manager: add logging and exception handling to scheduler

circle/manager/scheduler.py

+from logging import getLogger
+
 from django.db.models import Sum
 from django.db.models import Sum
 
 
+logger = getLogger(__name__)
+
 
 
 class NotEnoughMemoryException(Exception):
 class NotEnoughMemoryException(Exception):
 
 
@@ -24,20 +28,25 @@ def select_node(instance, nodes):
     '''
     '''
     # check required traits
     # check required traits
     nodes = [n for n in nodes
     nodes = [n for n in nodes
-             if n.enabled and has_traits(instance.req_traits.all(), n)]
+             if n.enabled and n.online
+             and has_traits(instance.req_traits.all(), n)]
     if not nodes:
     if not nodes:
+        logger.warning('select_node: no usable node for %s', unicode(instance))
         raise TraitsUnsatisfiableException()
         raise TraitsUnsatisfiableException()
 
 
     # check required RAM
     # check required RAM
     nodes = [n for n in nodes if has_enough_ram(instance.ram_size, n)]
     nodes = [n for n in nodes if has_enough_ram(instance.ram_size, n)]
     if not nodes:
     if not nodes:
+        logger.warning('select_node: no enough RAM for %s', unicode(instance))
         raise NotEnoughMemoryException()
         raise NotEnoughMemoryException()
 
 
     # sort nodes first by processor usage, then priority
     # sort nodes first by processor usage, then priority
     nodes.sort(key=lambda n: n.priority, reverse=True)
     nodes.sort(key=lambda n: n.priority, reverse=True)
     nodes.sort(key=free_cpu_time, reverse=True)
     nodes.sort(key=free_cpu_time, reverse=True)
+    result = nodes[0]
 
 
-    return nodes[0]
+    logger.info('select_node: %s for %s', unicode(result), unicode(instance))
+    return result
 
 
 
 
 def has_traits(traits, node):
 def has_traits(traits, node):
@@ -51,6 +60,7 @@ def has_enough_ram(ram_size, node):
     """True, if the node has enough memory to accomodate a guest requiring
     """True, if the node has enough memory to accomodate a guest requiring
        ram_size mebibytes of memory; otherwise, false.
        ram_size mebibytes of memory; otherwise, false.
     """
     """
+    try:
         total = node.ram_size
         total = node.ram_size
         used = (node.ram_usage / 100) * total
         used = (node.ram_usage / 100) * total
         unused = total - used
         unused = total - used
@@ -60,6 +70,10 @@ def has_enough_ram(ram_size, node):
         free = overcommit - reserved
         free = overcommit - reserved
 
 
         return ram_size < unused and ram_size < free
         return ram_size < unused and ram_size < free
+    except TypeError as e:
+        logger.warning('Got incorrect monitoring data for node %s. %s',
+                       unicode(node), unicode(e))
+        return False
 
 
 
 
 def free_cpu_time(node):
 def free_cpu_time(node):
@@ -67,7 +81,12 @@ def free_cpu_time(node):
 
 
     Higher values indicate more idle time.
     Higher values indicate more idle time.
     """
     """
+    try:
         activity = node.cpu_usage / 100
         activity = node.cpu_usage / 100
         inactivity = 1 - activity
         inactivity = 1 - activity
         cores = node.num_cores
         cores = node.num_cores
         return cores * inactivity
         return cores * inactivity
+    except TypeError as e:
+        logger.warning('Got incorrect monitoring data for node %s. %s',
+                       unicode(node), unicode(e))
+        return False  # monitoring data is incorrect