X-Git-Url: https://code.grnet.gr/git/ganeti-local/blobdiff_plain/827f753e27c17007fe424372bc14e64984f65419..c8fcde472922e4ee664d904e0bf1a583f1d5040d:/scripts/gnt-node diff --git a/scripts/gnt-node b/scripts/gnt-node index dc988cc..c60ab41 100755 --- a/scripts/gnt-node +++ b/scripts/gnt-node @@ -19,50 +19,108 @@ # 02110-1301, USA. +# pylint: disable-msg=W0401,W0614 +# W0401: Wildcard import ganeti.cli +# W0614: Unused import %s from wildcard import (since we need cli) + import sys from optparse import make_option from ganeti.cli import * +from ganeti import cli from ganeti import opcodes -from ganeti import logger from ganeti import utils from ganeti import constants from ganeti import errors from ganeti import bootstrap +#: default list of field for L{ListNodes} _LIST_DEF_FIELDS = [ "name", "dtotal", "dfree", "mtotal", "mnode", "mfree", "pinst_cnt", "sinst_cnt", ] +#: headers (and full field list for L{ListNodes} +_LIST_HEADERS = { + "name": "Node", "pinst_cnt": "Pinst", "sinst_cnt": "Sinst", + "pinst_list": "PriInstances", "sinst_list": "SecInstances", + "pip": "PrimaryIP", "sip": "SecondaryIP", + "dtotal": "DTotal", "dfree": "DFree", + "mtotal": "MTotal", "mnode": "MNode", "mfree": "MFree", + "bootid": "BootID", + "ctotal": "CTotal", "cnodes": "CNodes", "csockets": "CSockets", + "tags": "Tags", + "serial_no": "SerialNo", + "master_candidate": "MasterC", + "master": "IsMaster", + "offline": "Offline", "drained": "Drained", + } + + +@UsesRPC def AddNode(opts, args): - """Add node cli-to-processor bridge. + """Add a node to the cluster. + + @param opts: the command line options selected by the user + @type args: list + @param args: should contain only one element, the new node name + @rtype: int + @return: the desired exit code """ + cl = GetClient() dns_data = utils.HostInfo(args[0]) node = dns_data.name - - if not opts.readd: - op = opcodes.OpQueryNodes(output_fields=['name'], names=[node]) - try: - output = SubmitOpCode(op) - except (errors.OpPrereqError, errors.OpExecError): - pass - else: - logger.ToStderr("Node %s already in the cluster (as %s)" - " - please use --readd" % (args[0], output[0][0])) + readd = opts.readd + + try: + output = cl.QueryNodes(names=[node], fields=['name', 'sip'], + use_locking=False) + node_exists, sip = output[0] + except (errors.OpPrereqError, errors.OpExecError): + node_exists = "" + sip = None + + if readd: + if not node_exists: + ToStderr("Node %s not in the cluster" + " - please retry without '--readd'", node) return 1 + else: + if node_exists: + ToStderr("Node %s already in the cluster (as %s)" + " - please retry with '--readd'", node, node_exists) + return 1 + sip = opts.secondary_ip + + # read the cluster name from the master + output = cl.QueryConfigValues(['cluster_name']) + cluster_name = output[0] + + if readd: + # clear the offline and drain flags on the node + ToStdout("Resetting the 'offline' and 'drained' flags due to re-add") + op = opcodes.OpSetNodeParams(node_name=node, force=True, + offline=False, drained=False) + + result = SubmitOpCode(op, cl=cl) + if result: + ToStdout("Modified:") + for param, data in result: + ToStdout(" - %-5s -> %s", param, data) + else: + ToStderr("-- WARNING -- \n" + "Performing this operation is going to replace the ssh daemon" + " keypair\n" + "on the target machine (%s) with the ones of the" + " current one\n" + "and grant full intra-cluster ssh root access to/from it\n", node) - logger.ToStderr("-- WARNING -- \n" - "Performing this operation is going to replace the ssh daemon keypair\n" - "on the target machine (%s) with the ones of the current one\n" - "and grant full intra-cluster ssh root access to/from it\n" % node) - - bootstrap.SetupNodeDaemon(node) + bootstrap.SetupNodeDaemon(cluster_name, node, opts.ssh_key_check) - op = opcodes.OpAddNode(node_name=args[0], secondary_ip=opts.secondary_ip, + op = opcodes.OpAddNode(node_name=args[0], secondary_ip=sip, readd=opts.readd) SubmitOpCode(op) @@ -70,6 +128,12 @@ def AddNode(opts, args): def ListNodes(opts, args): """List nodes and their properties. + @param