Optimize the OpQueryNodes for names only
[ganeti-local] / scripts / gnt-cluster
index 412ceeb..31e6425 100755 (executable)
 import sys
 from optparse import make_option
 import pprint
+import os.path
 
 from ganeti.cli import *
 from ganeti import opcodes
+from ganeti import constants
+from ganeti import errors
+from ganeti import utils
+from ganeti import bootstrap
+from ganeti import ssh
+from ganeti import ssconf
 
 
 def InitCluster(opts, args):
@@ -35,13 +42,22 @@ def InitCluster(opts, args):
     args - list of arguments, expected to be [clustername]
 
   """
-  op = opcodes.OpInitCluster(cluster_name=args[0],
-                             secondary_ip=opts.secondary_ip,
-                             hypervisor_type=opts.hypervisor_type,
-                             vg_name=opts.vg_name,
-                             mac_prefix=opts.mac_prefix,
-                             def_bridge=opts.def_bridge)
-  SubmitOpCode(op)
+  if not opts.lvm_storage and opts.vg_name:
+    print ("Options --no-lvm-storage and --vg-name conflict.")
+    return 1
+
+  vg_name = opts.vg_name
+  if opts.lvm_storage and not opts.vg_name:
+    vg_name = constants.DEFAULT_VG
+
+  bootstrap.InitCluster(cluster_name=args[0],
+                        secondary_ip=opts.secondary_ip,
+                        hypervisor_type=opts.hypervisor_type,
+                        vg_name=vg_name,
+                        mac_prefix=opts.mac_prefix,
+                        def_bridge=opts.def_bridge,
+                        master_netdev=opts.master_netdev,
+                        file_storage_dir=opts.file_storage_dir)
   return 0
 
 
@@ -50,13 +66,39 @@ def DestroyCluster(opts, args):
 
   Args:
     opts - class with options as members
+
   """
   if not opts.yes_do_it:
     print ("Destroying a cluster is irreversibly. If you really want destroy"
-           "this cluster, supply the --yes-do-it option.")
+           " this cluster, supply the --yes-do-it option.")
     return 1
 
   op = opcodes.OpDestroyCluster()
+  master = SubmitOpCode(op)
+  # if we reached this, the opcode didn't fail; we can proceed to
+  # shutdown all the daemons
+  bootstrap.FinalizeClusterDestroy(master)
+  return 0
+
+
+def RenameCluster(opts, args):
+  """Rename the cluster.
+
+  Args:
+    opts - class with options as members, we use force only
+    args - list of arguments, expected to be [new_name]
+
+  """
+  name = args[0]
+  if not opts.force:
+    usertext = ("This will rename the cluster to '%s'. If you are connected"
+                " over the network to the cluster name, the operation is very"
+                " dangerous as the IP address will be removed from the node"
+                " and the change may not go through. Continue?") % name
+    if not AskUser(usertext):
+      return 1
+
+  op = opcodes.OpRenameCluster(name=name)
   SubmitOpCode(op)
   return 0
 
@@ -85,9 +127,8 @@ def ShowClusterMaster(opts, args):
     opts - class with options as members
 
   """
-  op = opcodes.OpQueryClusterInfo()
-  result = SubmitOpCode(op)
-  print (result["master"])
+  sstore = ssconf.SimpleStore()
+  print sstore.GetMasterNode()
   return 0
 
 
@@ -100,17 +141,12 @@ def ShowClusterConfig(opts, args):
 
   print ("Cluster name: %s" % result["name"])
 
-  print ("Architecture: %s (%s)" %
-            (result["architecture"][0], result["architecture"][1]))
-
   print ("Master node: %s" % result["master"])
 
-  print ("Instances:")
-  for name, node in result["instances"]:
-    print ("  - %s (on %s)" % (name, node))
-  print ("Nodes:")
-  for name in result["nodes"]:
-    print ("  - %s" % name)
+  print ("Architecture (this node): %s (%s)" %
+         (result["architecture"][0], result["architecture"][1]))
+
+  print ("Cluster hypervisor: %s" % result["hypervisor_type"])
 
   return 0
 
@@ -125,8 +161,20 @@ def ClusterCopyFile(opts, args):
     nodes - list containing the name of target nodes; if empty, all nodes
 
   """
-  op = opcodes.OpClusterCopyFile(filename=args[0], nodes=opts.nodes)
-  SubmitOpCode(op)
+  filename = args[0]
+  if not os.path.exists(filename):
+    raise errors.OpPrereqError("No such filename '%s'" % filename)
+
+  myname = utils.HostInfo().name
+
+  op = opcodes.OpQueryNodes(output_fields=["name"], names=opts.nodes)
+  results = [row[0] for row in SubmitOpCode(op) if row[0] != myname]
+  srun = ssh.SshRunner()
+  for node in results:
+    if not srun.CopyFileToNode(node, filename):
+      print >> sys.stderr, ("Copy of file %s to node %s failed" %
+                            (filename, node))
+
   return 0
 
 
@@ -141,15 +189,25 @@ def RunClusterCommand(opts, args):
 
   """
   command = " ".join(args)
-  nodes = opts.nodes
-  op = opcodes.OpRunClusterCommand(command=command, nodes=nodes)
-  result = SubmitOpCode(op)
-  for node, sshcommand, output, exit_code in result:
+  op = opcodes.OpQueryNodes(output_fields=["name"], names=opts.nodes)
+  nodes = [row[0] for row in SubmitOpCode(op)]
+
+  sstore = ssconf.SimpleStore()
+  master_node = sstore.GetMasterNode()
+  srun = ssh.SshRunner(sstore=sstore)
+
+  if master_node in nodes:
+    nodes.remove(master_node)
+    nodes.append(master_node)
+
+  for name in nodes:
+    result = srun.Run(name, "root", command)
     print ("------------------------------------------------")
-    print ("node: %s" % node)
-    print ("command: %s" % sshcommand)
-    print ("%s" % output)
-    print ("return code = %s" % exit_code)
+    print ("node: %s" % name)
+    print ("%s" % result.output)
+    print ("return code = %s" % result.exit_code)
+
+  return 0
 
 
 def VerifyCluster(opts, args):
@@ -159,9 +217,76 @@ def VerifyCluster(opts, args):
     opts - class with options as members
 
   """
-  op = opcodes.OpVerifyCluster()
+  skip_checks = []
+  if opts.skip_nplusone_mem:
+    skip_checks.append(constants.VERIFY_NPLUSONE_MEM)
+  op = opcodes.OpVerifyCluster(skip_checks=skip_checks)
+  if SubmitOpCode(op):
+    return 0
+  else:
+    return 1
+
+
+def VerifyDisks(opts, args):
+  """Verify integrity of cluster disks.
+
+  Args:
+    opts - class with options as members
+
+  """
+  op = opcodes.OpVerifyDisks()
   result = SubmitOpCode(op)
-  return result
+  if not isinstance(result, (list, tuple)) or len(result) != 4:
+    raise errors.ProgrammerError("Unknown result type for OpVerifyDisks")
+
+  nodes, nlvm, instances, missing = result
+
+  if nodes:
+    print "Nodes unreachable or with bad data:"
+    for name in nodes:
+      print "\t%s" % name
+  retcode = constants.EXIT_SUCCESS
+
+  if nlvm:
+    for node, text in nlvm.iteritems():
+      print ("Error on node %s: LVM error: %s" %
+             (node, text[-400:].encode('string_escape')))
+      retcode |= 1
+      print "You need to fix these nodes first before fixing instances"
+
+  if instances:
+    for iname in instances:
+      if iname in missing:
+        continue
+      op = opcodes.OpActivateInstanceDisks(instance_name=iname)
+      try:
+        print "Activating disks for instance '%s'" % iname
+        SubmitOpCode(op)
+      except errors.GenericError, err:
+        nret, msg = FormatError(err)
+        retcode |= nret
+        print >> sys.stderr, ("Error activating disks for instance %s: %s" %
+                              (iname, msg))
+
+  if missing:
+    for iname, ival in missing.iteritems():
+      all_missing = utils.all(ival, lambda x: x[0] in nlvm)
+      if all_missing:
+        print ("Instance %s cannot be verified as it lives on"
+               " broken nodes" % iname)
+      else:
+        print "Instance %s has missing logical volumes:" % iname
+        ival.sort()
+        for node, vol in ival:
+          if node in nlvm:
+            print ("\tbroken node %s /dev/xenvg/%s" % (node, vol))
+          else:
+            print ("\t%s /dev/xenvg/%s" % (node, vol))
+    print ("You need to run replace_disks for all the above"
+           " instances, if this message persist after fixing nodes.")
+    retcode |= 1
+
+  return retcode
 
 
 def MasterFailover(opts, args):
@@ -172,16 +297,50 @@ def MasterFailover(opts, args):
   master.
 
   """
-  op = opcodes.OpMasterFailover()
+  return bootstrap.MasterFailover()
+
+
+def SearchTags(opts, args):
+  """Searches the tags on all the cluster.
+
+  """
+  op = opcodes.OpSearchTags(pattern=args[0])
+  result = SubmitOpCode(op)
+  if not result:
+    return 1
+  result = list(result)
+  result.sort()
+  for path, tag in result:
+    print "%s %s" % (path, tag)
+
+
+def SetClusterParams(opts, args):
+  """Modify the cluster.
+
+  Args:
+    opts - class with options as members
+
+  """
+  if not (not opts.lvm_storage or opts.vg_name):
+    print "Please give at least one of the parameters."
+    return 1
+
+  vg_name = opts.vg_name
+  if not opts.lvm_storage and opts.vg_name:
+    print ("Options --no-lvm-storage and --vg-name conflict.")
+    return 1
+
+  op = opcodes.OpSetClusterParams(vg_name=opts.vg_name)
   SubmitOpCode(op)
+  return 0
 
 
 # this is an option common to more than one command, so we declare
 # it here and reuse it
 node_option = make_option("-n", "--node", action="append", dest="nodes",
-                          help="Node to copy to (if not given, all nodes)"
-                          ", can be given multiple times", metavar="<node>",
-                          default=[])
+                          help="Node to copy to (if not given, all nodes),"
+                               " can be given multiple times",
+                          metavar="<node>", default=[])
 
