Prevent errors when xenvg is broken cluster verify

[ganeti-local] / lib / backend.py

diff --git a/lib/backend.py b/lib/backend.py
index 48d0dc8..18d439e 100644 (file)
--- a/lib/backend.py
+++ b/lib/backend.py
@@ -437,9 +437,9 @@ def VerifyNode(what, cluster_name):
   if constants.NV_DRBDLIST in what:
     try:
       used_minors = bdev.DRBD8.GetUsedDevs().keys()
-    except errors.BlockDeviceError:
+    except errors.BlockDeviceError, err:
       logging.warning("Can't get used minors list", exc_info=True)
-      used_minors = []
+      used_minors = str(err)
     result[constants.NV_DRBDLIST] = used_minors
 
   return result
@@ -575,7 +575,6 @@ def GetInstanceList(hypervisor_list):
       results.extend(names)
     except errors.HypervisorError, err:
       logging.exception("Error enumerating instances for hypevisor %s", hname)
-      # FIXME: should we somehow not propagate this to the master?
       raise
 
   return results
@@ -661,9 +660,14 @@ def GetAllInstancesInfo(hypervisor_list):
           'state': state,
           'time': times,
           }
-        if name in output and output[name] != value:
-          raise errors.HypervisorError("Instance %s running duplicate"
-                                       " with different parameters" % name)
+        if name in output:
+          # we only check static parameters, like memory and vcpus,
+          # and not state and time which can change between the
+          # invocations of the different hypervisors
+          for key in 'memory', 'vcpus':
+            if value[key] != output[name][key]:
+              raise errors.HypervisorError("Instance %s is running twice"
+                                           " with different parameters" % name)
         output[name] = value
 
   return output
@@ -857,7 +861,7 @@ def _GatherAndLinkBlockDevs(instance):
   return block_devices
 
 
-def StartInstance(instance, extra_args):
+def StartInstance(instance):
   """Start an instance.
 
   @type instance: L{objects.Instance}
@@ -874,7 +878,7 @@ def StartInstance(instance, extra_args):
   try:
     block_devices = _GatherAndLinkBlockDevs(instance)
     hyper = hypervisor.GetHypervisor(instance.hypervisor)
-    hyper.StartInstance(instance, block_devices, extra_args)
+    hyper.StartInstance(instance, block_devices)
   except errors.BlockDeviceError, err:
     logging.exception("Failed to start instance")
     return (False, "Block device error: %s" % str(err))
@@ -886,7 +890,7 @@ def StartInstance(instance, extra_args):
   return (True, "Instance started successfully")
 
 
-def ShutdownInstance(instance):
+def InstanceShutdown(instance):
   """Shut an instance down.
 
   @note: this functions uses polling with a hardcoded timeout.
@@ -901,14 +905,15 @@ def ShutdownInstance(instance):
   running_instances = GetInstanceList([hv_name])
 
   if instance.name not in running_instances:
-    return True
+    return (True, "Instance already stopped")
 
   hyper = hypervisor.GetHypervisor(hv_name)
   try:
     hyper.StopInstance(instance)
   except errors.HypervisorError, err:
-    logging.error("Failed to stop instance: %s" % err)
-    return False
+    msg = "Failed to stop instance %s: %s" % (instance.name, err)
+    logging.error(msg)
+    return (False, msg)
 
   # test every 10secs for 2min
 
@@ -925,21 +930,23 @@ def ShutdownInstance(instance):
     try:
       hyper.StopInstance(instance, force=True)
     except errors.HypervisorError, err:
-      logging.exception("Failed to stop instance: %s" % err)
-      return False
+      msg = "Failed to force stop instance %s: %s" % (instance.name, err)
+      logging.error(msg)
+      return (False, msg)
 
     time.sleep(1)
     if instance.name in GetInstanceList([hv_name]):
-      logging.error("Could not shutdown instance '%s' even by destroy",
-                    instance.name)
-      return False
+      msg = ("Could not shutdown instance %s even by destroy" %
+             instance.name)
+      logging.error(msg)
+      return (False, msg)
 
   _RemoveBlockDevLinks(instance.name, instance.disks)
 
-  return True
+  return (True, "Instance has been shutdown successfully")
 
 
-def InstanceReboot(instance, reboot_type, extra_args):
+def InstanceReboot(instance, reboot_type):
   """Reboot an instance.
 
   @type instance: L{objects.Instance}
@@ -975,8 +982,10 @@ def InstanceReboot(instance, reboot_type, extra_args):
       return (False, msg)
   elif reboot_type == constants.INSTANCE_REBOOT_HARD:
     try:
-      ShutdownInstance(instance)
-      StartInstance(instance, extra_args)
+      stop_result = InstanceShutdown(instance)
+      if not stop_result[0]:
+        return stop_result
+      return StartInstance(instance)
     except errors.HypervisorError, err:
       msg = "Failed to hard reboot instance %s: %s" % (instance.name, err)
       logging.error(msg)
@@ -1663,7 +1672,6 @@ def OSEnvironment(instance, debug=0):
                                     str(disk))
     real_disk.Open()
     result['DISK_%d_PATH' % idx] = real_disk.dev_path
-    # FIXME: When disks will have read-only mode, populate this
     result['DISK_%d_ACCESS' % idx] = disk.mode
     if constants.HV_DISK_TYPE in instance.hvparams:
       result['DISK_%d_FRONTEND_TYPE' % idx] = \