Source Code for Module ganeti.watcher.nodemaint

  1  # 
  2  # 
  3   
  4  # Copyright (C) 2006, 2007, 2008, 2009, 2010, 2011 Google Inc. 
  5  # All rights reserved. 
  6  # 
  7  # Redistribution and use in source and binary forms, with or without 
  8  # modification, are permitted provided that the following conditions are 
  9  # met: 
 10  # 
 11  # 1. Redistributions of source code must retain the above copyright notice, 
 12  # this list of conditions and the following disclaimer. 
 13  # 
 14  # 2. Redistributions in binary form must reproduce the above copyright 
 15  # notice, this list of conditions and the following disclaimer in the 
 16  # documentation and/or other materials provided with the distribution. 
 17  # 
 18  # THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS 
 19  # IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 
 20  # TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 
 21  # PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR 
 22  # CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, 
 23  # EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, 
 24  # PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR 
 25  # PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF 
 26  # LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING 
 27  # NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS 
 28  # SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 
 29   
 30   
 31  """Module doing node maintenance for Ganeti watcher. 
 32   
 33  """ 
 34   
 35  import logging 
 36   
 37  from ganeti import constants 
 38  from ganeti import errors 
 39  from ganeti import hypervisor 
 40  from ganeti import netutils 
 41  from ganeti import ssconf 
 42  from ganeti import utils 
 43  from ganeti import confd 
 44  from ganeti.storage import drbd 
 45   
 46  import ganeti.confd.client # pylint: disable=W0611 
 47   
 48   
 49 -class NodeMaintenance(object): 
 50    """Talks to confd daemons and possible shutdown instances/drbd devices. 
 51   
 52    """ 
 53 -  def __init__(self): 
 54      self.store_cb = confd.client.StoreResultCallback() 
 55      self.filter_cb = confd.client.ConfdFilterCallback(self.store_cb) 
 56      self.confd_client = confd.client.GetConfdClient(self.filter_cb) 
 57   
 58    @staticmethod 
 59 -  def ShouldRun(): 
 60      """Checks whether node maintenance should run. 
 61   
 62      """ 
 63      try: 
 64        return ssconf.SimpleStore().GetMaintainNodeHealth() 
 65      except errors.ConfigurationError, err: 
 66        logging.error("Configuration error, not activating node maintenance: %s", 
 67                      err) 
 68        return False 
 69   
 70    @staticmethod 
 71 -  def GetRunningInstances(): 
 72      """Compute list of hypervisor/running instances. 
 73   
 74      """ 
 75      hyp_list = ssconf.SimpleStore().GetHypervisorList() 
 76      hvparams = ssconf.SimpleStore().GetHvparams() 
 77      results = [] 
 78      for hv_name in hyp_list: 
 79        try: 
 80          hv = hypervisor.GetHypervisor(hv_name) 
 81          ilist = hv.ListInstances(hvparams=hvparams) 
 82          results.extend([(iname, hv_name) for iname in ilist]) 
 83        except: # pylint: disable=W0702 
 84          logging.error("Error while listing instances for hypervisor %s", 
 85                        hv_name, exc_info=True) 
 86      return results 
 87   
 88    @staticmethod 
 89 -  def GetUsedDRBDs(): 
 90      """Get list of used DRBD minors. 
 91   
 92      """ 
 93      return drbd.DRBD8.GetUsedDevs() 
 94   
 95    @classmethod 
 96 -  def DoMaintenance(cls, role): 
 97      """Maintain the instance list. 
 98   
 99      """ 
100      if role == constants.CONFD_NODE_ROLE_OFFLINE: 
101        inst_running = cls.GetRunningInstances() 
102        cls.ShutdownInstances(inst_running) 
103        drbd_running = cls.GetUsedDRBDs() 
104        cls.ShutdownDRBD(drbd_running) 
105      else: 
106        logging.debug("Not doing anything for role %s", role) 
107   
108    @staticmethod 
109 -  def ShutdownInstances(inst_running): 
110      """Shutdown running instances. 
111   
112      """ 
113      names_running = set([i[0] for i in inst_running]) 
114      if names_running: 
115        logging.info("Following instances should not be running," 
116                     " shutting them down: %s", utils.CommaJoin(names_running)) 
117        # this dictionary will collapse duplicate instance names (only 
118        # xen pvm/vhm) into a single key, which is fine 
119        i2h = dict(inst_running) 
120        for name in names_running: 
121          hv_name = i2h[name] 
122          hv = hypervisor.GetHypervisor(hv_name) 
123          hv.StopInstance(None, force=True, name=name) 
124   
125    @staticmethod 
126 -  def ShutdownDRBD(drbd_running): 
127      """Shutdown active DRBD devices. 
128   
129      """ 
130      if drbd_running: 
131        logging.info("Following DRBD minors should not be active," 
132                     " shutting them down: %s", utils.CommaJoin(drbd_running)) 
133        for minor in drbd_running: 
134          drbd.DRBD8.ShutdownAll(minor) 
135   
136 -  def Exec(self): 
137      """Check node status versus cluster desired state. 
138   
139      """ 
140      my_name = netutils.Hostname.GetSysName() 
141      req = \ 
142        confd.client.ConfdClientRequest(type=constants.CONFD_REQ_NODE_ROLE_BYNAME, 
143                                        query=my_name) 
144      self.confd_client.SendRequest(req, async=False, coverage=-1) 
145      timed_out, _, _ = self.confd_client.WaitForReply(req.rsalt) 
146      if not timed_out: 
147        # should have a valid response 
148        status, result = self.store_cb.GetResponse(req.rsalt) 
149        assert status, "Missing result but received replies" 
150        if not self.filter_cb.consistent[req.rsalt]: 
151          logging.warning("Inconsistent replies, not doing anything") 
152          return 
153        self.DoMaintenance(result.server_reply.answer) 
154      else: 
155        logging.warning("Confd query timed out, cannot do maintenance actions") 
156