# Copyright 2012 OpenStack Foundation # # Licensed under the Apache License, Version 2.0 (the "License"); you may # not use this file except in compliance with the License. You may obtain # a copy of the License at # # http://www.apache.org/licenses/LICENSE-2.0 # # Unless required by applicable law or agreed to in writing, software # distributed under the License is distributed on an "AS IS" BASIS, WITHOUT # WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the # License for the specific language governing permissions and limitations # under the License. import os.path import time import traceback from cinderclient import exceptions as cinder_exceptions from eventlet import greenthread from eventlet.timeout import Timeout from novaclient import exceptions as nova_exceptions from oslo_log import log as logging from oslo_utils import netutils from swiftclient.client import ClientException from trove.backup import models as bkup_models from trove.backup.models import Backup from trove.backup.models import DBBackup from trove.backup.state import BackupState from trove.cluster.models import Cluster from trove.cluster.models import DBCluster from trove.cluster import tasks from trove.common import cfg from trove.common import crypto_utils as cu from trove.common import exception from trove.common.exception import BackupCreationError from trove.common.exception import GuestError from trove.common.exception import GuestTimeout from trove.common.exception import InvalidModelError from trove.common.exception import MalformedSecurityGroupRuleError 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 import instance as rd_instance from trove.common.instance import ServiceStatuses from trove.common.notification import ( DBaaSInstanceRestart, DBaaSInstanceUpgrade, EndNotification, StartNotification, TroveInstanceCreate, TroveInstanceModifyVolume, TroveInstanceModifyFlavor, TroveInstanceDelete) import trove.common.remote as remote from trove.common.remote import create_cinder_client from trove.common.remote import create_dns_client from trove.common.remote import create_guest_client from trove.common import server_group as srv_grp from trove.common.strategies.cluster import strategy from trove.common import template from trove.common import timeutils from trove.common import utils from trove.common.utils import try_recover from trove.extensions.mysql import models as mysql_models from trove.extensions.security_group.models import SecurityGroup from trove.extensions.security_group.models import SecurityGroupRule from trove.instance import models as inst_models from trove.instance.models import BuiltInstance from trove.instance.models import DBInstance from trove.instance.models import FreshInstance from trove.instance.models import Instance from trove.instance.models import InstanceServiceStatus from trove.instance.models import InstanceStatus from trove.instance.tasks import InstanceTasks from trove.module import models as module_models from trove.module import views as module_views from trove.quota.quota import run_with_quotas from trove import rpc LOG = logging.getLogger(__name__) CONF = cfg.CONF class NotifyMixin(object): """Notification Mixin This adds the ability to send usage events to an Instance object. """ def _get_service_id(self, datastore_manager, id_map): if datastore_manager in id_map: datastore_manager_id = id_map[datastore_manager] else: datastore_manager_id = cfg.UNKNOWN_SERVICE_ID LOG.error("Datastore ID for Manager (%s) is not configured", datastore_manager) return datastore_manager_id def send_usage_event(self, event_type, **kwargs): event_type = 'trove.instance.%s' % event_type publisher_id = CONF.host # Grab the instance size from the kwargs or from the nova client instance_size = kwargs.pop('instance_size', None) flavor = self.nova_client.flavors.get(self.flavor_id) server = kwargs.pop('server', None) if server is None: server = self.nova_client.servers.get(self.server_id) az = getattr(server, 'OS-EXT-AZ:availability_zone', None) # Default payload created_time = timeutils.isotime(self.db_info.created) payload = { 'availability_zone': az, 'created_at': created_time, 'name': self.name, 'instance_id': self.id, 'instance_name': self.name, 'instance_size': instance_size or flavor.ram, 'instance_type': flavor.name, 'instance_type_id': flavor.id, 'launched_at': created_time, 'nova_instance_id': self.server_id, 'region': CONF.region, 'state_description': self.status, 'state': self.status, 'tenant_id': self.tenant_id, 'user_id': self.context.user, } if CONF.get(self.datastore_version.manager).volume_support: payload.update({ 'volume_size': self.volume_size, 'nova_volume_id': self.volume_id }) payload['service_id'] = self._get_service_id( self.datastore_version.manager, CONF.notification_service_id) # Update payload with all other kwargs payload.update(kwargs) LOG.debug('Sending event: %(event_type)s, %(payload)s', {'event_type': event_type, 'payload': payload}) notifier = rpc.get_notifier( service="taskmanager", publisher_id=publisher_id) notifier.info(self.context, event_type, payload) class ConfigurationMixin(object): """Configuration Mixin Configuration related tasks for instances and resizes. """ def _render_config(self, flavor): config = template.SingleInstanceConfigTemplate( self.datastore_version, flavor, self.id) config.render() return config def _render_replica_source_config(self, flavor): config = template.ReplicaSourceConfigTemplate( self.datastore_version, flavor, self.id) config.render() return config def _render_replica_config(self, flavor): config = template.ReplicaConfigTemplate( self.datastore_version, flavor, self.id) config.render() return config def _render_config_dict(self, flavor): config = template.SingleInstanceConfigTemplate( self.datastore_version, flavor, self.id) ret = config.render_dict() LOG.debug("the default template dict of mysqld section: %s", ret) return ret class ClusterTasks(Cluster): def update_statuses_on_failure(self, cluster_id, shard_id=None, status=None): if CONF.update_status_on_fail: if shard_id: db_instances = DBInstance.find_all(cluster_id=cluster_id, shard_id=shard_id).all() else: db_instances = DBInstance.find_all( cluster_id=cluster_id).all() for db_instance in db_instances: db_instance.set_task_status( status or InstanceTasks.BUILDING_ERROR_SERVER) db_instance.save() @classmethod def get_ip(cls, instance): return instance.get_visible_ip_addresses()[0] def _all_instances_ready(self, instance_ids, cluster_id, 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]) 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]) 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]) def _all_instances_acquire_status( self, instance_ids, cluster_id, shard_id, expected_status, fast_fail_statuses=None): def _is_fast_fail_status(status): return ((fast_fail_statuses is not None) and ((status == fast_fail_statuses) or (status in fast_fail_statuses))) def _all_have_status(ids): for instance_id in ids: status = InstanceServiceStatus.find_by( instance_id=instance_id).get_status() task_status = DBInstance.find_by( id=instance_id).get_task_status() if (_is_fast_fail_status(status) or (task_status == InstanceTasks.BUILDING_ERROR_SERVER)): # if one has failed, no need to continue polling LOG.debug("Instance %(id)s has acquired a fast-fail " "status %(status)s and" " task_status %(task_status)s.", {'id': instance_id, 'status': status, 'task_status': task_status}) return True if status != expected_status: # if one is not in the expected state, continue polling LOG.debug("Instance %(id)s was %(status)s.", {'id': instance_id, 'status': status}) return False return True def _instance_ids_with_failures(ids): LOG.debug("Checking for service failures on instances: %s", ids) failed_instance_ids = [] for instance_id in ids: status = InstanceServiceStatus.find_by( instance_id=instance_id).get_status() task_status = DBInstance.find_by( id=instance_id).get_task_status() if (_is_fast_fail_status(status) or (task_status == InstanceTasks.BUILDING_ERROR_SERVER)): failed_instance_ids.append(instance_id) return failed_instance_ids LOG.debug("Polling until all instances acquire %(expected)s " "status: %(ids)s", {'expected': expected_status, 'ids': instance_ids}) try: utils.poll_until(lambda: instance_ids, lambda ids: _all_have_status(ids), sleep_time=CONF.usage_sleep_time, time_out=CONF.usage_timeout) except PollTimeOut: LOG.exception("Timed out while waiting for all instances " "to become %s.", expected_status) self.update_statuses_on_failure(cluster_id, shard_id) return False failed_ids = _instance_ids_with_failures(instance_ids) if failed_ids: LOG.error("Some instances failed: %s", failed_ids) self.update_statuses_on_failure(cluster_id, shard_id) return False LOG.debug("All instances have acquired the expected status %s.", expected_status) return True def delete_cluster(self, context, cluster_id): LOG.debug("begin delete_cluster for id: %s", cluster_id) def all_instances_marked_deleted(): db_instances = DBInstance.find_all(cluster_id=cluster_id, deleted=False).all() return len(db_instances) == 0 try: utils.poll_until(all_instances_marked_deleted, sleep_time=2, time_out=CONF.cluster_delete_time_out) except PollTimeOut: LOG.error("timeout for instances to be marked as deleted.") return LOG.debug("setting cluster %s as deleted.", cluster_id) cluster = DBCluster.find_by(id=cluster_id) cluster.deleted = True cluster.deleted_at = timeutils.utcnow() cluster.task_status = tasks.ClusterTasks.NONE cluster.save() LOG.debug("end delete_cluster for id: %s", cluster_id) def rolling_restart_cluster(self, context, cluster_id, delay_sec=0): LOG.debug("Begin rolling cluster restart for id: %s", cluster_id) def _restart_cluster_instance(instance): LOG.debug("Restarting instance with id: %s", instance.id) context.notification = ( DBaaSInstanceRestart(context, **request_info)) with StartNotification(context, instance_id=instance.id): with EndNotification(context): instance.update_db(task_status=InstanceTasks.REBOOTING) instance.restart() timeout = Timeout(CONF.cluster_usage_timeout) cluster_notification = context.notification request_info = cluster_notification.serialize(context) try: node_db_inst = DBInstance.find_all(cluster_id=cluster_id, deleted=False).all() for index, db_inst in enumerate(node_db_inst): if index > 0: LOG.debug( "Waiting (%ds) for restarted nodes to rejoin the " "cluster before proceeding.", delay_sec) time.sleep(delay_sec) instance = BuiltInstanceTasks.load(context, db_inst.id) _restart_cluster_instance(instance) except Timeout as t: if t is not timeout: raise # not my timeout LOG.exception("Timeout for restarting cluster.") raise except Exception: LOG.exception("Error restarting cluster.", cluster_id) raise finally: context.notification = cluster_notification timeout.cancel() self.reset_task() LOG.debug("End rolling restart for id: %s.", cluster_id) def rolling_upgrade_cluster(self, context, cluster_id, datastore_version): LOG.debug("Begin rolling cluster upgrade for id: %s.", cluster_id) def _upgrade_cluster_instance(instance): LOG.debug("Upgrading instance with id: %s.", instance.id) context.notification = ( DBaaSInstanceUpgrade(context, **request_info)) with StartNotification( context, instance_id=instance.id, datastore_version_id=datastore_version.id): with EndNotification(context): instance.update_db( datastore_version_id=datastore_version.id, task_status=InstanceTasks.UPGRADING) instance.upgrade(datastore_version) timeout = Timeout(CONF.cluster_usage_timeout) cluster_notification = context.notification request_info = cluster_notification.serialize(context) try: for db_inst in DBInstance.find_all(cluster_id=cluster_id, deleted=False).all(): instance = BuiltInstanceTasks.load( context, db_inst.id) _upgrade_cluster_instance(instance) self.reset_task() except Timeout as t: if t is not timeout: raise # not my timeout LOG.exception("Timeout for upgrading cluster.") self.update_statuses_on_failure( cluster_id, status=InstanceTasks.UPGRADING_ERROR) except Exception: LOG.exception("Error upgrading cluster %s.", cluster_id) self.update_statuses_on_failure( cluster_id, status=InstanceTasks.UPGRADING_ERROR) finally: context.notification = cluster_notification timeout.cancel() LOG.debug("End upgrade_cluster for id: %s.", cluster_id) class FreshInstanceTasks(FreshInstance, NotifyMixin, ConfigurationMixin): def _delete_resources(self, deleted_at): LOG.debug("Begin _delete_resources for instance %s", self.id) # If volume has "available" status, delete it manually. try: if self.volume_id: volume_client = create_cinder_client(self.context) volume = volume_client.volumes.get(self.volume_id) if volume.status == "available": LOG.info("Deleting volume %(v)s for instance: %(i)s.", {'v': self.volume_id, 'i': self.id}) volume.delete() except Exception: LOG.exception("Error deleting volume of instance %(id)s.", {'id': self.db_info.id}) LOG.debug("End _delete_resource for instance %s", self.id) def wait_for_instance(self, timeout, flavor): # Make sure the service becomes active before sending a usage # record to avoid over billing a customer for an instance that # fails to build properly. error_message = '' error_details = '' try: utils.poll_until(self._service_is_active, sleep_time=CONF.usage_sleep_time, time_out=timeout) LOG.info("Created instance %s successfully.", self.id) TroveInstanceCreate(instance=self, instance_size=flavor['ram']).notify() except PollTimeOut as ex: LOG.error("Failed to create instance %s. " "Timeout waiting for instance to become active. " "No usage create-event was sent.", self.id) self.update_statuses_on_time_out() error_message = "%s" % ex error_details = traceback.format_exc() except Exception as ex: LOG.exception("Failed to send usage create-event for " "instance %s.", self.id) error_message = "%s" % ex error_details = traceback.format_exc() finally: if error_message: inst_models.save_instance_fault( self.id, error_message, error_details, skip_delta=CONF.usage_sleep_time + 1) def create_instance(self, flavor, image_id, databases, users, datastore_manager, packages, volume_size, backup_id, availability_zone, root_password, nics, overrides, cluster_config, snapshot, volume_type, modules, scheduler_hints): # It is the caller's responsibility to ensure that # FreshInstanceTasks.wait_for_instance is called after # create_instance to ensure that the proper usage event gets sent LOG.info("Creating instance %s.", self.id) security_groups = None if CONF.trove_security_groups_support: try: security_groups = self._create_secgroup(datastore_manager) except Exception as e: log_fmt = "Error creating security group for instance: %s" exc_fmt = _("Error creating security group for instance: %s") err = inst_models.InstanceTasks.BUILDING_ERROR_SEC_GROUP self._log_and_raise(e, log_fmt, exc_fmt, self.id, err) else: LOG.debug("Successfully created security group for " "instance: %s", self.id) files = self.get_injected_files(datastore_manager) cinder_volume_type = volume_type or CONF.cinder_volume_type volume_info = self._create_server_volume( flavor['id'], image_id, security_groups, datastore_manager, volume_size, availability_zone, nics, files, cinder_volume_type, scheduler_hints) config = self._render_config(flavor) backup_info = None if backup_id is not None: backup = bkup_models.Backup.get_by_id(self.context, backup_id) backup_info = {'id': backup_id, 'instance_id': backup.instance_id, 'location': backup.location, 'type': backup.backup_type, 'checksum': backup.checksum, } self._guest_prepare(flavor['ram'], volume_info, packages, databases, users, backup_info, config.config_contents, root_password, overrides, cluster_config, snapshot, modules) 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 # re-sending the prepare call; retrying DNS is much easier. try: self._create_dns_entry() except Exception as e: log_fmt = "Error creating DNS entry for instance: %s" exc_fmt = _("Error creating DNS entry for instance: %s") err = inst_models.InstanceTasks.BUILDING_ERROR_DNS self._log_and_raise(e, log_fmt, exc_fmt, self.id, err) def attach_replication_slave(self, snapshot, flavor): LOG.debug("Calling attach_replication_slave for %s.", self.id) try: replica_config = self._render_replica_config(flavor) self.guest.attach_replication_slave(snapshot, replica_config.config_contents) except GuestError as e: log_fmt = "Error attaching instance %s as replica." exc_fmt = _("Error attaching instance %s as replica.") err = inst_models.InstanceTasks.BUILDING_ERROR_REPLICA self._log_and_raise(e, log_fmt, exc_fmt, self.id, err) def get_replication_master_snapshot(self, context, slave_of_id, flavor, backup_id=None, replica_number=1): # First check to see if we need to take a backup master = BuiltInstanceTasks.load(context, slave_of_id) backup_required = master.backup_required_for_replication() if backup_required: # if we aren't passed in a backup id, look it up to possibly do # an incremental backup, thus saving time if not backup_id: backup = Backup.get_last_completed( context, slave_of_id, include_incremental=True) if backup: backup_id = backup.id else: LOG.debug('Skipping replication backup, as none is required.') snapshot_info = { 'name': "Replication snapshot for %s" % self.id, 'description': "Backup image used to initialize " "replication slave", 'instance_id': slave_of_id, 'parent_id': backup_id, 'tenant_id': self.tenant_id, 'state': BackupState.NEW, 'datastore_version_id': self.datastore_version.id, 'deleted': False, 'replica_number': replica_number, } replica_backup_id = None if backup_required: # Only do a backup if it's the first replica if replica_number == 1: try: db_info = DBBackup.create(**snapshot_info) replica_backup_id = db_info.id except InvalidModelError: log_fmt = ("Unable to create replication snapshot record " "for instance: %s") exc_fmt = _("Unable to create replication snapshot record " "for instance: %s") LOG.exception(log_fmt, self.id) raise BackupCreationError(exc_fmt % self.id) if backup_id: # Look up the parent backup info or fail early if not # found or if the user does not have access to the parent. _parent = Backup.get_by_id(context, backup_id) parent = { 'location': _parent.location, 'checksum': _parent.checksum, } snapshot_info.update({ 'parent': parent, }) else: # we've been passed in the actual replica backup id, # so just use it replica_backup_id = backup_id try: snapshot_info.update({ 'id': replica_backup_id, 'datastore': master.datastore.name, 'datastore_version': master.datastore_version.name, }) snapshot = master.get_replication_snapshot( snapshot_info, flavor=master.flavor_id) snapshot.update({ 'config': self._render_replica_config(flavor).config_contents }) return snapshot except Exception as e_create: create_log_fmt = ( "Error creating replication snapshot from " "instance %(source)s for new replica %(replica)s.") create_exc_fmt = ( "Error creating replication snapshot from " "instance %(source)s for new replica %(replica)s.") create_fmt_content = { 'source': slave_of_id, 'replica': self.id } err = inst_models.InstanceTasks.BUILDING_ERROR_REPLICA e_create_fault = create_log_fmt % create_fmt_content e_create_stack = traceback.format_exc() # we persist fault details to source instance inst_models.save_instance_fault(slave_of_id, e_create_fault, e_create_stack) # if the delete of the 'bad' backup fails, it'll mask the # create exception, so we trap it here try: # Only try to delete the backup if it's the first replica if replica_number == 1 and backup_required: Backup.delete(context, replica_backup_id) except Exception as e_delete: LOG.error(create_log_fmt, create_fmt_content) # Make sure we log any unexpected errors from the create if not isinstance(e_create, TroveError): LOG.exception(e_create) delete_log_fmt = ( "An error occurred while deleting a bad " "replication snapshot from instance %(source)s.") delete_exc_fmt = _( "An error occurred while deleting a bad " "replication snapshot from instance %(source)s.") # we've already logged the create exception, so we'll raise # the delete (otherwise the create will be logged twice) self._log_and_raise(e_delete, delete_log_fmt, delete_exc_fmt, {'source': slave_of_id}, err) # the delete worked, so just log the original problem with create self._log_and_raise(e_create, create_log_fmt, create_exc_fmt, create_fmt_content, err) def report_root_enabled(self): mysql_models.RootHistory.create(self.context, self.id) def update_statuses_on_time_out(self): 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.save() LOG.error("Service status: %(status)s\n" "Service error description: %(desc)s", {'status': ServiceStatuses. FAILED_TIMEOUT_GUESTAGENT.api_status, 'desc': ServiceStatuses. FAILED_TIMEOUT_GUESTAGENT.description}) # Updating instance status db_info = DBInstance.find_by(id=self.id, deleted=False) db_info.set_task_status(InstanceTasks. BUILDING_ERROR_TIMEOUT_GA) db_info.save() LOG.error("Trove instance status: %(action)s\n" "Trove instance status description: %(text)s", {'action': InstanceTasks. BUILDING_ERROR_TIMEOUT_GA.action, 'text': InstanceTasks. BUILDING_ERROR_TIMEOUT_GA.db_text}) def _service_is_active(self): """ Check that the database guest is active. This function is meant to be called with poll_until to check that the guest is alive before sending a 'create' message. This prevents over billing a customer for an instance that they can never use. Returns: boolean if the service is active. Raises: TroveError if the service is in a failure state. """ 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): return True elif status not in [rd_instance.ServiceStatuses.NEW, rd_instance.ServiceStatuses.BUILDING, rd_instance.ServiceStatuses.UNKNOWN, rd_instance.ServiceStatuses.DELETED]: raise TroveError(_("Service not active, status: %s") % status) c_id = self.db_info.compute_instance_id server = self.nova_client.servers.get(c_id) server_status = server.status if server_status in [InstanceStatus.ERROR, InstanceStatus.FAILED]: server_fault_message = 'No fault found' try: server_fault_message = server.fault.get('message', 'Unknown') except AttributeError: pass server_message = "\nServer error: %s" % server_fault_message raise TroveError(_("Server not active, status: %(status)s" "%(srv_msg)s") % {'status': server_status, 'srv_msg': server_message}) return False def _build_sg_rules_mapping(self, rule_ports): final = [] cidr = CONF.trove_security_group_rule_cidr for port_or_range in set(rule_ports): from_, to_ = port_or_range[0], port_or_range[-1] final.append({'cidr': cidr, 'from_': str(from_), 'to_': str(to_)}) return final def _create_server_volume(self, flavor_id, image_id, security_groups, datastore_manager, volume_size, availability_zone, nics, files, volume_type, scheduler_hints): LOG.debug("Begin _create_server_volume for id: %s", self.id) server = None volume_info = self._build_volume_info(datastore_manager, volume_size=volume_size, volume_type=volume_type) block_device_mapping_v2 = volume_info['block_device'] try: server = self._create_server(flavor_id, image_id, security_groups, datastore_manager, block_device_mapping_v2, availability_zone, nics, files, scheduler_hints) server_id = server.id # Save server ID. self.update_db(compute_instance_id=server_id) except Exception as e: log_fmt = "Failed to create server for instance %s" exc_fmt = _("Failed to create server for instance %s") err = inst_models.InstanceTasks.BUILDING_ERROR_SERVER self._log_and_raise(e, log_fmt, exc_fmt, self.id, err) LOG.debug("End _create_server_volume for id: %s", self.id) return volume_info def _build_volume_info(self, datastore_manager, volume_size=None, volume_type=None): volume_info = None volume_support = self.volume_support device_path = self.device_path mount_point = CONF.get(datastore_manager).mount_point LOG.debug("trove volume support = %s", volume_support) if volume_support: try: volume_info = self._create_volume( volume_size, volume_type, datastore_manager) except Exception as e: log_fmt = "Failed to create volume for instance %s" exc_fmt = _("Failed to create volume for instance %s") err = inst_models.InstanceTasks.BUILDING_ERROR_VOLUME self._log_and_raise(e, log_fmt, exc_fmt, self.id, err) else: LOG.debug("device_path = %(path)s\n" "mount_point = %(point)s", { "path": device_path, "point": mount_point }) volume_info = { 'block_device': None, 'device_path': device_path, 'mount_point': mount_point, } return volume_info # We remove all translations for messages logging execpet those for # exception raising. And we cannot use _(xxxx) instead of _("xxxx") # because of H701 PEP8 checking. So we pass log format , exception # format, and format content in and do translations only if needed. def _log_and_raise(self, exc, log_fmt, exc_fmt, fmt_content, task_status): LOG.error("%(message)s\n%(exc)s\n%(trace)s", {"message": log_fmt % fmt_content, "exc": exc, "trace": traceback.format_exc()}) self.update_db(task_status=task_status) exc_message = '\n%s' % exc if exc else '' full_message = "%s%s" % (exc_fmt % fmt_content, exc_message) raise TroveError(message=full_message) def _create_volume(self, volume_size, volume_type, datastore_manager): LOG.debug("Begin _create_volume for id: %s", self.id) volume_client = create_cinder_client(self.context, self.region_name) volume_desc = ("datastore volume for %s" % self.id) volume_ref = volume_client.volumes.create( volume_size, name="datastore-%s" % self.id, description=volume_desc, volume_type=volume_type) # Record the volume ID in case something goes wrong. self.update_db(volume_id=volume_ref.id) utils.poll_until( lambda: volume_client.volumes.get(volume_ref.id), lambda v_ref: v_ref.status in ['available', 'error'], sleep_time=2, time_out=CONF.volume_time_out) v_ref = volume_client.volumes.get(volume_ref.id) if v_ref.status in ['error']: raise VolumeCreationFailure() LOG.debug("End _create_volume for id: %s", self.id) return self._build_volume(v_ref, datastore_manager) def _build_volume(self, v_ref, datastore_manager): LOG.debug("Created volume %s", v_ref) # TODO(zhaochao): from Liberty, Nova libvirt driver does not honor # user-supplied device name anymore, so we may need find a new # method to make sure the volume is correctly mounted inside the # guest, please refer to the 'intermezzo-problem-with-device-names' # section of Nova user referrence at: # https://docs.openstack.org/nova/latest/user/block-device-mapping.html bdm = CONF.block_device_mapping # use Nova block_device_mapping_v2, referrence: # https://developer.openstack.org/api-ref/compute/#create-server # setting the delete_on_terminate instance to true=1 block_device_v2 = [{ "uuid": v_ref.id, "source_type": "volume", "destination_type": "volume", "device_name": bdm, "volume_size": v_ref.size, "delete_on_termination": True }] created_volumes = [{'id': v_ref.id, 'size': v_ref.size}] device_path = self.device_path mount_point = CONF.get(datastore_manager).mount_point LOG.debug("block_device = %(device)s\n" "volume = %(volume)s\n" "device_path = %(path)s\n" "mount_point = %(point)s", {"device": block_device_v2, "volume": created_volumes, "path": device_path, "point": mount_point}) volume_info = {'block_device': block_device_v2, 'device_path': device_path, 'mount_point': mount_point} return volume_info def _prepare_userdata(self, datastore_manager): userdata = None cloudinit = os.path.join(CONF.get('cloudinit_location'), "%s.cloudinit" % datastore_manager) if os.path.isfile(cloudinit): with open(cloudinit, "r") as f: userdata = f.read() return userdata def _create_server(self, flavor_id, image_id, security_groups, datastore_manager, block_device_mapping_v2, availability_zone, nics, files={}, scheduler_hints=None): userdata = self._prepare_userdata(datastore_manager) name = self.hostname or self.name bdmap_v2 = block_device_mapping_v2 config_drive = CONF.use_nova_server_config_drive server = self.nova_client.servers.create( name, image_id, flavor_id, files=files, userdata=userdata, security_groups=security_groups, block_device_mapping_v2=bdmap_v2, availability_zone=availability_zone, nics=nics, config_drive=config_drive, scheduler_hints=scheduler_hints) LOG.debug("Created new compute instance %(server_id)s " "for instance %(id)s", {'server_id': server.id, 'id': self.id}) return server def _guest_prepare(self, flavor_ram, volume_info, packages, databases, users, backup_info=None, config_contents=None, root_password=None, overrides=None, cluster_config=None, snapshot=None, modules=None): LOG.debug("Entering guest_prepare") # Now wait for the response from the create to do additional work self.guest.prepare(flavor_ram, packages, databases, users, device_path=volume_info['device_path'], mount_point=volume_info['mount_point'], backup_info=backup_info, config_contents=config_contents, root_password=root_password, overrides=overrides, cluster_config=cluster_config, snapshot=snapshot, modules=modules) def _create_dns_entry(self): dns_support = CONF.trove_dns_support LOG.debug("trove dns support = %s", dns_support) if dns_support: LOG.debug("%(gt)s: Creating dns entry for instance: %(id)s", {'gt': greenthread.getcurrent(), 'id': self.id}) dns_client = create_dns_client(self.context) def get_server(): c_id = self.db_info.compute_instance_id return self.nova_client.servers.get(c_id) def ip_is_available(server): LOG.debug("Polling for ip addresses: $%s ", server.addresses) if server.addresses != {}: return True elif (server.addresses == {} and server.status != InstanceStatus.ERROR): return False elif (server.addresses == {} and server.status == InstanceStatus.ERROR): LOG.error("Failed to create DNS entry for instance " "%(instance)s. Server status was " "%(status)s).", {'instance': self.id, 'status': server.status}) raise TroveError(status=server.status) utils.poll_until(get_server, ip_is_available, sleep_time=1, time_out=CONF.dns_time_out) server = self.nova_client.servers.get( self.db_info.compute_instance_id) self.db_info.addresses = server.addresses LOG.debug("Creating dns entry...") ip = self.dns_ip_address if not ip: raise TroveError(_("Failed to create DNS entry for instance " "%s. No IP available.") % self.id) dns_client.create_instance_entry(self.id, ip) LOG.debug("Successfully created DNS entry for instance: %s", self.id) else: LOG.debug("%(gt)s: DNS not enabled for instance: %(id)s", {'gt': greenthread.getcurrent(), 'id': self.id}) def _create_secgroup(self, datastore_manager): security_group = SecurityGroup.create_for_instance( self.id, self.context, self.region_name) tcp_ports = CONF.get(datastore_manager).tcp_ports udp_ports = CONF.get(datastore_manager).udp_ports icmp = CONF.get(datastore_manager).icmp self._create_rules(security_group, tcp_ports, 'tcp') self._create_rules(security_group, udp_ports, 'udp') if icmp: self._create_rules(security_group, None, 'icmp') return [security_group["name"]] def _create_rules(self, s_group, ports, protocol): err = inst_models.InstanceTasks.BUILDING_ERROR_SEC_GROUP err_msg = _("Failed to create security group rules for instance " "%(instance_id)s: Invalid port format - " "FromPort = %(from)s, ToPort = %(to)s") def set_error_and_raise(port_or_range): from_port, to_port = port_or_range self.update_db(task_status=err) msg = err_msg % {'instance_id': self.id, 'from': from_port, 'to': to_port} raise MalformedSecurityGroupRuleError(message=msg) cidr = CONF.trove_security_group_rule_cidr if protocol == 'icmp': SecurityGroupRule.create_sec_group_rule( s_group, 