openstack
/
trove
Code Issues Proposed changes

Merge "Set status to ERROR if heartbeat expires"

This commit is contained in:
Zuul 2020-06-04 02:57:23 +00:00 committed by Gerrit Code Review
parent 816a7bca58 dc117d8dd8
commit 3202b321b3
29 changed files with 426 additions and 313 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
.zuul.yaml
View File

@ -14,14 +14,16 @@
- openstack-tox-pylint
- trove-tox-bandit-baseline:
voting: false
- trove-tempest
- trove-tempest:
voting: false
- trove-tempest-ipv6-only:
voting: false
gate:
queue: trove
jobs:
- openstack-tox-pylint
- trove-tempest
- trove-tempest:
voting: false
experimental:
jobs:
- trove-functional-mysql

2
doc/source/user/index.rst
View File

@ -17,6 +17,6 @@ handling complex administrative tasks.
backup-db-incremental.rst
manage-db-config.rst
set-up-replication.rst
set-up-clustering.rst
upgrade-datastore.rst
set-up-clustering.rst
upgrade-cluster-datastore.rst

5
doc/source/user/upgrade-cluster-datastore.rst
View File

@ -2,6 +2,11 @@
Upgrade cluster datastore
=========================
.. caution::
Database clustering function is still in experimental, should not be used
in production environment.
Upgrading datastore for cluster instances is very similar to upgrading
a single instance.

119
doc/source/user/upgrade-datastore.rst
View File

@ -8,58 +8,71 @@ configuration files of your database.
To perform datastore upgrade, you need:
- A Trove database instance to be upgrade.
- A guest image with the target datastore version.
- A Trove database instance to be upgrade.
This guide shows you how to upgrade MySQL datastore from 5.7.29 to 5.7.30 for a
database instance.
This example shows you how to upgrade Redis datastore (version 3.2.6)
for a single instance database.
.. warning::
.. note::
**Before** upgrading, make sure that:
- Your target datastore is binary compatible with the current
datastore. Each database provider has its own compatibilty
policy. Usually there shouldn't be any problem when
performing an upgrade within minor versions.
- You **do not** downgrade your datastore.
- Target versions is supported by Trove. For instance, Trove
doesn't support Cassandra >=2.2 at this moment so you
shouldn't perform an upgrade from 2.1 to 2.2.
Datastore upgrade could cause downtime of the database service.
Upgrading datastore
~~~~~~~~~~~~~~~~~~~
#. **Check instance status**
#. **Check datastore versions in the system**
In my environment, both datastore version 5.7.29 and 5.7.30 are defined for
MySQL.
.. code-block:: console
$ openstack datastore list
+--------------------------------------+-------+
| ID | Name |
+--------------------------------------+-------+
| 50bed39d-6788-4a0d-8d74-321012bb6b55 | mysql |
+--------------------------------------+-------+
$ openstack datastore version list mysql
+--------------------------------------+--------+
| ID | Name |
+--------------------------------------+--------+
| 70c68d0a-27e1-4fbd-bd3b-f29d42ce1a7d | 5.7.29 |
| cf91aa9a-2192-4ec4-b7ce-5cac3b1e7dbe | 5.7.30 |
+--------------------------------------+--------+
#. **Create a new instance with datastore version 5.7.29**
Make sure the instance status is HEALTHY before upgrading.
.. code-block:: console
$ openstack database instance create test-mysql-upgrade \
d2 \
--size 1 \
--nic net-id=$netid \
--datastore mysql --datastore_version 5.7.29 \
--databases testdb --users user:password
$ openstack database instance list
+--------------------------------------+------------+-----------+-------------------+---------+-----------+-----------+------+-----------+
| ID | Name | Datastore | Datastore Version | Status | Addresses | Flavor ID | Size | Region |
+--------------------------------------+------------+-----------+-------------------+---------+-----------+-----------+------+-----------+
| 55411e95-1670-497f-8d92-0179f3b4fdd4 | redis_test | redis | 3.2.6 | HEALTHY | 10.1.0.25 | 6 | 1 | RegionOne |
+--------------------------------------+------------+-----------+-------------------+---------+-----------+-----------+------+-----------+
+--------------------------------------+--------------------+-----------+-------------------+---------+-----------------------------------------------+-----------+------+-----------+---------+
| ID | Name | Datastore | Datastore Version | Status | Addresses | Flavor ID | Size | Region | Role |
+--------------------------------------+--------------------+-----------+-------------------+---------+-----------------------------------------------+-----------+------+-----------+---------+
| 32eb56b0-d10d-43e9-b59e-1e4b0979e5dd | test-mysql-upgrade | mysql | 5.7.29 | HEALTHY | [{'address': '10.0.0.54', 'type': 'private'}] | d2 | 1 | RegionOne | |
+--------------------------------------+--------------------+-----------+-------------------+---------+-----------------------------------------------+-----------+------+-----------+---------+
#. **Check if target version is available**
Use :command:`openstack datastore version list` command to list
all available versions your datastore.
Check the MySQL version by connecting with the database:
.. code-block:: console
$ openstack datastore version list redis
+--------------------------------------+-------+
| ID | Name |
+--------------------------------------+-------+
| 483debec-b7c3-4167-ab1d-1765795ed7eb | 3.2.6 |
| 507f666e-193c-4194-9d9d-da8342dcb4f1 | 3.2.7 |
+--------------------------------------+-------+
$ ip=10.0.0.54
$ mysql -u user -ppassword -h $ip testdb
mysql> SELECT @@GLOBAL.innodb_version;
+-------------------------+
| @@GLOBAL.innodb_version |
+-------------------------+
| 5.7.29 |
+-------------------------+
#. **Run upgrade**
@ -68,7 +81,7 @@ Upgrading datastore
.. code-block:: console
$ openstack database instance upgrade 55411e95-1670-497f-8d92-0179f3b4fdd4 3.2.7
$ openstack database instance upgrade 32eb56b0-d10d-43e9-b59e-1e4b0979e5dd cf91aa9a-2192-4ec4-b7ce-5cac3b1e7dbe
#. **Wait until status changes from UPGRADE to HEALTHY**
@ -78,24 +91,26 @@ Upgrading datastore
.. code-block:: console
$ openstack database instance list
+--------------------------------------+------------+-----------+-------------------+---------+-----------+-----------+------+-----------+
| ID | Name | Datastore | Datastore Version | Status | Addresses | Flavor ID | Size | Region |
+--------------------------------------+------------+-----------+-------------------+---------+-----------+-----------+------+-----------+
| 55411e95-1670-497f-8d92-0179f3b4fdd4 | redis_test | redis | 3.2.7 | UPGRADE | 10.1.0.25 | 6 | 5 | RegionOne |
+--------------------------------------+------------+-----------+-------------------+---------+-----------+-----------+------+-----------+
+--------------------------------------+--------------------+-----------+-------------------+---------+-----------------------------------------------+-----------+------+-----------+---------+
| ID | Name | Datastore | Datastore Version | Status | Addresses | Flavor ID | Size | Region | Role |
+--------------------------------------+--------------------+-----------+-------------------+---------+-----------------------------------------------+-----------+------+-----------+---------+
| 32eb56b0-d10d-43e9-b59e-1e4b0979e5dd | test-mysql-upgrade | mysql | 5.7.30 | UPGRADE | [{'address': '10.0.0.54', 'type': 'private'}] | d2 | 1 | RegionOne | |
+--------------------------------------+--------------------+-----------+-------------------+---------+-----------------------------------------------+-----------+------+-----------+---------+
$ openstack database instance list
+--------------------------------------+------------+-----------+-------------------+---------+-----------+-----------+------+-----------+
| ID | Name | Datastore | Datastore Version | Status | Addresses | Flavor ID | Size | Region |
+--------------------------------------+------------+-----------+-------------------+---------+-----------+-----------+------+-----------+
| 55411e95-1670-497f-8d92-0179f3b4fdd4 | redis_test | redis | 3.2.7 | HEALTHY | 10.1.0.25 | 6 | 5 | RegionOne |
+--------------------------------------+------------+-----------+-------------------+---------+-----------+-----------+------+-----------+
+--------------------------------------+--------------------+-----------+-------------------+---------+-----------------------------------------------+-----------+------+-----------+---------+
| ID | Name | Datastore | Datastore Version | Status | Addresses | Flavor ID | Size | Region | Role |
+--------------------------------------+--------------------+-----------+-------------------+---------+-----------------------------------------------+-----------+------+-----------+---------+
| 32eb56b0-d10d-43e9-b59e-1e4b0979e5dd | test-mysql-upgrade | mysql | 5.7.30 | HEALTHY | [{'address': '10.0.0.54', 'type': 'private'}] | d2 | 1 | RegionOne | |
+--------------------------------------+--------------------+-----------+-------------------+---------+-----------------------------------------------+-----------+------+-----------+---------+
Other datastores
~~~~~~~~~~~~~~~~
Check the MySQL version again:
Upgrade for other datastores works in the same way. Currently Trove
supports upgrades for the following datastores:
.. code-block:: console
- MySQL
- MariaDB
- Redis
$ mysql -u user -ppassword -h $ip testdb
mysql> SELECT @@GLOBAL.innodb_version;
+-------------------------+
| @@GLOBAL.innodb_version |
+-------------------------+
| 5.7.30 |
+-------------------------+

4
releasenotes/notes/victoria-expired-database-status.yaml Normal file
View File

@ -0,0 +1,4 @@
---
fixes:
- When the trove-guestagent failed to update the datastore service status,
the instance status should be ERROR.

