Set "disabled reason" for compute service.

Masakari never sets reason why compute service was disabled.
"disabled reason" was added in config.

Closes-Bug: 1936181
Change-Id: I998f7884195b93927773c7186d61c13670a53662

masakari/conf/engine_driver.py

@@ -27,6 +27,11 @@ host_recovery_group = cfg.OptGroup(
     title='Host failure recovery options',
     help="Configuration options for host failure recovery")
 
+process_recovery_group = cfg.OptGroup(
+    'process_failure',
+    title='Process failure recovery options',
+    help="Configuration options for process failure recovery")
+
 customized_recovery_flow_group = cfg.OptGroup(
     'taskflow_driver_recovery_flows',
     title='Customized recovery flow Options',
@@ -80,6 +85,10 @@ Operators can decide whether reserved_host should be added to aggregate group
 of failed compute host. When set to True, reserved host will be added to the
 aggregate group of failed compute host. When set to False, the reserved_host
 will not be added to the aggregate group of failed compute host."""),
+    cfg.StrOpt("service_disable_reason",
+               default="Masakari detected host failed.",
+               help="Compute disable reason in case Masakari detects host "
+                    "failure."),
 ]
 
 instance_failure_options = [
@@ -220,14 +229,23 @@ The allowed values for this option is comma separated dictionary of object
 names in between ``{`` and ``}``."""))
 ]
 
+process_failure_opts = [
+    cfg.StrOpt("service_disable_reason",
+               default="Masakari detected process failed.",
+               help="Compute disable reason in case Masakari detects process "
+                    "failure."),
+]
+
 
 def register_opts(conf):
     conf.register_group(instance_recovery_group)
     conf.register_group(host_recovery_group)
+    conf.register_group(process_recovery_group)
     conf.register_group(customized_recovery_flow_group)
     conf.register_group(taskflow_group)
     conf.register_opts(instance_failure_options, group=instance_recovery_group)
     conf.register_opts(host_failure_opts, group=host_recovery_group)
+    conf.register_opts(process_failure_opts, group=process_recovery_group)
     conf.register_opts(taskflow_driver_recovery_flows,
                        group=customized_recovery_flow_group)
     conf.register_opts(taskflow_options, group=taskflow_group)
@@ -237,6 +255,7 @@ def list_opts():
     return {
         instance_recovery_group.name: instance_failure_options,
         host_recovery_group.name: host_failure_opts,
+        process_recovery_group.name: process_failure_opts,
         taskflow_group.name: taskflow_options
     }
 

masakari/engine/drivers/taskflow/host_failure.py

@@ -48,7 +48,8 @@ class DisableComputeServiceTask(base.MasakariTask):
     def execute(self, host_name):
         msg = "Disabling compute service on host: '%s'" % host_name
         self.update_details(msg)
-        self.novaclient.enable_disable_service(self.context, host_name)
+        self.novaclient.enable_disable_service(self.context, host_name,
+            reason=CONF.host_failure.service_disable_reason)
         # Sleep until nova-compute service is marked as disabled.
         log_msg = ("Sleeping %(wait)s sec before starting recovery "
                "thread until nova recognizes the node down.")

masakari/engine/drivers/taskflow/process_failure.py

@@ -45,7 +45,8 @@ class DisableComputeNodeTask(base.MasakariTask):
         if not self.novaclient.is_service_disabled(self.context, host_name,
                                                    process_name):
             # disable compute node on given host
-            self.novaclient.enable_disable_service(self.context, host_name)
+            self.novaclient.enable_disable_service(self.context, host_name,
+                reason=CONF.process_failure.service_disable_reason)
             msg = "Disabled compute service on host: '%s'" % host_name
             self.update_details(msg, 1.0)
         else:

masakari/tests/unit/engine/drivers/taskflow/test_host_failure_flow.py

@@ -52,6 +52,7 @@ class HostFailureTestCase(test.TestCase):
         self.instance_host = "fake-host"
         self.novaclient = nova.API()
         self.fake_client = fakes.FakeNovaClient()
+        self.disabled_reason = CONF.host_failure.service_disable_reason
 
     def _verify_instance_evacuated(self, old_instance_list):
         for server in old_instance_list:
@@ -86,7 +87,7 @@ class HostFailureTestCase(test.TestCase):
         task.execute(self.instance_host)
 
         mock_enable_disable.assert_called_once_with(
-            self.ctxt, self.instance_host)
+            self.ctxt, self.instance_host, reason=self.disabled_reason)
 
     def _test_instance_list(self, instances_evacuation_count):
         task = host_failure.PrepareHAEnabledInstancesTask(self.ctxt,

masakari/tests/unit/engine/drivers/taskflow/test_process_failure_flow.py

@@ -20,12 +20,15 @@ Unit Tests for process failure TaskFlow
 from unittest import mock
 
 from masakari.compute import nova
+from masakari import conf
 from masakari import context
 from masakari.engine.drivers.taskflow import process_failure
 from masakari import exception
 from masakari import test
 from masakari.tests.unit import fakes
 
+CONF = conf.CONF
+
 
 class ProcessFailureTestCase(test.TestCase):
 
@@ -39,6 +42,7 @@ class ProcessFailureTestCase(test.TestCase):
         # overriding 'wait_period_after_service_update' to 2 seconds
         # to reduce the wait period.
         self.override_config('wait_period_after_service_update', 2)
+        self.disabled_reason = CONF.process_failure.service_disable_reason
 
     @mock.patch('masakari.compute.nova.novaclient')
     @mock.patch('masakari.engine.drivers.taskflow.base.MasakariTask.'

masakari/tests/unit/fakes.py

@@ -172,6 +172,13 @@ class FakeNovaClient(object):
                     services.append(service)
             return services
 
+        def disable_log_reason(self, service_id, reason):
+            for _service in self._services:
+                if _service.id == service_id:
+                    service = _service
+            service.status = 'disabled'
+            service.disabled_reason = reason
+
     def __init__(self):
         self.servers = FakeNovaClient.ServerManager()
         self.services = FakeNovaClient.Services()

releasenotes/notes/compute-disable-reason-9570734c0bb888cf.yaml

@@ -0,0 +1,6 @@
+---
+features:
+  - |
+    Nova compute service "disable reason" is now set
+    in case of host or process failure.
+    It can be customised per type of failure via config.