opts: the command line options selected by the user + @type args: list + @param args: should be an empty list + @rtype: int + @return: the desired exit code + """ if opts.output is None: selected_fields = _LIST_DEF_FIELDS @@ -78,32 +142,19 @@ def ListNodes(opts, args): else: selected_fields = opts.output.split(",") - op = opcodes.OpQueryNodes(output_fields=selected_fields, names=[]) - output = SubmitOpCode(op) + output = GetClient().QueryNodes(args, selected_fields, opts.do_locking) if not opts.no_headers: - headers = { - "name": "Node", "pinst_cnt": "Pinst", "sinst_cnt": "Sinst", - "pinst_list": "PriInstances", "sinst_list": "SecInstances", - "pip": "PrimaryIP", "sip": "SecondaryIP", - "dtotal": "DTotal", "dfree": "DFree", - "mtotal": "MTotal", "mnode": "MNode", "mfree": "MFree", - "bootid": "BootID", - "ctotal": "CTotal", - "tags": "Tags", - } + headers = _LIST_HEADERS else: headers = None - if opts.human_readable: - unitfields = ["dtotal", "dfree", "mtotal", "mnode", "mfree"] - else: - unitfields = None + unitfields = ["dtotal", "dfree", "mtotal", "mnode", "mfree"] numfields = ["dtotal", "dfree", "mtotal", "mnode", "mfree", "pinst_cnt", "sinst_cnt", - "ctotal"] + "ctotal", "serial_no"] list_type_fields = ("pinst_list", "sinst_list", "tags") # change raw values to nicer strings @@ -112,15 +163,20 @@ def ListNodes(opts, args): val = row[idx] if field in list_type_fields: val = ",".join(val) + elif field in ('master', 'master_candidate', 'offline', 'drained'): + if val: + val = 'Y' + else: + val = 'N' elif val is None: val = "?" row[idx] = str(val) data = GenerateTable(separator=opts.separator, headers=headers, fields=selected_fields, unitfields=unitfields, - numfields=numfields, data=output) + numfields=numfields, data=output, units=opts.units) for line in data: - logger.ToStdout(line) + ToStdout(line) return 0 @@ -128,77 +184,91 @@ def ListNodes(opts, args): def EvacuateNode(opts, args): """Relocate all secondary instance from a node. + @param opts: the command line options selected by the user + @type args: list + @param args: should be an empty list + @rtype: int + @return: the desired exit code + """ + cl = GetClient() force = opts.force + + dst_node = opts.dst_node + iallocator = opts.iallocator + + cnt = [dst_node, iallocator].count(None) + if cnt != 1: + raise errors.OpPrereqError("One and only one of the -n and -i" + " options must be passed") + selected_fields = ["name", "sinst_list"] - src_node, dst_node = args + src_node = args[0] - op = opcodes.OpQueryNodes(output_fields=selected_fields, names=[src_node]) - result = SubmitOpCode(op) + result = cl.QueryNodes(names=[src_node], fields=selected_fields, + use_locking=False) src_node, sinst = result[0] - op = opcodes.OpQueryNodes(output_fields=["name"], names=[dst_node]) - result = SubmitOpCode(op) - dst_node = result[0][0] - - if src_node == dst_node: - raise errors.OpPrereqError("Evacuate node needs different source and" - " target nodes (node %s given twice)" % - src_node) if not sinst: - logger.ToStderr("No secondary instances on node %s, exiting." % src_node) + ToStderr("No secondary instances on node %s, exiting.", src_node) return constants.EXIT_SUCCESS - sinst = utils.NiceSort(sinst) + if dst_node is not None: + result = cl.QueryNodes(names=[dst_node], fields=["name"], + use_locking=False) + dst_node = result[0][0] - retcode = constants.EXIT_SUCCESS + if src_node == dst_node: + raise errors.OpPrereqError("Evacuate node needs different source and" + " target nodes (node %s given twice)" % + src_node) + txt_msg = "to node %s" % dst_node + else: + txt_msg = "using iallocator %s" % iallocator + + sinst = utils.NiceSort(sinst) if not force and not AskUser("Relocate instance(s) %s from node\n" - " %s to node\n %s?" % + " %s %s?" % (",".join("'%s'" % name for name in sinst), - src_node, dst_node)): + src_node, txt_msg)): return constants.EXIT_CONFIRMATION - good_cnt = bad_cnt = 0 + ops = [] for iname in sinst: op = opcodes.OpReplaceDisks(instance_name=iname, remote_node=dst_node, - mode=constants.REPLACE_DISK_ALL, - disks=["sda", "sdb"]) - try: - logger.ToStdout("Replacing disks for instance %s" % iname) - SubmitOpCode(op) - logger.ToStdout("Instance %s has been relocated" % iname) - good_cnt += 1 - except errors.GenericError, err: - nret, msg = FormatError(err) - retcode |= nret - logger.ToStderr("Error replacing disks for instance %s: %s" % - (iname, msg)) - bad_cnt += 1 - - if retcode == constants.EXIT_SUCCESS: - logger.ToStdout("All %d instance(s) relocated successfully." % good_cnt) - else: - logger.ToStdout("There were errors during the relocation:\n" - "%d error(s) out of %d instance(s)." % - (bad_cnt, good_cnt + bad_cnt)) - return retcode + mode=constants.REPLACE_DISK_CHG, + iallocator=iallocator, + disks=[]) + ops.append(op) + + job_id = cli.SendJob(ops, cl=cl) + cli.PollJob(job_id, cl=cl) def FailoverNode(opts, args): """Failover all primary instance on a node. + @param opts: the command line options selected by the user + @type args: list + @param args: should be an empty list + @rtype: int + @return: the desired exit code + """ + cl = GetClient() force = opts.force selected_fields = ["name", "pinst_list"] - op = opcodes.OpQueryNodes(output_fields=selected_fields, names=args) - result = SubmitOpCode(op) + # these fields are static data anyway, so it doesn't matter, but + # locking=True should be safer + result = cl.QueryNodes(names=args, fields=selected_fields, + use_locking=False) node, pinst = result[0] if not pinst: - logger.ToStderr("No primary instances on node %s, exiting." % node) + ToStderr("No primary instances on node %s, exiting.", node) return 0 pinst = utils.NiceSort(pinst) @@ -209,68 +279,151 @@ def FailoverNode(opts, args): (",".join("'%s'" % name for name in pinst))): return 2 - good_cnt = bad_cnt = 0 + jex = JobExecutor(cl=cl) for iname in pinst: op = opcodes.OpFailoverInstance(instance_name=iname, ignore_consistency=opts.ignore_consistency) - try: - logger.ToStdout("Failing over instance %s" % iname) - SubmitOpCode(op) - logger.ToStdout("Instance %s has been failed over" % iname) - good_cnt += 1 - except errors.GenericError, err: - nret, msg = FormatError(err) - retcode |= nret - logger.ToStderr("Error failing over instance %s: %s" % (iname, msg)) - bad_cnt += 1 - - if retcode == 0: - logger.ToStdout("All %d instance(s) failed over successfully." % good_cnt) + jex.QueueJob(iname, op) + results = jex.GetResults() + bad_cnt = len([row for row in results if not row[0]]) + if bad_cnt == 0: + ToStdout("All %d instance(s) failed over successfully.", len(results)) + else: + ToStdout("There were errors during the failover:\n" + "%d error(s) out of %d instance(s).", bad_cnt, len(results)) + return retcode + + +def MigrateNode(opts, args): + """Migrate all primary instance on a node. + + """ + cl = GetClient() + force = opts.force + selected_fields = ["name", "pinst_list"] + + result = cl.QueryNodes(names=args, fields=selected_fields, use_locking=False) + node, pinst = result[0] + + if not pinst: + ToStdout("No primary instances on node %s, exiting." % node) + return 0 + + pinst = utils.NiceSort(pinst) + + retcode = 0 + + if not force and not AskUser("Migrate instance(s) %s?" % + (",".join("'%s'" % name for name in pinst))): + return 2 + + jex = JobExecutor(cl=cl) + for iname in pinst: + op = opcodes.OpMigrateInstance(instance_name=iname, live=opts.live, + cleanup=False) + jex.QueueJob(iname, op) + + results = jex.GetResults() + bad_cnt = len([row for row in results if not row[0]]) + if bad_cnt == 0: + ToStdout("All %d instance(s) migrated successfully.", len(results)) else: - logger.ToStdout("There were errors during the failover:\n" - "%d error(s) out of %d instance(s)." % - (bad_cnt, good_cnt + bad_cnt)) + ToStdout("There were errors during the migration:\n" + "%d error(s) out of %d instance(s).", bad_cnt, len(results)) return retcode def ShowNodeConfig(opts, args): """Show node information. - """ - op = opcodes.OpQueryNodes(output_fields=["name", "pip", "sip", - "pinst_list", "sinst_list"], - names=args) - result = SubmitOpCode(op) + @param opts: the command line options selected by the user + @type args: list + @param args: should either be an empty list, in which case + we show information about all nodes, or should contain + a list of nodes to be queried for information + @rtype: int + @return: the desired exit code - for name, primary_ip, secondary_ip, pinst, sinst in result: - logger.ToStdout("Node name: %s" % name) - logger.ToStdout(" primary ip: %s" % primary_ip) - logger.ToStdout(" secondary ip: %s" % secondary_ip) + """ + cl = GetClient() + result = cl.QueryNodes(fields=["name", "pip", "sip", + "pinst_list", "sinst_list", + "master_candidate", "drained", "offline"], + names=args, use_locking=False) + + for (name, primary_ip, secondary_ip, pinst, sinst, + is_mc, drained, offline) in result: + ToStdout("Node name: %s", name) + ToStdout(" primary ip: %s", primary_ip) + ToStdout(" secondary ip: %s", secondary_ip) + ToStdout(" master candidate: %s", is_mc) + ToStdout(" drained: %s", drained) + ToStdout(" offline: %s", offline) if pinst: - logger.ToStdout(" primary for instances:") - for iname in pinst: - logger.ToStdout(" - %s" % iname) + ToStdout(" primary for instances:") + for iname in utils.NiceSort(pinst): + ToStdout(" - %s", iname) else: - logger.ToStdout(" primary for no instances") + ToStdout(" primary for no instances") if sinst: - logger.ToStdout(" secondary for instances:") - for iname in sinst: - logger.ToStdout(" - %s" % iname) + ToStdout(" secondary for instances:") + for iname in utils.NiceSort(sinst): + ToStdout(" - %s", iname) else: - logger.ToStdout(" secondary for no instances") + ToStdout(" secondary for no instances") return 0 def RemoveNode(opts, args): - """Remove node cli-to-processor bridge.""" + """Remove a node from the cluster. + + @param opts: the command line options selected by the user + @type args: list + @param args: should contain only one element, the name of + the node to be removed + @rtype: int + @return: the desired exit code + + """ op = opcodes.OpRemoveNode(node_name=args[0]) SubmitOpCode(op) + return 0 + + +def PowercycleNode(opts, args): + """Remove a node from the cluster. + + @param opts: the command line options selected by the user + @type args: list + @param args: should contain only one element, the name of + the node to be removed + @rtype: int + @return: the desired exit code + + """ + node = args[0] + if (not opts.confirm and + not AskUser("Are you sure you want to hard powercycle node %s?" % node)): + return 2 + + op = opcodes.OpPowercycleNode(node_name=node, force=opts.force) + result = SubmitOpCode(op) + ToStderr(result) + return 0 def ListVolumes(opts, args): """List logical volumes on node(s). + @param opts: the command line options selected by the user + @type args: list + @param args: should either be an empty list, in which case + we list data for all nodes, or contain a list of nodes + to display data only for those + @rtype: int + @return: the desired exit code + """ if opts.output is None: selected_fields = ["node", "phys", "vg", @@ -288,23 +441,63 @@ def ListVolumes(opts, args): else: headers = None - if opts.human_readable: - unitfields = ["size"] - else: - unitfields = None + unitfields = ["size"] numfields = ["size"] data = GenerateTable(separator=opts.separator, headers=headers, fields=selected_fields, unitfields=unitfields, - numfields=numfields, data=output) + numfields=numfields, data=output, units=opts.units) for line in data: - logger.ToStdout(line) + ToStdout(line) return 0 +def SetNodeParams(opts, args): + """Modifies a node. + + @param opts: the command line options selected by the user + @type args: list + @param args: should contain only one element, the node name + @rtype: int + @return: the desired exit code + + """ + if [opts.master_candidate, opts.drained, opts.offline].count(None) == 3: + ToStderr("Please give at least one of the parameters.") + return 1 + + if opts.master_candidate is not None: + candidate = opts.master_candidate == 'yes' + else: + candidate = None + if opts.offline is not None: + offline = opts.offline == 'yes' + else: + offline = None + + if opts.drained