12
tools/trove-pylint.config
View File

@ -1239,6 +1239,12 @@
"Instance of 'DBInstance' has no 'encrypted_key' member",
"DBInstance.key"
],
[
"trove/instance/models.py",
"E1101",
"Instance of 'InstanceServiceStatus' has no 'updated_at' member",
"InstanceServiceStatus.is_uptodate"
],
[
"trove/instance/models.py",
"no-member",
@ -1323,6 +1329,12 @@
"Instance of 'DBInstance' has no 'encrypted_key' member",
"DBInstance.key"
],
[
"trove/instance/models.py",
"no-member",
"Instance of 'InstanceServiceStatus' has no 'updated_at' member",
"InstanceServiceStatus.is_uptodate"
],
[
"trove/instance/service.py",
"E1101",

11
trove/common/strategies/cluster/experimental/mongodb/taskmanager.py
View File

@ -17,17 +17,16 @@ from eventlet.timeout import Timeout
from oslo_log import log as logging
from trove.common import cfg
from trove.common.exception import PollTimeOut
from trove.common.instance import ServiceStatuses
from trove.common.strategies.cluster import base
from trove.common import utils
from trove.common.exception import PollTimeOut
from trove.common.strategies.cluster import base
from trove.instance import models
from trove.instance import tasks as inst_tasks
from trove.instance.models import DBInstance
from trove.instance.models import Instance
from trove.instance import tasks as inst_tasks
from trove.instance.service_status import ServiceStatuses
from trove.taskmanager import api as task_api
import trove.taskmanager.models as task_models
from trove.taskmanager import models as task_models
LOG = logging.getLogger(__name__)
CONF = cfg.CONF

6
trove/conductor/manager.py
View File

@ -19,12 +19,12 @@ from oslo_service import periodic_task
from trove.backup import models as bkup_models
from trove.common import cfg
from trove.common import exception as trove_exception
from trove.common.instance import ServiceStatus
from trove.common.rpc import version as rpc_version
from trove.common.serializable_notification import SerializableNotification
from trove.conductor.models import LastSeen
from trove.extensions.mysql import models as mysql_models
from trove.instance import models as inst_models
from trove.instance.service_status import ServiceStatus
LOG = logging.getLogger(__name__)
CONF = cfg.CONF
@ -89,8 +89,8 @@ class Manager(periodic_task.PeriodicTasks):
if self._message_too_old(instance_id, 'heartbeat', sent):
return
if payload.get('service_status') is not None:
status.set_status(ServiceStatus.from_description(
payload['service_status']))
status.set_status(
ServiceStatus.from_description(payload['service_status']))
status.save()
def update_backup(self, context, instance_id, backup_id,

18
trove/guestagent/api.py
View File

@ -380,6 +380,14 @@ class API(object):
self.agent_high_timeout, version=version,
upgrade_info=upgrade_info)
def upgrade(self, upgrade_info):
"""Upgrade database service."""
LOG.debug("Sending the call to upgrade database service.")
version = self.API_BASE_VERSION
return self._cast("upgrade", version=version,
upgrade_info=upgrade_info)
def restart(self):
"""Restart the database server."""
LOG.debug("Sending the call to restart the database process "
@ -419,16 +427,6 @@ class API(object):
self._call("stop_db", self.agent_low_timeout,
version=version)
def upgrade(self, instance_version, location, metadata=None):
"""Make an asynchronous call to self upgrade the guest agent."""
LOG.debug("Sending an upgrade call to nova-guest.")
version = self.API_BASE_VERSION
self._cast("upgrade", version=version,
instance_version=instance_version,
location=location,
metadata=metadata)
def get_volume_info(self):
"""Make a synchronous call to get volume info for the container."""
LOG.debug("Check Volume Info on instance %s.", self.id)

9
trove/guestagent/datastore/manager.py
View File

@ -25,7 +25,6 @@ from oslo_service import periodic_task
from trove.common import cfg
from trove.common import exception
from trove.common import instance
from trove.common.i18n import _
from trove.common.notification import EndNotification
from trove.guestagent import dbaas
@ -37,6 +36,7 @@ from trove.guestagent.common.operating_system import FileMode
from trove.guestagent.module import driver_manager
from trove.guestagent.module import module_manager
from trove.guestagent.strategies import replication as repl_strategy
from trove.instance import service_status
LOG = logging.getLogger(__name__)
CONF = cfg.CONF
@ -306,6 +306,10 @@ class Manager(periodic_task.PeriodicTasks):
"""
return {}
def upgrade(self, context, upgrade_info):
"""Upgrade the database."""
pass
def post_upgrade(self, context, upgrade_info):
"""Recovers the guest after the image is upgraded using information
from the pre_upgrade step
@ -588,7 +592,8 @@ class Manager(periodic_task.PeriodicTasks):
self.configuration_manager.apply_system_override(
config_man_values, change_id=apply_label, pre_user=True)
if restart_required:
self.status.set_status(instance.ServiceStatuses.RESTART_REQUIRED)
self.status.set_status(
service_status.ServiceStatuses.RESTART_REQUIRED)
else:
self.apply_overrides(context, cfg_values)

14
trove/guestagent/datastore/mysql_common/manager.py
View File

@ -22,7 +22,6 @@ from oslo_log import log as logging
from trove.common import cfg
from trove.common import configurations
from trove.common import exception
from trove.common import instance as rd_instance
from trove.common import utils
from trove.common.notification import EndNotification
from trove.guestagent import guest_log
@ -32,6 +31,7 @@ from trove.guestagent.datastore import manager
from trove.guestagent.strategies import replication as repl_strategy
from trove.guestagent.utils import docker as docker_util
from trove.guestagent.utils import mysql as mysql_util
from trove.instance import service_status
LOG = logging.getLogger(__name__)
CONF = cfg.CONF
@ -71,7 +71,7 @@ class MySqlManager(manager.Manager):
client.execute(cmd)
LOG.debug("Database service check: database query is responsive")
return rd_instance.ServiceStatuses.HEALTHY
return service_status.ServiceStatuses.HEALTHY
except Exception:
return super(MySqlManager, self).get_service_status()
@ -295,7 +295,7 @@ class MySqlManager(manager.Manager):
self.app.restore_backup(context, backup_info, restore_location)
except Exception:
LOG.error("Failed to restore from backup %s.", backup_info['id'])
self.status.set_status(rd_instance.ServiceStatuses.FAILED)
self.status.set_status(service_status.ServiceStatuses.FAILED)
raise
LOG.info("Finished restore data from backup %s", backup_info['id'])
@ -365,7 +365,7 @@ class MySqlManager(manager.Manager):
slave_config)
except Exception as err:
LOG.error("Error enabling replication, error: %s", str(err))
self.status.set_status(rd_instance.ServiceStatuses.FAILED)
self.status.set_status(service_status.ServiceStatuses.FAILED)
raise
def detach_replica(self, context, for_failover=False):
@ -431,3 +431,9 @@ class MySqlManager(manager.Manager):
def demote_replication_master(self, context):
LOG.info("Demoting replication master.")
self.replication.demote_master(self.app)
def upgrade(self, context, upgrade_info):
"""Upgrade the database."""
LOG.info('Starting to upgrade database, upgrade_info: %s',
upgrade_info)
self.app.upgrade(upgrade_info)

41
trove/guestagent/datastore/mysql_common/service.py
View File

@ -27,7 +27,6 @@ from sqlalchemy.sql.expression import text
from trove.backup.state import BackupState
from trove.common import cfg
from trove.common import exception
from trove.common import instance
from trove.common import utils
from trove.common.configurations import MySQLConfParser
from trove.common.db.mysql import models
@ -43,6 +42,7 @@ from trove.guestagent.datastore import service
from trove.guestagent.datastore.mysql_common import service as commmon_service
from trove.guestagent.utils import docker as docker_util
from trove.guestagent.utils import mysql as mysql_util
from trove.instance import service_status
LOG = logging.getLogger(__name__)
CONF = cfg.CONF
@ -77,24 +77,24 @@ class BaseMySqlAppStatus(service.BaseDbStatus):
cmd = 'mysql -uroot -p%s -e "select 1;"' % root_pass
try:
docker_util.run_command(self.docker_client, cmd)
return instance.ServiceStatuses.HEALTHY
return service_status.ServiceStatuses.HEALTHY
except Exception as exc:
LOG.warning('Failed to run docker command, error: %s',
str(exc))
container_log = docker_util.get_container_logs(
self.docker_client, tail='all')
LOG.warning('container log: %s', '\n'.join(container_log))
return instance.ServiceStatuses.RUNNING
LOG.debug('container log: \n%s', '\n'.join(container_log))
return service_status.ServiceStatuses.RUNNING
elif status == "not running":
return instance.ServiceStatuses.SHUTDOWN
return service_status.ServiceStatuses.SHUTDOWN
elif status == "paused":
return instance.ServiceStatuses.PAUSED
return service_status.ServiceStatuses.PAUSED
elif status == "exited":
return instance.ServiceStatuses.SHUTDOWN
return service_status.ServiceStatuses.SHUTDOWN
elif status == "dead":
return instance.ServiceStatuses.CRASHED
return service_status.ServiceStatuses.CRASHED
else:
return instance.ServiceStatuses.UNKNOWN
return service_status.ServiceStatuses.UNKNOWN
@six.add_metaclass(abc.ABCMeta)
@ -638,8 +638,9 @@ class BaseMySqlApp(object):
raise exception.TroveError(_("Failed to start mysql"))
if not self.status.wait_for_real_status_to_change_to(
instance.ServiceStatuses.HEALTHY,
CONF.state_change_wait_time, update_db):
service_status.ServiceStatuses.HEALTHY,
CONF.state_change_wait_time, update_db
):
raise exception.TroveError(_("Failed to start mysql"))
def start_db_with_conf_changes(self, config_contents):
@ -662,7 +663,7 @@ class BaseMySqlApp(object):
raise exception.TroveError("Failed to stop mysql")
if not self.status.wait_for_real_status_to_change_to(
instance.ServiceStatuses.SHUTDOWN,
service_status.ServiceStatuses.SHUTDOWN,
CONF.state_change_wait_time, update_db):
raise exception.TroveError("Failed to stop mysql")
@ -714,7 +715,7 @@ class BaseMySqlApp(object):
raise exception.TroveError("Failed to restart mysql")
if not self.status.wait_for_real_status_to_change_to(
instance.ServiceStatuses.HEALTHY,
service_status.ServiceStatuses.HEALTHY,
CONF.state_change_wait_time, update_db=False):
raise exception.TroveError("Failed to start mysql")
@ -949,6 +950,20 @@ class BaseMySqlApp(object):
q = "set global read_only = %s" % read_only
client.execute(text(str(q)))
def upgrade(self, upgrade_info):
"""Upgrade the database."""
new_version = upgrade_info.get('datastore_version')
LOG.info('Stopping db container for upgrade')
self.stop_db()
LOG.info('Deleting db container for upgrade')
docker_util.remove_container(self.docker_client)
LOG.info('Starting new db container with version %s for upgrade',
new_version)
self.start_db(update_db=True, ds_version=new_version)
class BaseMySqlRootAccess(object):
def __init__(self, mysql_app):

26
trove/guestagent/datastore/service.py
View File

@ -20,11 +20,11 @@ from oslo_utils import timeutils
from trove.common import cfg
from trove.common import context as trove_context
from trove.common import instance
from trove.common.i18n import _
from trove.conductor import api as conductor_api
from trove.guestagent.common import guestagent_utils
from trove.guestagent.common import operating_system
from trove.instance import service_status
LOG = logging.getLogger(__name__)
CONF = cfg.CONF
@ -74,7 +74,7 @@ class BaseDbStatus(object):
operating_system.write_file(prepare_start_file, '')
self.__refresh_prepare_completed()
self.set_status(instance.ServiceStatuses.BUILDING, True)
self.set_status(service_status.ServiceStatuses.BUILDING, True)
def set_ready(self):
prepare_end_file = guestagent_utils.build_file_path(
@ -92,9 +92,9 @@ class BaseDbStatus(object):
final_status = None
if error_occurred:
final_status = instance.ServiceStatuses.FAILED
final_status = service_status.ServiceStatuses.FAILED
elif post_processing:
final_status = instance.ServiceStatuses.INSTANCE_READY
final_status = service_status.ServiceStatuses.INSTANCE_READY
if final_status:
LOG.info("Set final status to %s.", final_status)
@ -126,8 +126,8 @@ class BaseDbStatus(object):
def is_running(self):
"""True if DB server is running."""
return (self.status is not None and
self.status in [instance.ServiceStatuses.RUNNING,
instance.ServiceStatuses.HEALTHY])
self.status in [service_status.ServiceStatuses.RUNNING,
service_status.ServiceStatuses.HEALTHY])
def set_status(self, status, force=False):
"""Use conductor to update the DB app status."""
@ -199,7 +199,7 @@ class BaseDbStatus(object):
"""
LOG.debug("Waiting for database to start up.")
if not self._wait_for_database_service_status(
instance.ServiceStatuses.RUNNING, timeout, update_db):
service_status.ServiceStatuses.RUNNING, timeout, update_db):
raise RuntimeError(_("Database failed to start."))
LOG.info("Database has started successfully.")
@ -229,7 +229,7 @@ class BaseDbStatus(object):
LOG.debug("Waiting for database to shutdown.")
if not self._wait_for_database_service_status(
instance.ServiceStatuses.SHUTDOWN, timeout, update_db):
service_status.ServiceStatuses.SHUTDOWN, timeout, update_db):
raise RuntimeError(_("Database failed to stop."))
LOG.info("Database has stopped successfully.")
@ -283,9 +283,19 @@ class BaseDbStatus(object):
# outside.
loop = True
# We need 3 (by default) consecutive success db connections for status
# 'HEALTHY'
healthy_count = 0
while loop:
self.status = self.get_actual_db_status()
if self.status == status:
if (status == service_status.ServiceStatuses.HEALTHY and
healthy_count < 2):
healthy_count += 1
time.sleep(CONF.state_change_poll_time)
continue
if update_db:
self.set_status(self.status)
return True

84
trove/instance/models.py
View File

@ -19,13 +19,13 @@ from datetime import datetime
from datetime import timedelta
import os.path
import re
import six
from novaclient import exceptions as nova_exceptions
from oslo_config.cfg import NoSuchOptError
from oslo_log import log as logging
from oslo_utils import encodeutils
from oslo_utils import netutils
import six
from sqlalchemy import func
from trove.backup.models import Backup
@ -33,15 +33,14 @@ from trove.common import cfg
from trove.common import clients
from trove.common import crypto_utils as cu
from trove.common import exception
from trove.common.i18n import _
from trove.common import instance as tr_instance
from trove.common import neutron
from trove.common import notification
from trove.common import server_group as srv_grp
from trove.common import template
from trove.common import timeutils
from trove.common.trove_remote import create_trove_client
from trove.common import utils
from trove.common.i18n import _
from trove.common.trove_remote import create_trove_client
from trove.configuration.models import Configuration
from trove.datastore import models as datastore_models
from trove.datastore.models import DatastoreVersionMetadata as dvm
@ -49,6 +48,7 @@ from trove.datastore.models import DBDatastoreVersionMetadata
from trove.db import get_db_api
from trove.db import models as dbmodels
from trove.extensions.security_group.models import SecurityGroup
from trove.instance import service_status as srvstatus
from trove.instance.tasks import InstanceTask
from trove.instance.tasks import InstanceTasks
from trove.module import models as module_models
@ -339,25 +339,25 @@ class SimpleInstance(object):
action = self.db_info.task_status.action
# Check if we are resetting status or force deleting
if (tr_instance.ServiceStatuses.UNKNOWN == self.datastore_status.status
if (srvstatus.ServiceStatuses.UNKNOWN == self.datastore_status.status
and action == InstanceTasks.DELETING.action):
return InstanceStatus.SHUTDOWN
elif (tr_instance.ServiceStatuses.UNKNOWN ==
elif (srvstatus.ServiceStatuses.UNKNOWN ==
self.datastore_status.status):
return InstanceStatus.ERROR
# Check for taskmanager status.
if 'BUILDING' == action:
if InstanceTasks.BUILDING.action == action:
if 'ERROR' == self.db_info.server_status:
return InstanceStatus.ERROR
return InstanceStatus.BUILD
if 'REBOOTING' == action:
if InstanceTasks.REBOOTING.action == action:
return InstanceStatus.REBOOT
if 'RESIZING' == action:
if InstanceTasks.RESIZING.action == action:
return InstanceStatus.RESIZE
if 'UPGRADING' == action:
if InstanceTasks.UPGRADING.action == action:
return InstanceStatus.UPGRADE
if 'RESTART_REQUIRED' == action:
if InstanceTasks.RESTART_REQUIRED.action == action:
return InstanceStatus.RESTART_REQUIRED
if InstanceTasks.PROMOTING.action == action:
return InstanceStatus.PROMOTE
@ -396,10 +396,10 @@ class SimpleInstance(object):
# Check against the service status.
# The service is only paused during a reboot.
if tr_instance.ServiceStatuses.PAUSED == self.datastore_status.status:
if srvstatus.ServiceStatuses.PAUSED == self.datastore_status.status:
return InstanceStatus.REBOOT
# If the service status is NEW, then we are building.
if tr_instance.ServiceStatuses.NEW == self.datastore_status.status:
if srvstatus.ServiceStatuses.NEW == self.datastore_status.status:
return InstanceStatus.BUILD
# For everything else we can look at the service status mapping.
@ -594,14 +594,19 @@ def load_instance(cls, context, id, needs_server=False,
def load_instance_with_info(cls, context, id, cluster_id=None):
db_info = get_db_info(context, id, cluster_id)
LOG.debug('Task status for instance %s: %s', id, db_info.task_status)
service_status = InstanceServiceStatus.find_by(instance_id=id)
if (db_info.task_status == InstanceTasks.NONE and
not service_status.is_uptodate()):
LOG.warning('Guest agent heartbeat for instance %s has expried', id)
service_status.status = \
srvstatus.ServiceStatuses.FAILED_TIMEOUT_GUESTAGENT
load_simple_instance_server_status(context, db_info)
load_simple_instance_addresses(context, db_info)
service_status = InstanceServiceStatus.find_by(instance_id=id)
LOG.debug("Instance %(instance_id)s service status is %(service_status)s.",
{'instance_id': id, 'service_status': service_status.status})
instance = cls(context, db_info, service_status)
load_guest_info(instance, context, id)
@ -879,7 +884,7 @@ class BaseInstance(SimpleInstance):
def set_servicestatus_deleted(self):
del_instance = InstanceServiceStatus.find_by(instance_id=self.id)
del_instance.set_status(tr_instance.ServiceStatuses.DELETED)
del_instance.set_status(srvstatus.ServiceStatuses.DELETED)
del_instance.save()
def set_instance_fault_deleted(self):
@ -956,7 +961,12 @@ class BaseInstance(SimpleInstance):
self.reset_task_status()
reset_instance = InstanceServiceStatus.find_by(instance_id=self.id)
reset_instance.set_status(tr_instance.ServiceStatuses.UNKNOWN)
reset_instance.set_status(srvstatus.ServiceStatuses.UNKNOWN)
reset_instance.save()
def set_service_status(self, status):
reset_instance = InstanceServiceStatus.find_by(instance_id=self.id)
reset_instance.set_status(status)
reset_instance.save()
@ -1267,7 +1277,7 @@ class Instance(BuiltInstance):
overrides = config.get_configuration_overrides()
service_status = InstanceServiceStatus.create(
instance_id=instance_id,
status=tr_instance.ServiceStatuses.NEW)
status=srvstatus.ServiceStatuses.NEW)
if CONF.trove_dns_support:
dns_client = clients.create_dns_client(context)
@ -1762,19 +1772,32 @@ class Instances(object):
db.server_status = "SHUTDOWN" # Fake it...
db.addresses = []
# volumes = find_volumes(server.id)
datastore_status = InstanceServiceStatus.find_by(
instance_id=db.id)
if not datastore_status.status: # This should never happen.
LOG.error("Server status could not be read for "
"instance id(%s).", db.id)
continue
LOG.debug("Server api_status(%s).",
datastore_status.status.api_status)
# Get the real-time service status.
LOG.debug('Task status for instance %s: %s', db.id,
db.task_status)
if db.task_status == InstanceTasks.NONE:
last_heartbeat_delta = (
timeutils.utcnow() - datastore_status.updated_at)
agent_expiry_interval = timedelta(
seconds=CONF.agent_heartbeat_expiry)
if last_heartbeat_delta > agent_expiry_interval:
LOG.warning(
'Guest agent heartbeat for instance %s has '
'expried', id)
datastore_status.status = \
srvstatus.ServiceStatuses.FAILED_TIMEOUT_GUESTAGENT
except exception.ModelNotFoundError:
LOG.error("Server status could not be read for "
"instance id(%s).", db.id)
continue
ret.append(load_instance(context, db, datastore_status,
server=server))
return ret
@ -2001,7 +2024,7 @@ class InstanceServiceStatus(dbmodels.DatabaseModelBase):
def _validate(self, errors):
if self.status is None:
errors['status'] = "Cannot be None."
if tr_instance.ServiceStatus.from_code(self.status_id) is None:
if srvstatus.ServiceStatus.from_code(self.status_id) is None:
errors['status_id'] = "Not valid."
def get_status(self):
@ -2012,7 +2035,7 @@ class InstanceServiceStatus(dbmodels.DatabaseModelBase):
status of the service
:rtype: trove.common.instance.ServiceStatus
"""
return tr_instance.ServiceStatus.from_code(self.status_id)
return srvstatus.ServiceStatus.from_code(self.status_id)
def set_status(self, value):
"""
@ -2027,6 +2050,15 @@ class InstanceServiceStatus(dbmodels.DatabaseModelBase):
self['updated_at'] = timeutils.utcnow()
return get_db_api().save(self)
def is_uptodate(self):
"""Check if the service status heartbeat is up to date."""
heartbeat_expiry = timedelta(seconds=CONF.agent_heartbeat_expiry)
last_update = (timeutils.utcnow() - self.updated_at)
if last_update < heartbeat_expiry:
return True
return False
status = property(get_status, set_status)
@ -2039,6 +2071,6 @@ def persisted_models():
MYSQL_RESPONSIVE_STATUSES = [
tr_instance.ServiceStatuses.RUNNING,
tr_instance.ServiceStatuses.HEALTHY
srvstatus.ServiceStatuses.RUNNING,
srvstatus.ServiceStatuses.HEALTHY
]

1
trove/common/instance.py → trove/instance/service_status.py
View File

@ -104,6 +104,7 @@ class ServiceStatuses(object):
RESTART_REQUIRED = ServiceStatus(0x20, 'restart required',
'RESTART_REQUIRED')
HEALTHY = ServiceStatus(0x21, 'healthy', 'HEALTHY')
UPGRADING = ServiceStatus(0x22, 'upgrading', 'UPGRADING')
# Dissuade further additions at run-time.

169
trove/taskmanager/models.py
View File

@ -14,6 +14,7 @@
import copy
import os.path
import time
import traceback
from cinderclient import exceptions as cinder_exceptions
@ -21,7 +22,6 @@ from eventlet import greenthread
from eventlet.timeout import Timeout
from oslo_log import log as logging
from swiftclient.client import ClientException
import time
from trove import rpc
from trove.backup import models as bkup_models
@ -33,9 +33,7 @@ from trove.cluster.models import Cluster
from trove.cluster.models import DBCluster
from trove.common import cfg
from trove.common import clients
from trove.common import crypto_utils as cu
from trove.common import exception
from trove.common import instance as rd_instance
from trove.common import neutron
from trove.common import template
from trove.common import timeutils
@ -51,7 +49,6 @@ from trove.common.exception import PollTimeOut
from trove.common.exception import TroveError
from trove.common.exception import VolumeCreationFailure
from trove.common.i18n import _
from trove.common.instance import ServiceStatuses
from trove.common.notification import DBaaSInstanceRestart
from trove.common.notification import DBaaSInstanceUpgrade
from trove.common.notification import EndNotification
@ -63,6 +60,7 @@ from trove.common.strategies.cluster import strategy
from trove.common.utils import try_recover
from trove.extensions.mysql import models as mysql_models
from trove.instance import models as inst_models
from trove.instance import service_status as srvstatus
from trove.instance.models import BuiltInstance
from trove.instance.models import DBInstance
from trove.instance.models import FreshInstance
@ -202,24 +200,36 @@ class ClusterTasks(Cluster):
shard_id=None):
"""Wait for all instances to get READY."""
return self._all_instances_acquire_status(
instance_ids, cluster_id, shard_id, ServiceStatuses.INSTANCE_READY,
fast_fail_statuses=[ServiceStatuses.FAILED,
ServiceStatuses.FAILED_TIMEOUT_GUESTAGENT])
instance_ids, cluster_id, shard_id,
srvstatus.ServiceStatuses.INSTANCE_READY,
fast_fail_statuses=[
srvstatus.ServiceStatuses.FAILED,
srvstatus.ServiceStatuses.FAILED_TIMEOUT_GUESTAGENT
]
)
def _all_instances_shutdown(self, instance_ids, cluster_id,
shard_id=None):
"""Wait for all instances to go SHUTDOWN."""
return self._all_instances_acquire_status(
instance_ids, cluster_id, shard_id, ServiceStatuses.SHUTDOWN,
fast_fail_statuses=[ServiceStatuses.FAILED,
ServiceStatuses.FAILED_TIMEOUT_GUESTAGENT])
instance_ids, cluster_id, shard_id,
srvstatus.ServiceStatuses.SHUTDOWN,
fast_fail_statuses=[
srvstatus.ServiceStatuses.FAILED,
srvstatus.ServiceStatuses.FAILED_TIMEOUT_GUESTAGENT
]
)
def _all_instances_running(self, instance_ids, cluster_id, shard_id=None):
"""Wait for all instances to become ACTIVE."""
return self._all_instances_acquire_status(
instance_ids, cluster_id, shard_id, ServiceStatuses.RUNNING,
fast_fail_statuses=[ServiceStatuses.FAILED,
ServiceStatuses.FAILED_TIMEOUT_GUESTAGENT])
instance_ids, cluster_id, shard_id,
srvstatus.ServiceStatuses.RUNNING,
fast_fail_statuses=[
srvstatus.ServiceStatuses.FAILED,
srvstatus.ServiceStatuses.FAILED_TIMEOUT_GUESTAGENT
]
)
def _all_instances_acquire_status(
self, instance_ids, cluster_id, shard_id, expected_status,
@ -427,7 +437,10 @@ class FreshInstanceTasks(FreshInstance, NotifyMixin, ConfigurationMixin):
utils.poll_until(self._service_is_active,
sleep_time=CONF.usage_sleep_time,
time_out=timeout)
LOG.info("Created instance %s successfully.", self.id)
if not self.db_info.task_status.is_error:
self.reset_task_status()
TroveInstanceCreate(instance=self,
instance_size=flavor['ram']).notify()
except (TroveError, PollTimeOut) as ex:
@ -582,9 +595,6 @@ class FreshInstanceTasks(FreshInstance, NotifyMixin, ConfigurationMixin):
if root_password:
self.report_root_enabled()
if not self.db_info.task_status.is_error:
self.reset_task_status()
# when DNS is supported, we attempt to add this after the
# instance is prepared. Otherwise, if DNS fails, instances
# end up in a poorer state and there's no tooling around
@ -723,12 +733,13 @@ class FreshInstanceTasks(FreshInstance, NotifyMixin, ConfigurationMixin):
if CONF.update_status_on_fail:
# Updating service status
service = InstanceServiceStatus.find_by(instance_id=self.id)
service.set_status(ServiceStatuses.FAILED_TIMEOUT_GUESTAGENT)
service.set_status(
srvstatus.ServiceStatuses.FAILED_TIMEOUT_GUESTAGENT)
service.save()
LOG.error(
"Service status: %s, service error description: %s",
ServiceStatuses.FAILED_TIMEOUT_GUESTAGENT.api_status,
ServiceStatuses.FAILED_TIMEOUT_GUESTAGENT.description
srvstatus.ServiceStatuses.FAILED_TIMEOUT_GUESTAGENT.api_status,
srvstatus.ServiceStatuses.FAILED_TIMEOUT_GUESTAGENT.description
)
# Updating instance status
@ -756,14 +767,14 @@ class FreshInstanceTasks(FreshInstance, NotifyMixin, ConfigurationMixin):
service = InstanceServiceStatus.find_by(instance_id=self.id)
status = service.get_status()
if (status == rd_instance.ServiceStatuses.RUNNING or
status == rd_instance.ServiceStatuses.INSTANCE_READY or
status == rd_instance.ServiceStatuses.HEALTHY):
if (status == srvstatus.ServiceStatuses.RUNNING or
status == srvstatus.ServiceStatuses.INSTANCE_READY or
status == srvstatus.ServiceStatuses.HEALTHY):
return True
elif status not in [rd_instance.ServiceStatuses.NEW,
rd_instance.ServiceStatuses.BUILDING,
rd_instance.ServiceStatuses.UNKNOWN,
rd_instance.ServiceStatuses.DELETED]:
elif status not in [srvstatus.ServiceStatuses.NEW,
srvstatus.ServiceStatuses.BUILDING,
srvstatus.ServiceStatuses.UNKNOWN,
srvstatus.ServiceStatuses.DELETED]:
raise TroveError(_("Service not active, status: %s") % status)
c_id = self.db_info.compute_instance_id
@ -1069,6 +1080,27 @@ class BuiltInstanceTasks(BuiltInstance, NotifyMixin, ConfigurationMixin):
associated with a compute server.
"""
def is_service_healthy(self):
"""Wait for the db service up and running.
This method is supposed to be called with poll_until against an
existing db instance.
"""
service = InstanceServiceStatus.find_by(instance_id=self.id)
status = service.get_status()
if service.is_uptodate():
if status in [srvstatus.ServiceStatuses.HEALTHY]:
return True
elif status in [
srvstatus.ServiceStatuses.FAILED,
srvstatus.ServiceStatuses.UNKNOWN,
srvstatus.ServiceStatuses.FAILED_TIMEOUT_GUESTAGENT
]:
raise TroveError('Database service error, status: %s' % status)
return False
def resize_volume(self, new_size):
LOG.info("Resizing volume for instance %(instance_id)s from "
"%(old_size)s GB to %(new_size)s GB.",
@ -1219,6 +1251,10 @@ class BuiltInstanceTasks(BuiltInstance, NotifyMixin, ConfigurationMixin):
LOG.info("Starting database on instance %s.", self.id)
self.guest.restart()
# Wait for database service up and running
utils.poll_until(self.is_service_healthy,
time_out=CONF.report_interval * 2)
LOG.info("Rebooted instance %s successfully.", self.id)
except Exception as e:
LOG.error("Failed to reboot instance %(id)s: %(e)s",
@ -1276,79 +1312,40 @@ class BuiltInstanceTasks(BuiltInstance, NotifyMixin, ConfigurationMixin):
This does not change the reference for this BuiltInstanceTask
"""
datastore_status = InstanceServiceStatus.find_by(instance_id=self.id)
datastore_status.status = rd_instance.ServiceStatuses.PAUSED
datastore_status.status = srvstatus.ServiceStatuses.PAUSED
datastore_status.save()
def upgrade(self, datastore_version):
LOG.info("Upgrading instance %s to new datastore version %s",
self.id, datastore_version)
def server_finished_rebuilding():
self.refresh_compute_server_info()
return not self.server_status_matches(['REBUILD'])
self.set_service_status(srvstatus.ServiceStatuses.UPGRADING)
try:
upgrade_info = self.guest.pre_upgrade()
upgrade_info = upgrade_info if upgrade_info else {}
upgrade_info.update({'datastore_version': datastore_version.name})
self.guest.upgrade(upgrade_info)
if self.volume_id:
volume = self.volume_client.volumes.get(self.volume_id)
volume_device = self._fix_device_path(
volume.attachments[0]['device'])
if volume:
upgrade_info['device'] = volume_device
# BUG(1650518): Cleanup in the Pike release some instances
# that we will be upgrading will be pre secureserialier
# and will have no instance_key entries. If this is one of
# those instances, make a key. That will make it appear in
# the injected files that are generated next. From this
# point, and until the guest comes up, attempting to send
# messages to it will fail because the RPC framework will
# encrypt messages to a guest which potentially doesn't
# have the code to handle it.
if CONF.enable_secure_rpc_messaging and (
self.db_info.encrypted_key is None):
encrypted_key = cu.encode_data(cu.encrypt_data(
cu.generate_random_key(),
CONF.inst_rpc_key_encr_key))
self.update_db(encrypted_key=encrypted_key)
LOG.debug("Generated unique RPC encryption key for "
"instance = %(id)s, key = %(key)s",
{'id': self.id, 'key': encrypted_key})
injected_files = self.get_injected_files(
datastore_version.manager)
LOG.debug("Rebuilding instance %(instance)s with image %(image)s.",
{'instance': self, 'image': datastore_version.image_id})
self.server.rebuild(datastore_version.image_id,
files=injected_files)
utils.poll_until(
server_finished_rebuilding,
sleep_time=5, time_out=600)
if not self.server_status_matches(['ACTIVE']):
raise TroveError(_("Instance %(instance)s failed to "
"upgrade to %(datastore_version)s"),
instance=self,
datastore_version=datastore_version)
LOG.info('Finished rebuilding server for instance %s', self.id)
self.guest.post_upgrade(upgrade_info)
# Wait for db instance healthy
LOG.info('Waiting for instance %s to be healthy after upgrading',
self.id)
utils.poll_until(self.is_service_healthy, time_out=600,
sleep_time=5)
self.reset_task_status()
LOG.info("Finished upgrading instance %s to new datastore "
"version %s",
self.id, datastore_version)
"version %s", self.id, datastore_version)
except Exception as e:
LOG.exception(e)
err = inst_models.InstanceTasks.BUILDING_ERROR_SERVER
self.update_db(task_status=err)
raise e
LOG.error('Failed to upgrade instance %s, error: %s', self.id, e)
self.update_db(
task_status=inst_models.InstanceTasks.BUILDING_ERROR_SERVER)
# Some cinder drivers appear to return "vdb" instead of "/dev/vdb".
# We need to account for that.
def _fix_device_path(self, device):
"""Get correct device path.
Some cinder drivers appear to return "vdb" instead of "/dev/vdb".
"""
if device.startswith("/dev"):
return device
else:
@ -1515,7 +1512,7 @@ class ResizeVolumeAction(object):
"status to failed.", {'func': orig_func.__name__,
'id': self.instance.id})
service = InstanceServiceStatus.find_by(instance_id=self.instance.id)
service.set_status(ServiceStatuses.FAILED)
service.set_status(srvstatus.ServiceStatuses.FAILED)
service.save()
def _recover_restart(self, orig_func):
@ -1790,7 +1787,7 @@ class ResizeActionBase(object):
def _datastore_is_offline(self):
self.instance._refresh_datastore_status()
return (self.instance.datastore_status_matches(
rd_instance.ServiceStatuses.SHUTDOWN))
srvstatus.ServiceStatuses.SHUTDOWN))
def _revert_nova_action(self):
LOG.debug("Instance %s calling Compute revert resize...",
@ -1811,7 +1808,7 @@ class ResizeActionBase(object):
def _guest_is_awake(self):
self.instance._refresh_datastore_status()
return not self.instance.datastore_status_matches(
rd_instance.ServiceStatuses.PAUSED)
srvstatus.ServiceStatuses.PAUSED)
def _perform_nova_action(self):
"""Calls Nova to resize or migrate an instance, and confirms."""

13
trove/tests/api/instances_actions.py
View File

@ -267,6 +267,15 @@ class RebootTestBase(ActionTestBase):
poll_until(is_finished_rebooting, time_out=TIME_OUT_TIME)
def wait_for_status(self, status, timeout=60):
def is_status():
instance = self.instance
if instance.status in status:
return True
return False
poll_until(is_status, time_out=timeout)
@test(groups=[tests.DBAAS_API_INSTANCE_ACTIONS],
depends_on_groups=[tests.DBAAS_API_DATABASES],
@ -312,9 +321,9 @@ class StopTests(RebootTestBase):
@test(depends_on=[test_ensure_mysql_is_running])
def test_stop_mysql(self):
"""Stops MySQL."""
"""Stops MySQL by admin."""
instance_info.dbaas_admin.management.stop(self.instance_id)
self.wait_for_failure_status()
self.wait_for_status(['SHUTDOWN'], timeout=60)
@test(depends_on=[test_stop_mysql])
def test_volume_info_while_mysql_is_down(self):

42
trove/tests/api/instances_resize.py
View File

@ -13,21 +13,21 @@
# License for the specific language governing permissions and limitations
# under the License.
from novaclient.exceptions import BadRequest
from novaclient.v2.servers import Server
from unittest import mock
from novaclient.exceptions import BadRequest
from novaclient.v2.servers import Server
from oslo_messaging._drivers.common import RPCException
from proboscis import test
from testtools import TestCase
from trove.common.exception import PollTimeOut
from trove.common.exception import TroveError
from trove.common import instance as rd_instance
from trove.common import template
from trove.common import utils
from trove.common.exception import PollTimeOut
from trove.common.exception import TroveError
from trove.datastore.models import DatastoreVersion
from trove.guestagent import api as guest
from trove.instance import service_status as srvstatus
from trove.instance.models import DBInstance
from trove.instance.models import InstanceServiceStatus
from trove.instance.tasks import InstanceTasks
@ -63,7 +63,7 @@ class ResizeTestBase(TestCase):
self.server,
datastore_status=InstanceServiceStatus.create(
instance_id=self.db_info.id,
status=rd_instance.ServiceStatuses.RUNNING))
status=srvstatus.ServiceStatuses.RUNNING))
self.instance.server.flavor = {'id': OLD_FLAVOR_ID}
self.guest = mock.MagicMock(spec=guest.API)
self.instance._guest = self.guest
@ -124,7 +124,7 @@ class ResizeTests(ResizeTestBase):
task_status=InstanceTasks.NONE)
def test_nova_wont_resize(self):
self._datastore_changes_to(rd_instance.ServiceStatuses.SHUTDOWN)
self._datastore_changes_to(srvstatus.ServiceStatuses.SHUTDOWN)
self.server.resize.side_effect = BadRequest(400)
self.server.status = "ACTIVE"
self.assertRaises(BadRequest, self.action.execute)
@ -135,7 +135,7 @@ class ResizeTests(ResizeTestBase):
task_status=InstanceTasks.NONE)
def test_nova_resize_timeout(self):
self._datastore_changes_to(rd_instance.ServiceStatuses.SHUTDOWN)
self._datastore_changes_to(srvstatus.ServiceStatuses.SHUTDOWN)
self.server.status = "ACTIVE"
with mock.patch.object(utils, 'poll_until') as mock_poll_until:
@ -150,7 +150,7 @@ class ResizeTests(ResizeTestBase):
task_status=InstanceTasks.NONE)
def test_nova_doesnt_change_flavor(self):
self._datastore_changes_to(rd_instance.ServiceStatuses.SHUTDOWN)
self._datastore_changes_to(srvstatus.ServiceStatuses.SHUTDOWN)
with mock.patch.object(utils, 'poll_until') as mock_poll_until:
self.poll_until_side_effects.extend([
@ -177,7 +177,7 @@ class ResizeTests(ResizeTestBase):
task_status=InstanceTasks.NONE)
def test_nova_resize_fails(self):
self._datastore_changes_to(rd_instance.ServiceStatuses.SHUTDOWN)
self._datastore_changes_to(srvstatus.ServiceStatuses.SHUTDOWN)
with mock.patch.object(utils, 'poll_until') as mock_poll_until:
self.poll_until_side_effects.extend([
@ -200,7 +200,7 @@ class ResizeTests(ResizeTestBase):
task_status=InstanceTasks.NONE)
def test_nova_resizes_in_weird_state(self):
self._datastore_changes_to(rd_instance.ServiceStatuses.SHUTDOWN)
self._datastore_changes_to(srvstatus.ServiceStatuses.SHUTDOWN)
with mock.patch.object(utils, 'poll_until') as mock_poll_until:
self.poll_until_side_effects.extend([
@ -224,7 +224,7 @@ class ResizeTests(ResizeTestBase):
task_status=InstanceTasks.NONE)
def test_guest_is_not_okay(self):
self._datastore_changes_to(rd_instance.ServiceStatuses.SHUTDOWN)
self._datastore_changes_to(srvstatus.ServiceStatuses.SHUTDOWN)
with mock.patch.object(utils, 'poll_until') as mock_poll_until:
self.poll_until_side_effects.extend([
@ -237,7 +237,7 @@ class ResizeTests(ResizeTestBase):
self.instance.set_datastore_status_to_paused.side_effect = (
lambda: self._datastore_changes_to(
rd_instance.ServiceStatuses.PAUSED))
srvstatus.ServiceStatuses.PAUSED))
self.assertRaises(PollTimeOut, self.action.execute)
@ -257,7 +257,7 @@ class ResizeTests(ResizeTestBase):
task_status=InstanceTasks.NONE)
def test_mysql_is_not_okay(self):
self._datastore_changes_to(rd_instance.ServiceStatuses.SHUTDOWN)
self._datastore_changes_to(srvstatus.ServiceStatuses.SHUTDOWN)
with mock.patch.object(utils, 'poll_until') as mock_poll_until:
self.poll_until_side_effects.extend([
@ -269,7 +269,7 @@ class ResizeTests(ResizeTestBase):
self.instance.set_datastore_status_to_paused.side_effect = (
lambda: self._datastore_changes_to(
rd_instance.ServiceStatuses.SHUTDOWN))
srvstatus.ServiceStatuses.SHUTDOWN))
self._start_mysql()
self.assertRaises(PollTimeOut, self.action.execute)
@ -290,7 +290,7 @@ class ResizeTests(ResizeTestBase):
task_status=InstanceTasks.NONE)
def test_confirm_resize_fails(self):
self._datastore_changes_to(rd_instance.ServiceStatuses.SHUTDOWN)
self._datastore_changes_to(srvstatus.ServiceStatuses.SHUTDOWN)
with mock.patch.object(utils, 'poll_until') as mock_poll_until:
self.poll_until_side_effects.extend([
@ -303,7 +303,7 @@ class ResizeTests(ResizeTestBase):
self.instance.set_datastore_status_to_paused.side_effect = (
lambda: self._datastore_changes_to(
rd_instance.ServiceStatuses.RUNNING))
srvstatus.ServiceStatuses.RUNNING))
self.server.confirm_resize.side_effect = BadRequest(400)
self._start_mysql()
@ -322,7 +322,7 @@ class ResizeTests(ResizeTestBase):
task_status=InstanceTasks.NONE)
def test_revert_nova_fails(self):
self._datastore_changes_to(rd_instance.ServiceStatuses.SHUTDOWN)
self._datastore_changes_to(srvstatus.ServiceStatuses.SHUTDOWN)
with mock.patch.object(utils, 'poll_until') as mock_poll_until:
self.poll_until_side_effects.extend([
@ -335,7 +335,7 @@ class ResizeTests(ResizeTestBase):
self.instance.set_datastore_status_to_paused.side_effect = (
lambda: self._datastore_changes_to(
rd_instance.ServiceStatuses.PAUSED))
srvstatus.ServiceStatuses.PAUSED))
self.assertRaises(PollTimeOut, self.action.execute)
@ -363,7 +363,7 @@ class MigrateTests(ResizeTestBase):
self.action = models.MigrateAction(self.instance)
def test_successful_migrate(self):
self._datastore_changes_to(rd_instance.ServiceStatuses.SHUTDOWN)
self._datastore_changes_to(srvstatus.ServiceStatuses.SHUTDOWN)
with mock.patch.object(utils, 'poll_until') as mock_poll_until:
self.poll_until_side_effects.extend([
@ -375,7 +375,7 @@ class MigrateTests(ResizeTestBase):
self.instance.set_datastore_status_to_paused.side_effect = (
lambda: self._datastore_changes_to(
rd_instance.ServiceStatuses.RUNNING))
srvstatus.ServiceStatuses.RUNNING))
self.action.execute()

