAL) + raise errors.OpPrereqError("Error: %s" % vgstatus, errors.ECODE_INVAL) - if drbd_helper is not None: - try: - 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) - 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) + drbd_enabled = constants.DT_DRBD8 in enabled_disk_templates + _InitCheckDrbdHelper(drbd_helper, drbd_enabled) - if constants.ENABLE_FILE_STORAGE: - file_storage_dir = _InitFileStorage(file_storage_dir) - else: - file_storage_dir = "" + 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_SHARED_FILE_STORAGE: - shared_file_storage_dir = _InitFileStorage(shared_file_storage_dir) - else: - shared_file_storage_dir = "" + file_storage_dir = _PrepareFileStorage(enabled_disk_templates, + file_storage_dir) + shared_file_storage_dir = _PrepareSharedFileStorage(enabled_disk_templates, + 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, @@ -410,7 +620,7 @@ def InitCluster(cluster_name, mac_prefix, # pylint: disable=R0913, R0914 (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) objects.UpgradeBeParams(beparams) @@ -420,6 +630,7 @@ def InitCluster(cluster_name, mac_prefix, # pylint: disable=R0913, R0914 objects.NIC.CheckParameterSyntax(nicparams) full_ipolicy = objects.FillIPolicy(constants.IPOLICY_DEFAULTS, ipolicy) + _RestrictIpolicyToEnabledDiskTemplates(full_ipolicy, enabled_disk_templates) if ndparams is not None: utils.ForceDictType(ndparams, constants.NDS_PARAMETER_TYPES) @@ -461,12 +672,32 @@ def InitCluster(cluster_name, mac_prefix, # pylint: disable=R0913, R0914 unknown_params = param_keys - default_param_keys raise errors.OpPrereqError("Invalid parameters for disk template %s:" " %s" % (template, - utils.CommaJoin(unknown_params))) + utils.CommaJoin(unknown_params)), + errors.ECODE_INVAL) utils.ForceDictType(dt_params, constants.DISK_DT_TYPES) + if template == constants.DT_DRBD8 and vg_name is not None: + # The default METAVG value is equal to the VG name set at init time, + # if provided + dt_params[constants.DRBD_DEFAULT_METAVG] = vg_name + + 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 - sshline = utils.ReadFile(constants.SSH_HOST_RSA_PUB) - sshkey = sshline.split(" ")[1] + rsa_sshkey = "" + dsa_sshkey = "" + if os.path.isfile(pathutils.SSH_HOST_RSA_PUB): + sshline = utils.ReadFile(pathutils.SSH_HOST_RSA_PUB) + rsa_sshkey = sshline.split(" ")[1] + if os.path.isfile(pathutils.SSH_HOST_DSA_PUB): + sshline = utils.ReadFile(pathutils.SSH_HOST_DSA_PUB) + dsa_sshkey = sshline.split(" ")[1] + if not rsa_sshkey and not dsa_sshkey: + raise errors.OpPrereqError("Failed to find SSH public keys", + errors.ECODE_ENVIRON) if modify_etc_hosts: utils.AddHostToEtcHosts(hostname.name, hostname.ip) @@ -494,12 +725,12 @@ def InitCluster(cluster_name, mac_prefix, # pylint: disable=R0913, R0914 # init of cluster config file cluster_config = objects.Cluster( serial_no=1, - rsahostkeypub=sshkey, + rsahostkeypub=rsa_sshkey, + dsahostkeypub=dsa_sshkey, highest_used_port=(constants.FIRST_DRBD_PORT - 1), mac_prefix=mac_prefix, volume_group_name=vg_name, tcpudp_port_pool=set(), - master_node=hostname.name, master_ip=clustername.ip, master_netmask=master_netmask, master_netdev=master_netdev, @@ -527,6 +758,7 @@ def InitCluster(cluster_name, mac_prefix, # pylint: disable=R0913, R0914 ipolicy=full_ipolicy, hv_state_static=hv_state, disk_state_static=disk_state, + enabled_disk_templates=enabled_disk_templates, ) master_node_config = objects.Node(name=hostname.name, primary_ip=hostname.ip, @@ -538,15 +770,15 @@ def InitCluster(cluster_name, mac_prefix, # pylint: disable=R0913, R0914 ) InitConfig(constants.CONFIG_VERSION, cluster_config, master_node_config) cfg = config.ConfigWriter(offline=True) - ssh.WriteKnownHostsFile(cfg, constants.SSH_KNOWN_HOSTS_FILE) + ssh.WriteKnownHostsFile(cfg, pathutils.SSH_KNOWN_HOSTS_FILE) cfg.Update(cfg.GetClusterInfo(), logging.error) - backend.WriteSsconfFiles(cfg.GetSsconfValues()) + ssconf.WriteSsconfFiles(cfg.GetSsconfValues()) # set up the inter-node password and certificate _InitGanetiServerSetup(hostname.name) logging.debug("Starting daemons") - result = utils.RunCmd([constants.DAEMON_UTIL, "start-all"]) + 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" % @@ -556,7 +788,7 @@ def InitCluster(cluster_name, mac_prefix, # pylint: disable=R0913, R0914 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 @@ -577,13 +809,14 @@ def InitConfig(version, cluster_config, master_node_config, _INITCONF_ECID) master_node_config.uuid = uuid_generator.Generate([], utils.NewUUID, _INITCONF_ECID) + cluster_config.master_node = master_node_config.uuid nodes = { - master_node_config.name: master_node_config, + master_node_config.uuid: 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], + members=[master_node_config.uuid], diskparams={}, ) nodegroups = { @@ -595,6 +828,7 @@ def InitConfig(version, cluster_config, master_node_config, nodegroups=nodegroups, nodes=nodes, instances={}, + networks={}, serial_no=1, ctime=now, mtime=now) utils.WriteFile(cfg_file, @@ -602,7 +836,7 @@ def InitConfig(version, cluster_config, master_node_config, mode=0600) -def FinalizeClusterDestroy(master): +def FinalizeClusterDestroy(master_uuid): """Execute the last steps of cluster destroy This function shuts down all the daemons, completing the destroy @@ -613,29 +847,31 @@ def FinalizeClusterDestroy(master): modify_ssh_setup = cfg.GetClusterInfo().modify_ssh_setup runner = rpc.BootstrapRunner() + master_name = cfg.GetNodeName(master_uuid) + master_params = cfg.GetMasterNetworkParameters() - master_params.name = master + master_params.uuid = master_uuid ems = cfg.GetUseExternalMipScript() - result = runner.call_node_deactivate_master_ip(master_params.name, - master_params, ems) + result = runner.call_node_deactivate_master_ip(master_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) + result = runner.call_node_stop_master(master_name) msg = result.fail_msg if msg: logging.warning("Could not disable the master role: %s", msg) - result = runner.call_node_leave_cluster(master, modify_ssh_setup) + result = runner.call_node_leave_cluster(master_name, modify_ssh_setup) 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 @@ -644,38 +880,19 @@ def SetupNodeDaemon(cluster_name, node, ssh_key_check): @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 """ - family = ssconf.SimpleStore().GetPrimaryIPFamily() - sshrunner = ssh.SshRunner(cluster_name, - ipv6=(family == netutils.IP6Address.family)) - - bind_address = constants.IP4_ADDRESS_ANY - if family == netutils.IP6Address.family: - bind_address = constants.IP6_ADDRESS_ANY - - # 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 - sshrunner.CopyFileToNode(node, constants.NODED_CERT_FILE) - sshrunner.CopyFileToNode(node, constants.RAPI_CERT_FILE) - sshrunner.CopyFileToNode(node, constants.SPICE_CERT_FILE) - sshrunner.CopyFileToNode(node, constants.SPICE_CACERT_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, - utils.ShellQuote(bind_address))) - - result = sshrunner.Run(node, "root", mycommand, batch=False, - ask_key=ssh_key_check, - use_cluster_key=True, - 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) @@ -695,7 +912,7 @@ def MasterFailover(no_voting=False): sstore = ssconf.SimpleStore() old_master, new_master = ssconf.GetMasterAndMyself(sstore) - node_list = sstore.GetNodeList() + node_names = sstore.GetNodeList() mc_list = sstore.GetMasterCandidates() if old_master == new_master: @@ -714,7 +931,7 @@ def MasterFailover(no_voting=False): errors.ECODE_STATE) if not no_voting: - vote_list = GatherMasterVotes(node_list) + vote_list = GatherMasterVotes(node_names) if vote_list: voted_master = vote_list[0][0] @@ -739,8 +956,20 @@ def MasterFailover(no_voting=False): # configuration data cfg = config.ConfigWriter(accept_foreign=True) + old_master_node = cfg.GetNodeInfoByName(old_master) + if old_master_node is None: + raise errors.OpPrereqError("Could not find old master node '%s' in" + " cluster configuration." % old_master, + errors.ECODE_NOENT) + cluster_info = cfg.GetClusterInfo() - cluster_info.master_node = new_master + new_master_node = cfg.GetNodeInfoByName(new_master) + if new_master_node is None: + raise errors.OpPrereqError("Could not find new master node '%s' in" + " cluster configuration." % new_master, + errors.ECODE_NOENT) + + cluster_info.master_node = new_master_node.uuid # this will also regenerate the ssconf files, since we updated the # cluster info cfg.Update(cluster_info, logging.error) @@ -758,9 +987,9 @@ def MasterFailover(no_voting=False): runner = rpc.BootstrapRunner() master_params = cfg.GetMasterNetworkParameters() - master_params.name = old_master + master_params.uuid = old_master_node.uuid ems = cfg.GetUseExternalMipScript() - result = runner.call_node_deactivate_master_ip(master_params.name, + result = runner.call_node_deactivate_master_ip(old_master, master_params, ems) msg = result.fail_msg @@ -771,7 +1000,7 @@ def MasterFailover(no_voting=False): 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...") @@ -824,7 +1053,7 @@ def GetMaster(): return old_master -def GatherMasterVotes(node_list): +def GatherMasterVotes(node_names): """Check the agreement on who is the master. This function will return a list of (node, number of votes), ordered @@ -838,8 +1067,8 @@ def GatherMasterVotes(node_list): since we use the same source for configuration information for both backend and boostrap, we'll always vote for ourselves. - @type node_list: list - @param node_list: the list of nodes to query for master info; the current + @type node_names: list + @param node_names: the list of nodes to query for master info; the current node will be removed if it is in the list @rtype: list @return: list of (node, votes) @@ -847,30 +1076,31 @@ def GatherMasterVotes(node_list): """ myself = netutils.Hostname.GetSysName() try: - node_list.remove(myself) + node_names.remove(myself) except ValueError: pass - if not node_list: + if not node_names: # no nodes left (eventually after removing myself) return [] - results = rpc.BootstrapRunner().call_master_info(node_list) + results = rpc.BootstrapRunner().call_master_info(node_names) if not isinstance(results, dict): # this should not happen (unless internal error in rpc) logging.critical("Can't complete rpc call, aborting master startup") - return [(None, len(node_list))] + return [(None, len(node_names))] votes = {} - for node in results: - nres = results[node] + for node_name in results: + nres = results[node_name] data = nres.payload msg = nres.fail_msg fail = False if msg: - logging.warning("Error contacting node %s: %s", node, msg) + logging.warning("Error contacting node %s: %s", node_name, 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) + logging.warning("Invalid data received from node %s: %s", + node_name, data) fail = True if fail: if None not in votes: