Arnike/watcher
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Restrict existing strategies to their default scope

Browse Source 
Diffrent stratege has diffrent default scope, restrict them to their
default scope will avoid usage problems.
1)workload_balancing/thermal_optimization/airflow_optimization goals
  react on enabled nodes, so restrict default scope to compute nodes
  with up state and enabled status.
2)server_consolidation goal react on enabled or disabled nodes, So
  restrict default scope to compute nodes with up state and
  enabled/disabled status.

Change-Id: I7437dee699ee2d3dd227a047196d4d8db811b81e
Closes-Bug: #1714002
This commit is contained in:
suzhengwei
2017-06-29 14:40:49 +08:00
parent daee2336a4
commit 4cb2b45e3a
6 changed files with 50 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
watcher/decision_engine/strategy/strategies/basic_consolidation.py
View File

@@ -197,6 +197,14 @@ class BasicConsolidation(base.ServerConsolidationBaseStrategy):
def gnocchi(self, gnocchi):
self._gnocchi = gnocchi
def get_available_compute_nodes(self):
default_node_scope = [element.ServiceState.ENABLED.value,
element.ServiceState.DISABLED.value]
return {uuid: cn for uuid, cn in
self.compute_model.get_all_compute_nodes().items()
if cn.state == element.ServiceState.ONLINE.value and
cn.status in default_node_scope}
def check_migration(self, source_node, destination_node,
instance_to_migrate):
"""Check if the migration is possible
@@ -428,7 +436,7 @@ class BasicConsolidation(base.ServerConsolidationBaseStrategy):
def compute_score_of_nodes(self):
"""Calculate score of nodes based on load by VMs"""
score = []
for node in self.compute_model.get_all_compute_nodes().values():
for node in self.get_available_compute_nodes().values():
if node.status == element.ServiceState.ENABLED.value:
self.number_of_enabled_nodes += 1
@@ -502,7 +510,7 @@ class BasicConsolidation(base.ServerConsolidationBaseStrategy):
if not self.compute_model:
raise exception.ClusterStateNotDefined()
if len(self.compute_model.get_all_compute_nodes()) == 0:
if len(self.get_available_compute_nodes()) == 0:
raise exception.ClusterEmpty()
if self.compute_model.stale:

9
watcher/decision_engine/strategy/strategies/outlet_temp_control.py
View File

@@ -171,6 +171,13 @@ class OutletTempControl(base.ThermalOptimizationBaseStrategy):
choices=["ceilometer", "gnocchi"])
]
def get_available_compute_nodes(self):
default_node_scope = [element.ServiceState.ENABLED.value]
return {uuid: cn for uuid, cn in
self.compute_model.get_all_compute_nodes().items()
if cn.state == element.ServiceState.ONLINE.value and
cn.status in default_node_scope}
def calc_used_resource(self, node):
"""Calculate the used vcpus, memory and disk based on VM flavors"""
instances = self.compute_model.get_node_instances(node)
@@ -186,7 +193,7 @@ class OutletTempControl(base.ThermalOptimizationBaseStrategy):
def group_hosts_by_outlet_temp(self):
"""Group hosts based on outlet temp meters"""
nodes = self.compute_model.get_all_compute_nodes()
nodes = self.get_available_compute_nodes()
size_cluster = len(nodes)
if size_cluster == 0:
raise wexc.ClusterEmpty()

9
watcher/decision_engine/strategy/strategies/uniform_airflow.py
View File

@@ -214,6 +214,13 @@ class UniformAirflow(base.BaseStrategy):
choices=["ceilometer", "gnocchi"])
]
def get_available_compute_nodes(self):
default_node_scope = [element.ServiceState.ENABLED.value]
return {uuid: cn for uuid, cn in
self.compute_model.get_all_compute_nodes().items()
if cn.state == element.ServiceState.ONLINE.value and
cn.status in default_node_scope}
def calculate_used_resource(self, node):
"""Compute the used vcpus, memory and disk based on instance flavors"""
instances = self.compute_model.get_node_instances(node)
@@ -334,7 +341,7 @@ class UniformAirflow(base.BaseStrategy):
def group_hosts_by_airflow(self):
"""Group hosts based on airflow meters"""
nodes = self.compute_model.get_all_compute_nodes()
nodes = self.get_available_compute_nodes()
if not nodes:
raise wexc.ClusterEmpty()
overload_hosts = []

20
watcher/decision_engine/strategy/strategies/vm_workload_consolidation.py Executable file → Normal file
View File

@@ -169,6 +169,14 @@ class VMWorkloadConsolidation(base.ServerConsolidationBaseStrategy):
choices=["ceilometer", "gnocchi"])
]
def get_available_compute_nodes(self):
default_node_scope = [element.ServiceState.ENABLED.value,
element.ServiceState.DISABLED.value]
return {uuid: cn for uuid, cn in
self.compute_model.get_all_compute_nodes().items()
if cn.state == element.ServiceState.ONLINE.value and
cn.status in default_node_scope}
def get_instance_state_str(self, instance):
"""Get instance state in string format.
@@ -273,7 +281,7 @@ class VMWorkloadConsolidation(base.ServerConsolidationBaseStrategy):
:return: None
"""
for node in self.compute_model.get_all_compute_nodes().values():
for node in self.get_available_compute_nodes().values():
if (len(self.compute_model.get_node_instances(node)) == 0 and
node.status !=
element.ServiceState.DISABLED.value):
@@ -422,7 +430,7 @@ class VMWorkloadConsolidation(base.ServerConsolidationBaseStrategy):
RCU is an average of relative utilizations (rhu) of active nodes.
:return: {'cpu': <0,1>, 'ram': <0,1>, 'disk': <0,1>}
"""
nodes = self.compute_model.get_all_compute_nodes().values()
nodes = self.get_available_compute_nodes().values()
rcu = {}
counters = {}
for node in nodes:
@@ -534,7 +542,7 @@ class VMWorkloadConsolidation(base.ServerConsolidationBaseStrategy):
:param cc: dictionary containing resource capacity coefficients
"""
sorted_nodes = sorted(
self.compute_model.get_all_compute_nodes().values(),
self.get_available_compute_nodes().values(),
key=lambda x: self.get_node_utilization(x)['cpu'])
for node in reversed(sorted_nodes):
if self.is_overloaded(node, cc):
@@ -567,7 +575,7 @@ class VMWorkloadConsolidation(base.ServerConsolidationBaseStrategy):
:param cc: dictionary containing resource capacity coefficients
"""
sorted_nodes = sorted(
self.compute_model.get_all_compute_nodes().values(),
self.get_available_compute_nodes().values(),
key=lambda x: self.get_node_utilization(x)['cpu'])
asc = 0
for node in sorted_nodes:
@@ -630,7 +638,7 @@ class VMWorkloadConsolidation(base.ServerConsolidationBaseStrategy):
rcu_after = self.get_relative_cluster_utilization()
info = {
"compute_nodes_count": len(
self.compute_model.get_all_compute_nodes()),
self.get_available_compute_nodes()),
'number_of_migrations': self.number_of_migrations,
'number_of_released_nodes':
self.number_of_released_nodes,
@@ -643,7 +651,7 @@ class VMWorkloadConsolidation(base.ServerConsolidationBaseStrategy):
def post_execute(self):
self.solution.set_efficacy_indicators(
compute_nodes_count=len(
self.compute_model.get_all_compute_nodes()),
self.get_available_compute_nodes()),
released_compute_nodes_count=self.number_of_released_nodes,
instance_migrations_count=self.number_of_migrations,
)

9
watcher/decision_engine/strategy/strategies/workload_balance.py
View File

@@ -180,6 +180,13 @@ class WorkloadBalance(base.WorkloadStabilizationBaseStrategy):
choices=["ceilometer", "gnocchi"])
]
def get_available_compute_nodes(self):
default_node_scope = [element.ServiceState.ENABLED.value]
return {uuid: cn for uuid, cn in
self.compute_model.get_all_compute_nodes().items()
if cn.state == element.ServiceState.ONLINE.value and
cn.status in default_node_scope}
def calculate_used_resource(self, node):
"""Calculate the used vcpus, memory and disk based on VM flavors"""
instances = self.compute_model.get_node_instances(node)
@@ -271,7 +278,7 @@ class WorkloadBalance(base.WorkloadStabilizationBaseStrategy):
and also generate the instance workload map.
"""
nodes = self.compute_model.get_all_compute_nodes()
nodes = self.get_available_compute_nodes()
cluster_size = len(nodes)
if not nodes:
raise wexc.ClusterEmpty()

4
watcher/tests/decision_engine/model/data/scenario_1_with_metrics.xml
View File

@@ -1,8 +1,8 @@
<ModelRoot>
<ComputeNode hostname="hostname_0" uuid="Node_0" id="0" state="enabled" human_id="" status="enabled" vcpus="40" disk="250" disk_capacity="250" memory="64">
<ComputeNode hostname="hostname_0" uuid="Node_0" id="0" state="up" human_id="" status="enabled" vcpus="40" disk="250" disk_capacity="250" memory="64">
<Instance human_id="" state="active" uuid="INSTANCE_0" vcpus="10" disk="20" disk_capacity="20" memory="2" metadata='{"optimize": true,"top": "floor", "nested": {"x": "y"}}'/>
</ComputeNode>
<ComputeNode hostname="hostname_1" uuid="Node_1" id="1" state="enabled" human_id="" status="enabled" vcpus="40" disk="250" disk_capacity="250" memory="64">
<ComputeNode hostname="hostname_1" uuid="Node_1" id="1" state="up" human_id="" status="enabled" vcpus="40" disk="250" disk_capacity="250" memory="64">
<Instance human_id="" state="active" uuid="INSTANCE_1" vcpus="10" disk="20" disk_capacity="20" memory="2" metadata='{"optimize": true,"top": "floor", "nested": {"x": "y"}}'/>
</ComputeNode>
</ModelRoot>