13
trove/tests/api/mgmt/instances_actions.py
View File

@ -12,23 +12,24 @@
# License for the specific language governing permissions and limitations
# under the License.
from unittest import mock
from novaclient.v2.servers import Server
from proboscis import after_class
from proboscis.asserts import assert_equal
from proboscis.asserts import assert_raises
from proboscis import before_class
from proboscis import SkipTest
from proboscis import test
from unittest import mock
from proboscis.asserts import assert_equal
from proboscis.asserts import assert_raises
from trove.backup import models as backup_models
from trove.backup import state
from trove.common.context import TroveContext
from trove.common import exception
import trove.common.instance as tr_instance
from trove.common.context import TroveContext
from trove.extensions.mgmt.instances.models import MgmtInstance
from trove.extensions.mgmt.instances.service import MgmtInstanceController
from trove.instance import models as imodels
from trove.instance import service_status as srvstatus
from trove.instance.models import DBInstance
from trove.instance.tasks import InstanceTasks
from trove.tests.config import CONFIG
@ -65,7 +66,7 @@ class MgmtInstanceBase(object):
self.db_info,
self.server,
datastore_status=imodels.InstanceServiceStatus(
tr_instance.ServiceStatuses.RUNNING))
srvstatus.ServiceStatuses.RUNNING))
def _make_request(self, path='/', context=None, **kwargs):
from webob import Request

10
trove/tests/fakes/guestagent.py
View File

@ -20,7 +20,7 @@ import eventlet
from oslo_log import log as logging
from trove.common import exception as rd_exception
from trove.common import instance as rd_instance
from trove.instance import service_status as srvstatus
from trove.tests.util import unquote_user_host
DB = {}
@ -236,9 +236,9 @@ class FakeGuest(object):
def update_db():
status = InstanceServiceStatus.find_by(instance_id=self.id)
if instance_name.endswith('GUEST_ERROR'):
status.status = rd_instance.ServiceStatuses.FAILED
status.status = srvstatus.ServiceStatuses.FAILED
else:
status.status = rd_instance.ServiceStatuses.HEALTHY
status.status = srvstatus.ServiceStatuses.HEALTHY
status.save()
AgentHeartBeat.create(instance_id=self.id)
eventlet.spawn_after(3.5, update_db)
@ -246,8 +246,8 @@ class FakeGuest(object):
def _set_task_status(self, new_status='HEALTHY'):
from trove.instance.models import InstanceServiceStatus
print("Setting status to %s" % new_status)
states = {'HEALTHY': rd_instance.ServiceStatuses.HEALTHY,
'SHUTDOWN': rd_instance.ServiceStatuses.SHUTDOWN,
states = {'HEALTHY': srvstatus.ServiceStatuses.HEALTHY,
'SHUTDOWN': srvstatus.ServiceStatuses.SHUTDOWN,
}
status = InstanceServiceStatus.find_by(instance_id=self.id)
status.status = states[new_status]

13
trove/tests/fakes/nova.py
View File

@ -13,18 +13,17 @@
# License for the specific language governing permissions and limitations
# under the License.
import collections
import uuid
import eventlet
from novaclient import exceptions as nova_exceptions
from oslo_log import log as logging
from trove.common.exception import PollTimeOut
from trove.common import instance as rd_instance
from trove.instance import service_status as srvstatus
from trove.tests.fakes.common import authorize
import collections
import eventlet
import uuid
LOG = logging.getLogger(__name__)
FAKE_HOSTS = ["fake_host_1", "fake_host_2"]
@ -326,7 +325,7 @@ class FakeServers(object):
instance = DBInstance.find_by(compute_instance_id=id)
LOG.debug("Setting server %s to running", instance.id)
status = InstanceServiceStatus.find_by(instance_id=instance.id)
status.status = rd_instance.ServiceStatuses.RUNNING
status.status = srvstatus.ServiceStatuses.RUNNING
status.save()
eventlet.spawn_after(time_from_now, set_server_running)

3
trove/tests/unittests/conductor/test_methods.py
View File

@ -18,14 +18,13 @@ from oslo_utils import timeutils
from trove.backup import models as bkup_models
from trove.backup import state
from trove.common import exception as t_exception
from trove.common.instance import ServiceStatuses
from trove.common import utils
from trove.conductor import manager as conductor_manager
from trove.instance import models as t_models
from trove.instance.service_status import ServiceStatuses
from trove.tests.unittests import trove_testtools
from trove.tests.unittests.util import util
# See LP bug #1255178
OLD_DBB_SAVE = bkup_models.DBBackup.save

6
trove/tests/unittests/instance/test_instance_models.py
View File

@ -11,15 +11,14 @@
# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
# License for the specific language governing permissions and limitations
# under the License.
from mock import Mock
from mock import patch
import uuid
from mock import Mock, patch
from trove.backup import models as backup_models
from trove.common import cfg
from trove.common import clients
from trove.common import exception
from trove.common.instance import ServiceStatuses
from trove.common import neutron
from trove.datastore import models as datastore_models
from trove.instance import models
@ -29,6 +28,7 @@ from trove.instance.models import Instance
from trove.instance.models import instance_encryption_key_cache
from trove.instance.models import InstanceServiceStatus
from trove.instance.models import SimpleInstance
from trove.instance.service_status import ServiceStatuses
from trove.instance.tasks import InstanceTasks
from trove.taskmanager import api as task_api
from trove.tests.fakes import nova

5
trove/tests/unittests/instance/test_instance_status.py
View File

@ -13,15 +13,16 @@
# License for the specific language governing permissions and limitations
# under the License.
#
from trove.common.instance import ServiceStatuses
import uuid
from trove.datastore import models
from trove.instance.models import InstanceServiceStatus
from trove.instance.models import InstanceStatus
from trove.instance.models import SimpleInstance
from trove.instance.service_status import ServiceStatuses
from trove.instance.tasks import InstanceTasks
from trove.tests.unittests import trove_testtools
from trove.tests.unittests.util import util
import uuid
class FakeInstanceTask(object):