is not None: + drained = opts.drained == 'yes' + else: + drained = None + op = opcodes.OpSetNodeParams(node_name=args[0], + master_candidate=candidate, + offline=offline, + drained=drained, + force=opts.force) + + # even if here we process the result, we allow submit only + result = SubmitOrSend(op, opts) + + if result: + ToStdout("Modified node %s", args[0]) + for param, data in result: + ToStdout(" - %-5s -> %s", param, data) + return 0 + + commands = { 'add': (AddNode, ARGS_ONE, [DEBUG_OPT, @@ -314,14 +507,26 @@ commands = { make_option("--readd", dest="readd", default=False, action="store_true", help="Readd old node after replacing it"), + make_option("--no-ssh-key-check", dest="ssh_key_check", + default=True, action="store_false", + help="Disable SSH key fingerprint checking"), ], - "[-s ip] [--readd] ", "Add a node to the cluster"), - 'evacuate': (EvacuateNode, ARGS_FIXED(2), - [DEBUG_OPT, FORCE_OPT], - "[-f] ", - "Relocate the secondary instances from the first node" - " to the second one (only for instances with drbd disk template" - ), + "[-s ip] [--readd] [--no-ssh-key-check] ", + "Add a node to the cluster"), + 'evacuate': (EvacuateNode, ARGS_ONE, + [DEBUG_OPT, FORCE_OPT, + make_option("-n", "--new-secondary", dest="dst_node", + help="New secondary node", metavar="NODE", + default=None), + make_option("-I", "--iallocator", metavar="", + help="Select new secondary for the instance" + " automatically using the" + " iallocator plugin", + default=None, type="string"), + ], + "[-f] {-I | -n } ", + "Relocate the secondary instances from a node" + " to other nodes (only for instances with drbd disk template)"), 'failover': (FailoverNode, ARGS_ONE, [DEBUG_OPT, FORCE_OPT, make_option("--ignore-consistency", dest="ignore_consistency", @@ -332,16 +537,45 @@ commands = { "[-f] ", "Stops the primary instances on a node and start them on their" " secondary node (only for instances with drbd disk template)"), + 'migrate': (MigrateNode, ARGS_ONE, + [DEBUG_OPT, FORCE_OPT, + make_option("--non-live", dest="live", + default=True, action="store_false", + help="Do a non-live migration (this usually means" + " freeze the instance, save the state," + " transfer and only then resume running on the" + " secondary node)"), + ], + "[-f] ", + "Migrate all the primary instance on a node away from it" + " (only for instances of type drbd)"), 'info': (ShowNodeConfig, ARGS_ANY, [DEBUG_OPT], "[...]", "Show information about the node(s)"), - 'list': (ListNodes, ARGS_NONE, - [DEBUG_OPT, NOHDR_OPT, SEP_OPT, USEUNITS_OPT, FIELDS_OPT], - "", "Lists the nodes in the cluster. The available fields" - " are (see the man page for details): name, pinst_cnt, pinst_list," - " sinst_cnt, sinst_list, pip, sip, dtotal, dfree, mtotal, mnode," - " mfree, bootid, cpu_count. The default field list is" - " (in order): %s." % ", ".join(_LIST_DEF_FIELDS), - ), + 'list': (ListNodes, ARGS_ANY, + [DEBUG_OPT, NOHDR_OPT, SEP_OPT, USEUNITS_OPT, FIELDS_OPT, SYNC_OPT], + "[nodes...]", + "Lists the nodes in the cluster. The available fields" + " are (see the man page for details): %s" + " The default field list is (in order): %s." % + (", ".join(_LIST_HEADERS), ", ".join(_LIST_DEF_FIELDS))), + 'modify': (SetNodeParams, ARGS_ONE, + [DEBUG_OPT, FORCE_OPT, + SUBMIT_OPT, + make_option("-C", "--master-candidate", dest="master_candidate", + choices=('yes', 'no'), default=None, + metavar="yes|no", + help="Set the master_candidate flag on the node"), + + make_option("-O", "--offline", dest="offline", metavar="yes|no", + choices=('yes', 'no'), default=None, + help="Set the offline flag on the node"), + make_option("-D", "--drained", dest="drained", metavar="yes|no", + choices=('yes', 'no'), default=None, + help="Set the drained flag on the node"), + ], + "", "Alters the parameters of an instance"), + 'powercycle': (PowercycleNode, ARGS_ONE, [DEBUG_OPT, FORCE_OPT, CONFIRM_OPT], + "", "Tries to forcefully powercycle a node"), 'remove': (RemoveNode, ARGS_ONE, [DEBUG_OPT], "", "Removes a node from the cluster"), 'volumes': (ListVolumes, ARGS_ANY,