'icmp', None, None, cidr, self.context, self.region_name) else: for port_or_range in set(ports): try: from_, to_ = (None, None) from_, to_ = port_or_range[0], port_or_range[-1] SecurityGroupRule.create_sec_group_rule( s_group, protocol, int(from_), int(to_), cidr, self.context, self.region_name) except (ValueError, TroveError): set_error_and_raise([from_, to_]) class BuiltInstanceTasks(BuiltInstance, NotifyMixin, ConfigurationMixin): """ Performs the various asynchronous instance related tasks. """ def _delete_resources(self, deleted_at): LOG.debug("Begin _delete_resources for instance %s", self.id) server_id = self.db_info.compute_instance_id old_server = self.nova_client.servers.get(server_id) try: # The server may have already been marked as 'SHUTDOWN' # but check for 'ACTIVE' in case of any race condition # We specifically don't want to attempt to stop db if # the server is in 'ERROR' or 'FAILED" state, as it will # result in a long timeout if self.server_status_matches(['ACTIVE', 'SHUTDOWN'], server=self): LOG.debug("Stopping datastore on instance %s before deleting " "any resources.", self.id) self.guest.stop_db() except Exception: LOG.exception("Error stopping the datastore before attempting " "to delete instance id %s.", self.id) try: self.server.delete() except Exception as ex: LOG.exception("Error during delete compute server %s", self.server.id) try: dns_support = CONF.trove_dns_support LOG.debug("trove dns support = %s", dns_support) if dns_support: dns_api = create_dns_client(self.context) dns_api.delete_instance_entry(instance_id=self.db_info.id) except Exception as ex: LOG.exception("Error during dns entry of instance %(id)s: " "%(ex)s", {'id': self.db_info.id, 'ex': ex}) try: srv_grp.ServerGroup.delete(self.context, self.server_group) except Exception: LOG.exception("Error during delete server group for %s", self.id) # Poll until the server is gone. def server_is_finished(): try: server = self.nova_client.servers.get(server_id) if not self.server_status_matches(['SHUTDOWN', 'ACTIVE'], server=server): LOG.error("Server %(server_id)s entered ERROR status " "when deleting instance %(instance_id)s!", {'server_id': server.id, 'instance_id': self.id}) return False except nova_exceptions.NotFound: return True try: utils.poll_until(server_is_finished, sleep_time=2, time_out=CONF.server_delete_time_out) except PollTimeOut: LOG.exception("Failed to delete instance %(instance_id)s: " "Timeout deleting compute server %(server_id)s", {'instance_id': self.id, 'server_id': server_id}) # If volume has been resized it must be manually removed in cinder try: if self.volume_id: volume_client = create_cinder_client(self.context, self.region_name) volume = volume_client.volumes.get(self.volume_id) if volume.status == "available": LOG.info("Deleting volume %(v)s for instance: %(i)s.", {'v': self.volume_id, 'i': self.id}) volume.delete() except Exception: LOG.exception("Error deleting volume of instance %(id)s.", {'id': self.db_info.id}) TroveInstanceDelete(instance=self, deleted_at=timeutils.isotime(deleted_at), server=old_server).notify() LOG.debug("End _delete_resources for instance %s", self.id) def server_status_matches(self, expected_status, server=None): if not server: server = self.server return server.status.upper() in ( status.upper() for status in expected_status) 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.", {'instance_id': self.id, 'old_size': self.volume_size, 'new_size': new_size}) action = ResizeVolumeAction(self, self.volume_size, new_size) action.execute() LOG.info("Resized volume for instance %s successfully.", self.id) def resize_flavor(self, old_flavor, new_flavor): LOG.info("Resizing instance %(instance_id)s from flavor " "%(old_flavor)s to %(new_flavor)s.", {'instance_id': self.id, 'old_flavor': old_flavor['id'], 'new_flavor': new_flavor['id']}) action = ResizeAction(self, old_flavor, new_flavor) action.execute() LOG.info("Resized instance %s successfully.", self.id) def migrate(self, host): LOG.info("Initiating migration to host %s.", host) action = MigrateAction(self, host) action.execute() def create_backup(self, backup_info): LOG.info("Initiating backup for instance %s.", self.id) self.guest.create_backup(backup_info) def backup_required_for_replication(self): LOG.debug("Seeing if replication backup is required for instance %s.", self.id) return self.guest.backup_required_for_replication() def get_replication_snapshot(self, snapshot_info, flavor): def _get_replication_snapshot(): LOG.debug("Calling get_replication_snapshot on %s.", self.id) try: rep_source_config = self._render_replica_source_config(flavor) result = self.guest.get_replication_snapshot( snapshot_info, rep_source_config.config_contents) LOG.debug("Got replication snapshot from guest successfully.") return result except Exception: LOG.exception("Failed to get replication snapshot from %s.", self.id) raise return run_with_quotas(self.context.tenant, {'backups': 1}, _get_replication_snapshot) def detach_replica(self, master, for_failover=False): LOG.debug("Calling detach_replica on %s", self.id) try: self.guest.detach_replica(for_failover) self.update_db(slave_of_id=None) self.slave_list = None except (GuestError, GuestTimeout): LOG.exception("Failed to detach replica %s.", self.id) raise finally: if not for_failover: self.reset_task_status() def attach_replica(self, master): LOG.debug("Calling attach_replica on %s", self.id) try: replica_info = master.guest.get_replica_context() flavor = self.nova_client.flavors.get(self.flavor_id) slave_config = self._render_replica_config(flavor).config_contents self.guest.attach_replica(replica_info, slave_config) self.update_db(slave_of_id=master.id) self.slave_list = None except (GuestError, GuestTimeout): LOG.exception("Failed to attach replica %s.", self.id) raise def make_read_only(self, read_only): LOG.debug("Calling make_read_only on %s", self.id) self.guest.make_read_only(read_only) def _get_floating_ips(self): """Returns floating ips as a dict indexed by the ip.""" floating_ips = {} network_floating_ips = self.neutron_client.list_floatingips() for ip in network_floating_ips.get('floatingips'): floating_ips.update( {ip.get('floating_ip_address'): ip.get('id')}) LOG.debug("In _get_floating_ips(), returning %s", floating_ips) return floating_ips def detach_public_ips(self): LOG.debug("Begin detach_public_ips for instance %s", self.id) removed_ips = [] floating_ips = self._get_floating_ips() for ip in self.get_visible_ip_addresses(): if ip in floating_ips: fip_id = floating_ips[ip] self.neutron_client.update_floatingip( fip_id, {'floatingip': {'port_id': None}}) removed_ips.append(fip_id) return removed_ips def attach_public_ips(self, ips): LOG.debug("Begin attach_public_ips for instance %s", self.id) server_id = self.db_info.compute_instance_id # NOTE(zhaochao): in Nova's addFloatingIp, the new floating ip will # always be associated with the first IPv4 fixed address of the Nova # instance, we're doing the same thing here, after add_floating_ip is # removed from novaclient. server_ports = (self.neutron_client.list_ports(device_id=server_id) .get('ports')) fixed_address, port_id = next( (fixed_ip['ip_address'], port['id']) for port in server_ports for fixed_ip in port.get('fixed_ips') if netutils.is_valid_ipv4(fixed_ip['ip_address'])) for fip_id in ips: self.neutron_client.update_floatingip( fip_id, {'floatingip': { 'port_id': port_id, 'fixed_ip_address': fixed_address}}) def enable_as_master(self): LOG.debug("Calling enable_as_master on %s", self.id) flavor = self.nova_client.flavors.get(self.flavor_id) replica_source_config = self._render_replica_source_config(flavor) self.update_db(slave_of_id=None) self.slave_list = None self.guest.enable_as_master(replica_source_config.config_contents) def get_last_txn(self): LOG.debug("Calling get_last_txn on %s", self.id) return self.guest.get_last_txn() def