42
trove/tests/unittests/mgmt/test_models.py
View File

@ -13,26 +13,32 @@
# License for the specific language governing permissions and limitations
# under the License.
#
from testtools.matchers import Equals
from testtools.matchers import Is
from testtools.matchers import Not
import uuid
from mock import MagicMock, patch, ANY
from mock import ANY
from mock import MagicMock
from mock import patch
from novaclient.client import Client
from novaclient.v2.flavors import FlavorManager, Flavor
from novaclient.v2.servers import Server, ServerManager
from novaclient.v2.flavors import Flavor
from novaclient.v2.flavors import FlavorManager
from novaclient.v2.servers import Server
from novaclient.v2.servers import ServerManager
from oslo_config import cfg
from testtools.matchers import Equals, Is, Not
from trove import rpc
from trove.backup.models import Backup
from trove.common import clients
from trove.common import exception
from trove.common import instance as rd_instance
from trove.datastore import models as datastore_models
import trove.extensions.mgmt.instances.models as mgmtmodels
from trove.guestagent.api import API
from trove.instance import service_status as srvstatus
from trove.instance.models import DBInstance
from trove.instance.models import InstanceServiceStatus
from trove.instance.tasks import InstanceTasks
from trove import rpc
from trove.tests.unittests import trove_testtools
from trove.tests.unittests.util import util
@ -98,12 +104,12 @@ class MockMgmtInstanceTest(trove_testtools.TestCase):
compute_instance_id='compute_id_1',
server_id='server_id_1',
tenant_id='tenant_id_1',
server_status=rd_instance.ServiceStatuses.
server_status=srvstatus.ServiceStatuses.
BUILDING.api_status,
deleted=False)
instance.save()
service_status = InstanceServiceStatus(
rd_instance.ServiceStatuses.RUNNING,
srvstatus.ServiceStatuses.RUNNING,
id=str(uuid.uuid4()),
instance_id=instance.id,
)
@ -122,7 +128,7 @@ class TestNotificationTransformer(MockMgmtInstanceTest):
@patch('trove.instance.models.LOG')
def test_transformer(self, mock_logging):
status = rd_instance.ServiceStatuses.BUILDING.api_status
status = srvstatus.ServiceStatuses.BUILDING.api_status
instance, service_status = self.build_db_instance(
status, InstanceTasks.BUILDING)
payloads = mgmtmodels.NotificationTransformer(
@ -184,7 +190,7 @@ class TestNovaNotificationTransformer(MockMgmtInstanceTest):
Equals('unknown'))
def test_transformer(self):
status = rd_instance.ServiceStatuses.BUILDING.api_status
status = srvstatus.ServiceStatuses.BUILDING.api_status
instance, service_status = self.build_db_instance(
status, InstanceTasks.BUILDING)
@ -223,7 +229,7 @@ class TestNovaNotificationTransformer(MockMgmtInstanceTest):
@patch('trove.extensions.mgmt.instances.models.LOG')
def test_transformer_invalid_datastore_manager(self, mock_logging):
status = rd_instance.ServiceStatuses.BUILDING.api_status
status = srvstatus.ServiceStatuses.BUILDING.api_status
instance, service_status = self.build_db_instance(
status, InstanceTasks.BUILDING)
version = datastore_models.DBDatastoreVersion.get_by(
@ -268,9 +274,9 @@ class TestNovaNotificationTransformer(MockMgmtInstanceTest):
self.addCleanup(self.do_cleanup, instance, service_status)
def test_transformer_shutdown_instance(self):
status = rd_instance.ServiceStatuses.SHUTDOWN.api_status
status = srvstatus.ServiceStatuses.SHUTDOWN.api_status
instance, service_status = self.build_db_instance(status)
service_status.set_status(rd_instance.ServiceStatuses.SHUTDOWN)
service_status.set_status(srvstatus.ServiceStatuses.SHUTDOWN)
server = MagicMock(spec=Server)
server.user_id = 'test_user_id'
@ -296,9 +302,9 @@ class TestNovaNotificationTransformer(MockMgmtInstanceTest):
self.addCleanup(self.do_cleanup, instance, service_status)
def test_transformer_no_nova_instance(self):
status = rd_instance.ServiceStatuses.SHUTDOWN.api_status
status = srvstatus.ServiceStatuses.SHUTDOWN.api_status
instance, service_status = self.build_db_instance(status)
service_status.set_status(rd_instance.ServiceStatuses.SHUTDOWN)
service_status.set_status(srvstatus.ServiceStatuses.SHUTDOWN)
mgmt_instance = mgmtmodels.SimpleMgmtInstance(self.context,
instance,
None,
@ -321,7 +327,7 @@ class TestNovaNotificationTransformer(MockMgmtInstanceTest):
self.addCleanup(self.do_cleanup, instance, service_status)
def test_transformer_flavor_cache(self):
status = rd_instance.ServiceStatuses.BUILDING.api_status
status = srvstatus.ServiceStatuses.BUILDING.api_status
instance, service_status = self.build_db_instance(
status, InstanceTasks.BUILDING)
@ -366,7 +372,7 @@ class TestMgmtInstanceTasks(MockMgmtInstanceTest):
super(TestMgmtInstanceTasks, cls).setUpClass()
def test_public_exists_events(self):
status = rd_instance.ServiceStatuses.BUILDING.api_status
status = srvstatus.ServiceStatuses.BUILDING.api_status
instance, service_status = self.build_db_instance(
status, task_status=InstanceTasks.BUILDING)
server = MagicMock(spec=Server)
@ -443,7 +449,7 @@ class TestMgmtInstanceDeleted(MockMgmtInstanceTest):
class TestMgmtInstancePing(MockMgmtInstanceTest):
def test_rpc_ping(self):
status = rd_instance.ServiceStatuses.RUNNING.api_status
status = srvstatus.ServiceStatuses.RUNNING.api_status
instance, service_status = self.build_db_instance(
status, task_status=InstanceTasks.NONE)
mgmt_instance = mgmtmodels.MgmtInstance(instance,

5
trove/tests/unittests/taskmanager/test_clusters.py
View File

@ -21,17 +21,16 @@ from mock import patch
from trove.cluster.models import ClusterTasks as ClusterTaskStatus
from trove.cluster.models import DBCluster
from trove.common import utils
from trove.common.strategies.cluster.experimental.mongodb.taskmanager import (
MongoDbClusterTasks as ClusterTasks)
from trove.common import utils
from trove.datastore import models as datastore_models
from trove.instance.models import BaseInstance
from trove.instance.models import DBInstance
from trove.instance.models import Instance
from trove.instance.models import InstanceServiceStatus
from trove.instance.models import InstanceTasks
# from trove.taskmanager.models import BuiltInstanceTasks
from trove.taskmanager.models import ServiceStatuses
from trove.instance.service_status import ServiceStatuses
from trove.tests.unittests import trove_testtools

2
trove/tests/unittests/taskmanager/test_galera_clusters.py
View File

@ -29,7 +29,7 @@ from trove.instance.models import DBInstance
from trove.instance.models import Instance
from trove.instance.models import InstanceServiceStatus
from trove.instance.models import InstanceTasks
from trove.taskmanager.models import ServiceStatuses
from trove.instance.service_status import ServiceStatuses
from trove.tests.unittests import trove_testtools
from trove.tests.unittests.util import util

54
trove/tests/unittests/taskmanager/test_models.py
View File

@ -16,29 +16,34 @@ from tempfile import NamedTemporaryFile
from unittest import mock
from cinderclient import exceptions as cinder_exceptions
import cinderclient.v2.client as cinderclient
from cinderclient.v2 import volumes as cinderclient_volumes
from mock import Mock, MagicMock, patch, PropertyMock, call
import cinderclient.v2.client as cinderclient
from mock import call
from mock import MagicMock
from mock import Mock
from mock import patch
from mock import PropertyMock
import neutronclient.v2_0.client as neutronclient
from novaclient import exceptions as nova_exceptions
import novaclient.v2.flavors
import novaclient.v2.servers
from oslo_config import cfg
from swiftclient.client import ClientException
from testtools.matchers import Equals, Is
from testtools.matchers import Equals
from testtools.matchers import Is
import trove.backup.models
from trove import rpc
from trove.backup import models as backup_models
from trove.backup import state
import trove.backup.models
from trove.common import timeutils
from trove.common import utils
import trove.common.context
from trove.common.exception import GuestError
from trove.common.exception import PollTimeOut
from trove.common.exception import TroveError
from trove.common.instance import ServiceStatuses
from trove.common.notification import TroveInstanceModifyVolume
import trove.common.template as template
from trove.common import timeutils
from trove.common import utils
from trove.datastore import models as datastore_models
import trove.db.models
from trove.extensions.common import models as common_models
@ -48,8 +53,8 @@ from trove.instance.models import BaseInstance
from trove.instance.models import DBInstance
from trove.instance.models import InstanceServiceStatus
from trove.instance.models import InstanceStatus
from trove.instance.service_status import ServiceStatuses
from trove.instance.tasks import InstanceTasks
from trove import rpc
from trove.taskmanager import models as taskmanager_models
from trove.tests.unittests import trove_testtools
from trove.tests.unittests.util import util
@ -962,27 +967,20 @@ class BuiltInstanceTasksTest(trove_testtools.TestCase):
self.instance_task.demote_replication_master()
self.instance_task._guest.demote_replication_master.assert_any_call()
@patch.multiple(taskmanager_models.BuiltInstanceTasks,
get_injected_files=Mock(return_value="the-files"))
def test_upgrade(self, *args):
pre_rebuild_server = self.instance_task.server
dsv = Mock(image_id='foo_image')
mock_volume = Mock(attachments=[{'device': '/dev/mock_dev'}])
with patch.object(self.instance_task._volume_client.volumes, "get",
Mock(return_value=mock_volume)):
mock_server = Mock(status='ACTIVE')
with patch.object(self.instance_task._nova_client.servers,
'get', Mock(return_value=mock_server)):
with patch.multiple(self.instance_task._guest,
pre_upgrade=Mock(return_value={}),
post_upgrade=Mock()):
self.instance_task.upgrade(dsv)
@patch('trove.taskmanager.models.BuiltInstanceTasks.set_service_status')
@patch('trove.taskmanager.models.BuiltInstanceTasks.is_service_healthy')
@patch('trove.taskmanager.models.BuiltInstanceTasks.reset_task_status')
def test_upgrade(self, mock_resetstatus, mock_check, mock_setstatus):
dsv = MagicMock()
attrs = {'name': 'new_version'}
dsv.configure_mock(**attrs)
mock_check.return_value = True
self.instance_task._guest.pre_upgrade.return_value = {}
self.instance_task._guest.pre_upgrade.assert_called_with()
pre_rebuild_server.rebuild.assert_called_with(
dsv.image_id, files="the-files")
self.instance_task._guest.post_upgrade.assert_called_with(
mock_volume.attachments[0])
self.instance_task.upgrade(dsv)
self.instance_task._guest.upgrade.assert_called_once_with(
{'datastore_version': 'new_version'})
def test_fix_device_path(self):
self.assertEqual("/dev/vdb", self.instance_task.

4
trove/tests/unittests/taskmanager/test_vertica_clusters.py
View File

@ -16,6 +16,7 @@ import datetime
from mock import Mock
from mock import patch
from trove import rpc
from trove.cluster.models import ClusterTasks as ClusterTaskStatus
from trove.cluster.models import DBCluster
import trove.common.context as context
@ -32,8 +33,7 @@ from trove.instance.models import DBInstance
from trove.instance.models import Instance
from trove.instance.models import InstanceServiceStatus
from trove.instance.models import InstanceTasks
from trove import rpc
from trove.taskmanager.models import ServiceStatuses
from trove.instance.service_status import ServiceStatuses
from trove.tests.unittests import trove_testtools