Merge branch 'issue-102-rebased' into 'master'

Fix node scalability #102

circle/dashboard/templates/dashboard/node-list/column-vm.html

 {% load i18n %} 
 
 <div id="node-list-column-vm">
-  <a class="real-link" href="{% url "dashboard.views.node-detail" pk=record.pk %}#virtualmachines">{{ record.instance_set.count }}</a>
+  <a class="real-link" href="{% url "dashboard.views.node-detail" pk=record.pk %}#virtualmachines">{{ value }}</a>
 </div>

circle/dashboard/views.py

@@ -14,6 +14,7 @@ from django.core.exceptions import (
 )
 from django.core import signing
 from django.core.urlresolvers import reverse, reverse_lazy
+from django.db.models import Count
 from django.http import HttpResponse, HttpResponseRedirect, Http404
 from django.shortcuts import redirect, render, get_object_or_404
 from django.views.decorators.http import require_GET
@@ -921,10 +922,13 @@ class VmList(LoginRequiredMixin, SingleTableView):
 
 class NodeList(LoginRequiredMixin, SuperuserRequiredMixin, SingleTableView):
     template_name = "dashboard/node-list.html"
-    model = Node
     table_class = NodeListTable
     table_pagination = False
 
+    def get_queryset(self):
+        return Node.objects.annotate(
+            number_of_VMs=Count('instance_set')).select_related('host')
+
 
 class GroupList(LoginRequiredMixin, SuperuserRequiredMixin, SingleTableView):
     template_name = "dashboard/group-list.html"

circle/vm/models/node.py

@@ -25,6 +25,17 @@ from django.utils import timezone
 logger = getLogger(__name__)
 
 
+def node_available(function):
+    """Decorate methods to ignore disabled Nodes.
+    """
+    def decorate(self, *args, **kwargs):
+        if self.enabled and self.online:
+            return function(self, *args, **kwargs)
+        else:
+            return None
+    return decorate
+
+
 class Node(TimeStampedModel):
 
     """A VM host machine, a hypervisor.
@@ -55,19 +66,22 @@ class Node(TimeStampedModel):
     def __unicode__(self):
         return self.name
 
-    @method_cache(10, 5)
+    @method_cache(10)
     def get_online(self):
         """Check if the node is online.
 
-        Runs a remote ping task if the worker is running.
+        Check if node is online by queue is available.
         """
         try:
-            return self.remote_query(vm_tasks.ping, timeout=1, default=False)
-        except WorkerNotFound:
+            self.get_remote_queue_name("vm")
+        except:
             return False
+        else:
+            return True
 
     online = property(get_online)
 
+    @node_available
     @method_cache(300)
     def get_num_cores(self):
         """Number of CPU threads available to the virtual machines.
@@ -106,6 +120,7 @@ class Node(TimeStampedModel):
             self.get_num_cores(invalidate_cache=True)
             self.get_ram_size(invalidate_cache=True)
 
+    @node_available
     @method_cache(300)
     def get_ram_size(self):
         """Bytes of total memory in the node.
@@ -115,6 +130,7 @@ class Node(TimeStampedModel):
     ram_size = property(get_ram_size)
 
     @property
+    @node_available
     def ram_size_with_overcommit(self):
         """Bytes of total memory including overcommit margin.
         """
@@ -198,6 +214,7 @@ class Node(TimeStampedModel):
             else:
                 return default
 
+    @node_available
     def get_monitor_info(self):
         try:
             handler = GraphiteHandler()
@@ -229,17 +246,21 @@ class Node(TimeStampedModel):
         return collected
 
     @property
+    @node_available
     def cpu_usage(self):
         return float(self.get_monitor_info()["cpu.usage"]) / 100
 
     @property
+    @node_available
     def ram_usage(self):
         return float(self.get_monitor_info()["memory.usage"]) / 100
 
     @property
+    @node_available
     def byte_ram_usage(self):
         return self.ram_usage * self.ram_size
 
+    @node_available
     def update_vm_states(self):
         """Update state of Instances running on this Node.
 
@@ -284,7 +305,8 @@ class Node(TimeStampedModel):
 
     @classmethod
     def get_state_count(cls, online, enabled):
-        return len([1 for i in cls.objects.filter(enabled=enabled).all()
+        return len([1 for i in
+                    cls.objects.filter(enabled=enabled).select_related('host')
                     if i.online == online])
 
     @permalink

circle/vm/tasks/vm_tasks.py

+from django.core.cache import cache
+from logging import getLogger
+
 from manager.mancelery import celery
 
+logger = getLogger(__name__)
+
 
 def check_queue(node_hostname, queue_id):
-    drivers = ['vmdriver', 'netdriver', 'agentdriver']
-    worker_list = [node_hostname + "." + d for d in drivers]
+    """True if the queue is alive.
+
+    Example: check_queue('node01', 'vm'):
+    :param node_hostname: Short hostname of the node.
+    :param queue_id: Queue identifier (eg. vm).
+    """
+    # drivers = ['vmdriver', 'netdriver', 'agentdriver']
+    # worker_list = [node_hostname + "." + d for d in drivers]
     queue_name = node_hostname + "." + queue_id
-    inspect = celery.control.inspect(worker_list)
-    active_queues = inspect.active_queues()
+    active_queues = get_queues()
     if active_queues is None:
         return False
     # v is List of List of queues dict
@@ -18,6 +28,22 @@ def check_queue(node_hostname, queue_id):
     return False
 
 
+def get_queues():
+    """Get active celery queues.
+
+    Result is cached for 10 seconds!
+    """
+    key = __name__ + u'queues'
+    result = cache.get(key)
+    if result is None:
+        inspect = celery.control.inspect()
+        inspect.timeout = 0.1
+        result = inspect.active_queues()
+        logger.debug('Queue list of length %d cached.', len(result))
+        cache.set(key, result, 10)
+    return result
+
+
 @celery.task(name='vmdriver.create')
 def deploy(params):
     pass