get_latest_txn_id(self): LOG.debug("Calling get_latest_txn_id on %s", self.id) return self.guest.get_latest_txn_id() def wait_for_txn(self, txn): LOG.debug("Calling wait_for_txn on %s", self.id) if txn: self.guest.wait_for_txn(txn) def cleanup_source_on_replica_detach(self, replica_info): LOG.debug("Calling cleanup_source_on_replica_detach on %s", self.id) self.guest.cleanup_source_on_replica_detach(replica_info) def demote_replication_master(self): LOG.debug("Calling demote_replication_master on %s", self.id) self.guest.demote_replication_master() def reboot(self): try: # Issue a guest stop db call to shutdown the db if running LOG.debug("Stopping datastore on instance %s.", self.id) try: self.guest.stop_db() except (exception.GuestError, exception.GuestTimeout) as e: # Acceptable to be here if db was already in crashed state # Also we check guest state before issuing reboot LOG.debug(str(e)) self._refresh_datastore_status() if not (self.datastore_status_matches( rd_instance.ServiceStatuses.SHUTDOWN) or self.datastore_status_matches( rd_instance.ServiceStatuses.CRASHED)): # We will bail if db did not get stopped or is blocked LOG.error("Cannot reboot instance. DB status is %s.", self.datastore_status.status) return LOG.debug("The guest service status is %s.", self.datastore_status.status) LOG.info("Rebooting instance %s.", self.id) self.server.reboot() # Poll nova until instance is active reboot_time_out = CONF.reboot_time_out def update_server_info(): self.refresh_compute_server_info() return self.server_status_matches(['ACTIVE']) utils.poll_until( update_server_info, sleep_time=2, time_out=reboot_time_out) # Set the status to PAUSED. The guest agent will reset the status # when the reboot completes and MySQL is running. self.set_datastore_status_to_paused() LOG.info("Rebooted instance %s successfully.", self.id) except Exception as e: LOG.error("Failed to reboot instance %(id)s: %(e)s", {'id': self.id, 'e': str(e)}) finally: LOG.debug("Rebooting FINALLY %s", self.id) self.reset_task_status() def restart(self): LOG.info("Initiating datastore restart on instance %s.", self.id) try: self.guest.restart() except GuestError: LOG.error("Failed to initiate datastore restart on instance " "%s.", self.id) finally: self.reset_task_status() def guest_log_list(self): LOG.info("Retrieving guest log list for instance %s.", self.id) try: return self.guest.guest_log_list() except GuestError: LOG.error("Failed to retrieve guest log list for instance " "%s.", self.id) finally: self.reset_task_status() def guest_log_action(self, log_name, enable, disable, publish, discard): LOG.info("Processing guest log for instance %s.", self.id) try: return self.guest.guest_log_action(log_name, enable, disable, publish, discard) except GuestError: LOG.error("Failed to process guest log for instance %s.", self.id) finally: self.reset_task_status() def refresh_compute_server_info(self): """Refreshes the compute server field.""" server = self.nova_client.servers.get(self.server.id) self.server = server def _refresh_datastore_status(self): """ Gets the latest instance service status from datastore and updates the reference on this BuiltInstanceTask reference """ self.datastore_status = InstanceServiceStatus.find_by( instance_id=self.id) def set_datastore_status_to_paused(self): """ Updates the InstanceServiceStatus for this BuiltInstance to PAUSED. 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.save() def upgrade(self, datastore_version): LOG.debug("Upgrading instance %s to new datastore version %s", self, datastore_version) def server_finished_rebuilding(): self.refresh_compute_server_info() return not self.server_status_matches(['REBUILD']) try: upgrade_info = self.guest.pre_upgrade() 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=2, 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}) self.guest.post_upgrade(upgrade_info) self.reset_task_status() except Exception as e: LOG.exception(e) err = inst_models.InstanceTasks.BUILDING_ERROR_SERVER self.update_db(task_status=err) raise e # Some cinder drivers appear to return "vdb" instead of "/dev/vdb". # We need to account for that. def _fix_device_path(self, device): if device.startswith("/dev"): return device else: return "/dev/%s" % device class BackupTasks(object): @classmethod def _parse_manifest(cls, manifest): # manifest is in the format 'container/prefix' # where prefix can be 'path' or 'lots/of/paths' try: container_index = manifest.index('/') prefix_index = container_index + 1 except ValueError: return None, None container = manifest[:container_index] prefix = manifest[prefix_index:] return container, prefix @classmethod def delete_files_from_swift(cls, context, filename): container = CONF.backup_swift_container client = remote.create_swift_client(context) obj = client.head_object(container, filename) if 'x-static-large-object' in obj: # Static large object LOG.debug("Deleting large object file: %(cont)s/%(filename)s", {'cont': container, 'filename': filename}) client.delete_object(container, filename, query_string='multipart-manifest=delete') else: # Single object LOG.debug("Deleting object file: %(cont)s/%(filename)s", {'cont': container, 'filename': filename}) client.delete_object(container, filename) @classmethod def delete_backup(cls, context, backup_id): """Delete backup from swift.""" LOG.info("Deleting backup %s.", backup_id) backup = bkup_models.Backup.get_by_id(context, backup_id) try: filename = backup.filename if filename: BackupTasks.delete_files_from_swift(context, filename) except ValueError: backup.delete() except ClientException as e: if e.http_status == 404: # Backup already deleted in swift backup.delete() else: LOG.exception("Error occurred when deleting from swift. " "Details: %s", e) backup.state = bkup_models.BackupState.DELETE_FAILED backup.save() raise TroveError(_("Failed to delete swift object for backup " "%s.") % backup_id) else: backup.delete() LOG.info("Deleted backup %s successfully.", backup_id) class ModuleTasks(object): @classmethod def reapply_module(cls, context, module_id, md5, include_clustered, batch_size, batch_delay, force): """Reapply module.""" LOG.info("Reapplying module %s.", module_id) batch_size = batch_size or CONF.module_reapply_max_batch_size batch_delay = batch_delay or CONF.module_reapply_min_batch_delay # Don't let non-admin bypass the safeguards if not context.is_admin: batch_size = min(batch_size, CONF.module_reapply_max_batch_size) batch_delay = max(batch_delay, CONF.module_reapply_min_batch_delay) modules = module_models.Modules.load_by_ids(context, [module_id]) current_md5 = modules[0].md5 LOG.debug("MD5: %(md5)s Force: %(f)s.", {'md5': md5, 'f': force}) # Process all the instances instance_modules = module_models.InstanceModules.load_all( context, module_id=module_id, md5=md5) total_count = instance_modules.count() reapply_count = 0 skipped_count = 0 if instance_modules: module_list = module_views.convert_modules_to_list(modules) for instance_module in instance_modules: instance_id = instance_module.instance_id if (instance_module.md5 != current_md5 or force) and ( not md5 or md5 == instance_module.md5): instance = BuiltInstanceTasks.load(context, instance_id, needs_server=False) if instance and ( include_clustered or not instance.cluster_id): try: module_models.Modules.validate( modules, instance.datastore.id, instance.datastore_version.id) client = create_guest_client(context, instance_id) client.module_apply(module_list) Instance.add_instance_modules( context, instance_id, modules) reapply_count += 1 except exception.ModuleInvalid as ex: LOG.info("Skipping: %s", ex) skipped_count += 1 # Sleep if we've fired off too many in a row. if (batch_size and not reapply_count % batch_size and (reapply_count + skipped_count) < total_count): LOG.debug("Applied module to %(cnt)d of %(total)d " "instances - sleeping for %(batch)ds", {'cnt': reapply_count, 'total': total_count, 'batch': batch_delay}) time.sleep(batch_delay) else: LOG.debug("Instance '%s' not found or doesn't match " "criteria, skipping reapply.", instance_id) skipped_count += 1 else: LOG.debug("Instance '%s' does not match " "criteria, skipping reapply.", instance_id) skipped_count += 1 LOG.info("Reapplied module to %(num)d instances " "(skipped %(skip)d).", {'num': reapply_count, 'skip': skipped_count}) class ResizeVolumeAction(object): """Performs volume resize action.""" def __init__(self, instance, old_size, new_size): self.instance = instance self.old_size = int(old_size) self.new_size = int(new_size) def get_mount_point(self): mount_point = CONF.get( self.instance.datastore_version.manager).mount_point return mount_point def get_device_path(self): return self.instance.device_path def _fail(self, orig_func): LOG.exception("%(func)s encountered an error when " "attempting to resize the volume for " "instance %(id)s. Setting service " "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.save() def _recover_restart(self, orig_func): LOG.exception("%(func)s encountered an error when attempting to " "resize the volume for instance %(id)s. Trying to " "recover by restarting the " "guest.", {'func': orig_func.__name__, 'id': self.instance.id}) self.instance.restart() def _recover_mount_restart(self, orig_func): LOG.exception("%(func)s encountered an error when attempting to " "resize the volume for instance %(id)s. Trying to " "recover by mounting the volume and then restarting " "the guest.", {'func': orig_func.__name__, 'id': self.instance.id}) self._mount_volume() self.instance.restart() def _recover_full(self, orig_func): LOG.exception("%(func)s encountered an error when attempting to " "resize the volume for instance %(id)s. Trying to " "recover by attaching and" " mounting the volume and then restarting the " "guest.", {'func': orig_func.__name__, 'id': self.instance.id}) self._attach_volume() self._mount_volume() self.instance.restart() def _stop_db(self): LOG.debug("Instance %s calling stop_db.", self.instance.id) self.instance.guest.stop_db() @try_recover def _unmount_volume(self): LOG.debug("Unmounting the volume on instance %(id)s", { 'id': self.instance.id}) mount_point = self.get_mount_point() device_path = self.get_device_path() self.instance.guest.unmount_volume(device_path=device_path, mount_point=mount_point) LOG.debug("Successfully unmounted the volume %(vol_id)s for " "instance %(id)s", {'vol_id': self.instance.volume_id, 'id': self.instance.id}) @try_recover def _detach_volume(self): LOG.debug("Detach volume %(vol_id)s from instance %(id)s", { 'vol_id': self.instance.volume_id, 'id': self.instance.id}) self.instance.nova_client.volumes.delete_server_volume( self.instance.server.id, self.instance.volume_id) def volume_available(): volume = self.instance.volume_client.volumes.get( self.instance.volume_id) return volume.status == 'available' utils.poll_until(volume_available, sleep_time=2, time_out=CONF.volume_time_out) LOG.debug("Successfully detached volume %(vol_id)s from instance " "%(id)s", {'vol_id': self.instance.volume_id, 'id': self.instance.id}) @try_recover def _attach_volume(self): device_path = self.get_device_path() LOG.debug("Attach volume %(vol_id)s to instance %(id)s at " "%(dev)s", {'vol_id': self.instance.volume_id, 'id': self.instance.id, 'dev': device_path}) self.instance.nova_client.volumes.create_server_volume( self.instance.server.id, self.instance.volume_id, device_path) def volume_in_use(): volume = self.instance.volume_client.volumes.get( self.instance.volume_id) return volume.status == 'in-use' utils.poll_until(volume_in_use, sleep_time=2, time_out=CONF.volume_time_out) LOG.debug("Successfully attached volume %(vol_id)s to instance " "%(id)s", {'vol_id': self.instance.volume_id, 'id': self.instance.id}) @try_recover def _resize_fs(self): LOG.debug("Resizing the filesystem for instance %(id)s", { 'id': self.instance.id}) mount_point = self.get_mount_point() device_path = self.get_device_path() self.instance.guest.resize_fs(device_path=device_path, mount_point=mount_point) LOG.debug("Successfully resized volume %(vol_id)s filesystem for " "instance %(id)s", {'vol_id': self.instance.volume_id, 'id': self.instance.id}) @try_recover def _mount_volume(self): LOG.debug("Mount the volume on instance %(id)s", { 'id': self.instance.id}) mount_point = self.get_mount_point() device_path = self.get_device_path() self.instance.guest.mount_volume(device_path=device_path, mount_point=mount_point) LOG.debug("Successfully mounted the volume %(vol_id)s on instance " "%(id)s", {'vol_id': self.instance.volume_id, 'id': self.instance.id}) @try_recover def _extend(self): LOG.debug("Extending volume %(vol_id)s for instance %(id)s to " "size %(size)s", {'vol_id': self.instance.volume_id, 'id': self.instance.id, 'size': self.new_size}) self.instance.volume_client.volumes.extend(self.instance.volume_id, self.new_size) LOG.debug("Successfully extended the volume %(vol_id)s for instance " "%(id)s", {'vol_id': self.instance.volume_id, 'id': self.instance.id}) def _verify_extend(self): try: volume = self.instance.volume_client.volumes.get( self.instance.volume_id) if not volume: msg = (_('Failed to get volume %(vol_id)s') % { 'vol_id': self.instance.volume_id}) raise cinder_exceptions.ClientException(msg) def volume_is_new_size(): volume = self.instance.volume_client.volumes.get( self.instance.volume_id) return volume.size == self.new_size utils.poll_until(volume_is_new_size, sleep_time=2, time_out=CONF.volume_time_out) self.instance.update_db(volume_size=self.new_size) except PollTimeOut: LOG.exception("Timeout trying to extend the volume %(vol_id)s " "for instance %(id)s", {'vol_id': self.instance.volume_id, 'id': self.instance.id}) volume = self.instance.volume_client.volumes.get( self.instance.volume_id) if volume.status == 'extending': self._fail(self._verify_extend) elif volume.size != self.new_size: self.instance.update_db(volume_size=volume.size) self._recover_full(self._verify_extend) raise except Exception: LOG.exception("Error encountered trying to verify extend for " "the volume %(vol_id)s for instance %(id)s", {'vol_id': self.instance.volume_id, 'id': self.instance.id}) self._recover_full(self._verify_extend) raise def _resize_active_volume(self): LOG.debug("Begin _resize_active_volume for id: %(id)s", { 'id': self.instance.id}) self._stop_db() self._unmount_volume(recover_func=self._recover_restart) self._detach_volume(recover_func=self._recover_mount_restart) self._extend(recover_func=self._recover_full) self._verify_extend() # if anything fails after this point, recovery is futile self._attach_volume(recover_func=self._fail) self._resize_fs(recover_func=self._fail) self._mount_volume(recover_func=self._fail) self.instance.restart() LOG.debug("End _resize_active_volume for id: %(id)s", { 'id': self.instance.id}) def execute(self): LOG.debug("%(gt)s: Resizing instance %(id)s volume for server " "%(server_id)s from %(old_volume_size)s to " "%(new_size)r GB", {'gt': greenthread.getcurrent(), 'id': self.instance.id, 'server_id': self.instance.server.id, 'old_volume_size': self.old_size, 'new_size': self.new_size}) if self.instance.server.status == InstanceStatus.ACTIVE: self._resize_active_volume() self.instance.reset_task_status() # send usage event for size reported by cinder volume = self.instance.volume_client.volumes.get( self.instance.volume_id) launched_time = timeutils.isotime(self.instance.updated) modified_time = timeutils.isotime(self.instance.updated) TroveInstanceModifyVolume(instance=self.instance, old_volume_size=self.old_size, launched_at=launched_time, modify_at=modified_time, volume_size=volume.size, ).notify() else: self.instance.reset_task_status() msg = _("Failed to resize