 commands = {
   'init': (InitCluster, ARGS_ONE,
@@ -192,8 +351,12 @@ commands = {
                         " addresses",
                         metavar="ADDRESS", default=None),
             make_option("-t", "--hypervisor-type", dest="hypervisor_type",
-                        help="Specify the hypervisor type (xen-3.0, fake)",
-                        metavar="TYPE", choices=["xen-3.0", "fake"],
+                        help="Specify the hypervisor type "
+                        "(xen-3.0, kvm, fake, xen-hvm-3.1)",
+                        metavar="TYPE", choices=["xen-3.0",
+                                                 "kvm",
+                                                 "fake",
+                                                 "xen-hvm-3.1"],
                         default="xen-3.0",),
             make_option("-m", "--mac-prefix", dest="mac_prefix",
                         help="Specify the mac prefix for the instance IP"
@@ -204,12 +367,29 @@ commands = {
                         help="Specify the volume group name "
                         " (cluster-wide) for disk allocation [xenvg]",
                         metavar="VG",
-                        default="xenvg",),
+                        default=None,),
             make_option("-b", "--bridge", dest="def_bridge",
                         help="Specify the default bridge name (cluster-wide)"
-                        " to connect the instances to [xen-br0]",
+                          " to connect the instances to [%s]" %
+                          constants.DEFAULT_BRIDGE,
                         metavar="BRIDGE",
-                        default="xen-br0",),
+                        default=constants.DEFAULT_BRIDGE,),
+            make_option("--master-netdev", dest="master_netdev",
+                        help="Specify the node interface (cluster-wide)"
+                          " on which the master IP address will be added "
+                          " [%s]" % constants.DEFAULT_BRIDGE,
+                        metavar="NETDEV",
+                        default=constants.DEFAULT_BRIDGE,),
+            make_option("--file-storage-dir", dest="file_storage_dir",
+                        help="Specify the default directory (cluster-wide)"
+                             " for storing the file-based disks [%s]" %
+                             constants.DEFAULT_FILE_STORAGE_DIR,
+                        metavar="DIR",
+                        default=constants.DEFAULT_FILE_STORAGE_DIR,),
+            make_option("--no-lvm-storage", dest="lvm_storage",
+                        help="No support for lvm based instances"
+                             " (cluster-wide)",
+                        action="store_false", default=True,),
             ],
            "[opts...] <cluster_name>",
            "Initialises a new cluster configuration"),
@@ -220,8 +400,18 @@ commands = {
                            action="store_true"),
               ],
               "", "Destroy cluster"),
-  'verify': (VerifyCluster, ARGS_NONE, [DEBUG_OPT],
+  'rename': (RenameCluster, ARGS_ONE, [DEBUG_OPT, FORCE_OPT],
+               "<new_name>",
+               "Renames the cluster"),
+  'verify': (VerifyCluster, ARGS_NONE, [DEBUG_OPT,
+             make_option("--no-nplus1-mem", dest="skip_nplusone_mem",
+                         help="Skip N+1 memory redundancy tests",
+                         action="store_true",
+                         default=False,),
+             ],
              "", "Does a check on the cluster configuration"),
+  'verify-disks': (VerifyDisks, ARGS_NONE, [DEBUG_OPT],
+                   "", "Does a check on the cluster disk status"),
   'masterfailover': (MasterFailover, ARGS_NONE, [DEBUG_OPT],
                      "", "Makes the current node the master"),
   'version': (ShowClusterVersion, ARGS_NONE, [DEBUG_OPT],
@@ -236,8 +426,30 @@ commands = {
               "Runs a command on all (or only some) nodes"),
   'info': (ShowClusterConfig, ARGS_NONE, [DEBUG_OPT],
                  "", "Show cluster configuration"),
+  'list-tags': (ListTags, ARGS_NONE,
+                [DEBUG_OPT], "", "List the tags of the cluster"),
+  'add-tags': (AddTags, ARGS_ANY, [DEBUG_OPT, TAG_SRC_OPT],
+               "tag...", "Add tags to the cluster"),
+  'remove-tags': (RemoveTags, ARGS_ANY, [DEBUG_OPT, TAG_SRC_OPT],
+                  "tag...", "Remove tags from the cluster"),
+  'search-tags': (SearchTags, ARGS_ONE,
+                  [DEBUG_OPT], "", "Searches the tags on all objects on"
+                  " the cluster for a given pattern (regex)"),
+  'modify': (SetClusterParams, ARGS_NONE,
+             [DEBUG_OPT,
+              make_option("-g", "--vg-name", dest="vg_name",
+                          help="Specify the volume group name "
+                          " (cluster-wide) for disk allocation "
+                          "and enable lvm based storage",
+                          metavar="VG",),
+              make_option("--no-lvm-storage", dest="lvm_storage",
+                          help="Disable support for lvm based instances"
+                               " (cluster-wide)",
+                          action="store_false", default=True,),
+              ],
+             "[opts...]",
+             "Alters the parameters of the cluster"),
   }
 
 if __name__ == '__main__':
-  retcode = GenericMain(commands)
-  sys.exit(retcode)
+  sys.exit(GenericMain(commands, override={"tag_type": constants.TAG_CLUSTER}))