instance %(id)s volume for server " "%(server_id)s. The instance must be in state %(state)s " "not %(inst_state)s.") % { 'id': self.instance.id, 'server_id': self.instance.server.id, 'state': InstanceStatus.ACTIVE, 'inst_state': self.instance.server.status} raise TroveError(msg) class ResizeActionBase(object): """Base class for executing a resize action.""" def __init__(self, instance): """ Creates a new resize action for a given instance :param instance: reference to existing instance that will be resized :type instance: trove.taskmanager.models.BuiltInstanceTasks """ self.instance = instance def _assert_guest_is_ok(self): # The guest will never set the status to PAUSED. self.instance.set_datastore_status_to_paused() # Now we wait until it sets it to anything at all, # so we know it's alive. utils.poll_until( self._guest_is_awake, sleep_time=2, time_out=CONF.resize_time_out) def _assert_nova_status_is_ok(self): # Make sure Nova thinks things went well. if not self.instance.server_status_matches(["VERIFY_RESIZE"]): msg = "Migration failed! status=%(act_status)s and " \ "not %(exp_status)s" % { "act_status": self.instance.server.status, "exp_status": 'VERIFY_RESIZE'} raise TroveError(msg) def _assert_datastore_is_ok(self): # Tell the guest to turn on datastore, and ensure the status becomes # RUNNING. self._start_datastore() utils.poll_until( self._datastore_is_online, sleep_time=2, time_out=CONF.resize_time_out) def _assert_datastore_is_offline(self): # Tell the guest to turn off MySQL, and ensure the status becomes # SHUTDOWN. self.instance.guest.stop_db(do_not_start_on_reboot=True) utils.poll_until( self._datastore_is_offline, sleep_time=2, time_out=CONF.resize_time_out) def _assert_processes_are_ok(self): """Checks the procs; if anything is wrong, reverts the operation.""" # Tell the guest to turn back on, and make sure it can start. self._assert_guest_is_ok() LOG.debug("Nova guest is ok.") self._assert_datastore_is_ok() LOG.debug("Datastore is ok.") def _confirm_nova_action(self): LOG.debug("Instance %s calling Compute confirm resize...", self.instance.id) self.instance.server.confirm_resize() def _datastore_is_online(self): self.instance._refresh_datastore_status() return self.instance.is_datastore_running def _datastore_is_offline(self): self.instance._refresh_datastore_status() return (self.instance.datastore_status_matches( rd_instance.ServiceStatuses.SHUTDOWN)) def _revert_nova_action(self): LOG.debug("Instance %s calling Compute revert resize...", self.instance.id) self.instance.server.revert_resize() def execute(self): """Initiates the action.""" try: LOG.debug("Instance %s calling stop_db...", self.instance.id) self._assert_datastore_is_offline() self._perform_nova_action() finally: if self.instance.db_info.task_status != ( inst_models.InstanceTasks.NONE): self.instance.reset_task_status() def _guest_is_awake(self): self.instance._refresh_datastore_status() return not self.instance.datastore_status_matches( rd_instance.ServiceStatuses.PAUSED) def _perform_nova_action(self): """Calls Nova to resize or migrate an instance, and confirms.""" LOG.debug("Begin resize method _perform_nova_action instance: %s", self.instance.id) need_to_revert = False try: LOG.debug("Initiating nova action") self._initiate_nova_action() LOG.debug("Waiting for nova action") self._wait_for_nova_action() LOG.debug("Asserting nova status is ok") self._assert_nova_status_is_ok() need_to_revert = True LOG.debug("* * * REVERT BARRIER PASSED * * *") LOG.debug("Asserting nova action success") self._assert_nova_action_was_successful() LOG.debug("Asserting processes are OK") self._assert_processes_are_ok() LOG.debug("Confirming nova action") self._confirm_nova_action() except Exception: LOG.exception("Exception during nova action.") if need_to_revert: LOG.error("Reverting action for instance %s", self.instance.id) self._revert_nova_action() self._wait_for_revert_nova_action() if self.instance.server_status_matches(['ACTIVE']): LOG.error("Restarting datastore.") self.instance.guest.restart() else: LOG.error("Cannot restart datastore because " "Nova server status is not ACTIVE") LOG.error("Error resizing instance %s.", self.instance.id) raise LOG.debug("Recording success") self._record_action_success() LOG.debug("End resize method _perform_nova_action instance: %s", self.instance.id) def _wait_for_nova_action(self): # Wait for the flavor to change. def update_server_info(): self.instance.refresh_compute_server_info() return not self.instance.server_status_matches(['RESIZE']) utils.poll_until( update_server_info, sleep_time=2, time_out=CONF.resize_time_out) def _wait_for_revert_nova_action(self): # Wait for the server to return to ACTIVE after revert. def update_server_info(): self.instance.refresh_compute_server_info() return self.instance.server_status_matches(['ACTIVE']) utils.poll_until( update_server_info, sleep_time=2, time_out=CONF.revert_time_out) class ResizeAction(ResizeActionBase): def __init__(self, instance, old_flavor, new_flavor): """ :type instance: trove.taskmanager.models.BuiltInstanceTasks :type old_flavor: dict :type new_flavor: dict """ super(ResizeAction, self).__init__(instance) self.old_flavor = old_flavor self.new_flavor = new_flavor self.new_flavor_id = new_flavor['id'] def _assert_nova_action_was_successful(self): # Do check to make sure the status and flavor id are correct. if str(self.instance.server.flavor['id']) != str(self.new_flavor_id): msg = "Assertion failed! flavor_id=%s and not %s" \ % (self.instance.server.flavor['id'], self.new_flavor_id) raise TroveError(msg) def _initiate_nova_action(self): self.instance.server.resize(self.new_flavor_id) def _revert_nova_action(self): LOG.debug("Instance %s calling Compute revert resize... " "Repairing config.", self.instance.id) try: config = self.instance._render_config(self.old_flavor) config = {'config_contents': config.config_contents} self.instance.guest.reset_configuration(config) except GuestTimeout: LOG.exception("Error sending reset_configuration call.") LOG.debug("Reverting resize.") super(ResizeAction, self)._revert_nova_action() def _record_action_success(self): LOG.debug("Updating instance %(id)s to flavor_id %(flavor_id)s.", {'id': self.instance.id, 'flavor_id': self.new_flavor_id}) self.instance.update_db(flavor_id=self.new_flavor_id, task_status=inst_models.InstanceTasks.NONE) update_time = timeutils.isotime(self.instance.updated) TroveInstanceModifyFlavor(instance=self.instance, old_instance_size=self.old_flavor['ram'], instance_size=self.new_flavor['ram'], launched_at=update_time, modify_at=update_time, server=self.instance.server).notify() def _start_datastore(self): config = self.instance._render_config(self.new_flavor) self.instance.guest.start_db_with_conf_changes(config.config_contents) class MigrateAction(ResizeActionBase): def __init__(self, instance, host=None): super(MigrateAction, self).__init__(instance) self.instance = instance self.host = host def _assert_nova_action_was_successful(self): LOG.debug("Currently no assertions for a Migrate Action") def _initiate_nova_action(self): LOG.debug("Migrating instance %(instance)s without flavor change ...\n" "Forcing migration to host(%(host)s)", {"instance": self.instance.id, "host": self.host}) self.instance.server.migrate(force_host=self.host) def _record_action_success(self): LOG.debug("Successfully finished Migration to " "%(hostname)s: %(id)s", {'hostname': self.instance.hostname, 'id': self.instance.id}) def _start_datastore(self): self.instance.guest.restart() def load_cluster_tasks(context, cluster_id): manager = Cluster.manager_from_cluster_id(context, cluster_id) strat = strategy.load_taskmanager_strategy(manager) task_manager_cluster_tasks_class = strat.task_manager_cluster_tasks_class return ClusterTasks.load(context, cluster_id, task_manager_cluster_tasks_class)