service.py

# Copyright 2010 United States Government as represented by the
# Administrator of the National Aeronautics and Space Administration.
# Copyright 2011 Justin Santa Barbara
# All Rights Reserved.
#
#    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.

"""Generic Node base class for all workers that run on hosts."""


import inspect
import os
import random
import subprocess
import sys
import time
from typing import Optional

from oslo_concurrency import processutils
from oslo_config import cfg
from oslo_db import exception as db_exc
from oslo_log import log as logging
import oslo_messaging as messaging
from oslo_service import service
from oslo_service import wsgi
from oslo_utils import importutils
osprofiler_initializer = importutils.try_import('osprofiler.initializer')
profiler = importutils.try_import('osprofiler.profiler')
profiler_opts = importutils.try_import('osprofiler.opts')


from cinder.common import constants
from cinder import context
from cinder import coordination
from cinder import exception
from cinder.i18n import _
from cinder import objects
from cinder.objects import base as objects_base
from cinder.objects import fields
from cinder import rpc
from cinder import version
from cinder.volume import volume_utils

if os.name == 'nt':
    from os_win import utilsfactory as os_win_utilsfactory
else:
    os_win_utilsfactory = None


LOG = logging.getLogger(__name__)

service_opts = [
    cfg.IntOpt('report_interval',
               default=10,
               help='Interval, in seconds, between nodes reporting state '
                    'to datastore'),
    cfg.IntOpt('periodic_interval',
               default=60,
               help='Interval, in seconds, between running periodic tasks'),
    cfg.IntOpt('periodic_fuzzy_delay',
               default=60,
               help='Range, in seconds, to randomly delay when starting the'
                    ' periodic task scheduler to reduce stampeding.'
                    ' (Disable by setting to 0)'),
    cfg.StrOpt('osapi_volume_listen',
               default="0.0.0.0",
               help='IP address on which OpenStack Volume API listens'),
    cfg.PortOpt('osapi_volume_listen_port',
                default=8776,
                help='Port on which OpenStack Volume API listens'),
    cfg.IntOpt('osapi_volume_workers',
               help='Number of workers for OpenStack Volume API service. '
                    'The default is equal to the number of CPUs available.'),
    cfg.BoolOpt('osapi_volume_use_ssl',
                default=False,
                help='Wraps the socket in a SSL context if True is set. '
                     'A certificate file and key file must be specified.'), ]


CONF = cfg.CONF
CONF.register_opts(service_opts)
if profiler_opts:
    profiler_opts.set_defaults(CONF)


def setup_profiler(binary: str, host: str) -> None:
    if (osprofiler_initializer is None or
            profiler is None or
            profiler_opts is None):
        LOG.debug('osprofiler is not present')
        return

    if CONF.profiler.enabled:
        osprofiler_initializer.init_from_conf(
            conf=CONF,
            context=context.get_admin_context().to_dict(),
            project="cinder",
            service=binary,
            host=host
        )
        LOG.warning(
            "OSProfiler is enabled.\nIt means that person who knows "
            "any of hmac_keys that are specified in "
            "/etc/cinder/cinder.conf can trace his requests. \n"
            "In real life only operator can read this file so there "
            "is no security issue. Note that even if person can "
            "trigger profiler, only admin user can retrieve trace "
            "information.\n"
            "To disable OSProfiler set in cinder.conf:\n"
            "[profiler]\nenabled=false")


class Service(service.Service):
    """Service object for binaries running on hosts.

    A service takes a manager and enables rpc by listening to queues based
    on topic. It also periodically runs tasks on the manager and reports
    it state to the database services table.
    """
    # Make service_id a class attribute so it can be used for clean up
    service_id = None

    def __init__(self, host: str, binary: str, topic: str,
                 manager: str,
                 report_interval: Optional[int] = None,
                 periodic_interval: Optional[int] = None,
                 periodic_fuzzy_delay: Optional[int] = None,
                 service_name: Optional[str] = None,
                 coordination: bool = False,
                 cluster: Optional[str] = None,
                 *args, **kwargs):
        super(Service, self).__init__()

        if not rpc.initialized():
            rpc.init(CONF)

        self.cluster = cluster
        self.host = host
        self.binary = binary
        self.topic = topic
        self.manager_class_name = manager
        self.coordination = coordination
        manager_class = importutils.import_class(self.manager_class_name)
        if CONF.profiler.enabled:
            manager_class = profiler.trace_cls("rpc")(manager_class)

        self.service = None
        self.manager = manager_class(host=self.host,
                                     cluster=self.cluster,
                                     service_name=service_name,
                                     *args, **kwargs)
        self.availability_zone: str = self.manager.availability_zone
        self.model_disconnected: bool

        # NOTE(geguileo): We need to create the Service DB entry before we
        # create the manager, otherwise capped versions for serializer and rpc
        # client would use existing DB entries not including us, which could
        # result in us using None (if it's the first time the service is run)
        # or an old version (if this is a normal upgrade of a single service).
        ctxt = context.get_admin_context()
        try:
            service_ref = objects.Service.get_by_args(ctxt, host, binary)
            service_ref.rpc_current_version = manager_class.RPC_API_VERSION
            obj_version = objects_base.OBJ_VERSIONS.get_current()
            service_ref.object_current_version = obj_version

            # added_to_cluster attribute marks when we consider that we have
            # just added a host to a cluster so we can include resources into
            # that cluster.  We consider that we have added the host when we
            # didn't have data in the cluster DB field and our current
            # configuration has a cluster value.  We don't want to do anything
            # automatic if the cluster is changed, in those cases we'll want
            # to use cinder manage command and to it manually.
            self.added_to_cluster = (not service_ref.cluster_name and cluster)

            if service_ref.cluster_name != cluster:
                LOG.info('This service has been moved from cluster '
                         '%(cluster_svc)s to %(cluster_cfg)s. Resources '
                         'will %(opt_no)sbe moved to the new cluster',
                         {'cluster_svc': service_ref.cluster_name,
                          'cluster_cfg': cluster,
                          'opt_no': '' if self.added_to_cluster else 'NOT '})

            if self.added_to_cluster:
                # We pass copy service's disable status in the cluster if we
                # have to create it.
                self._ensure_cluster_exists(ctxt, service_ref)
                service_ref.cluster_name = cluster
            service_ref.save()
            Service.service_id = service_ref.id
            self.origin_service_id = service_ref.id
        except exception.NotFound:
            self._create_service_ref(ctxt, manager_class.RPC_API_VERSION)
            # Service entry Entry didn't exist because it was manually removed
            # or it's the first time running, to be on the safe side we say we
            # were added if we are clustered.
            self.added_to_cluster = bool(cluster)  # type: ignore

        self.report_interval = report_interval
        self.periodic_interval = periodic_interval
        self.periodic_fuzzy_delay = periodic_fuzzy_delay
        self.basic_config_check()
        self.saved_args, self.saved_kwargs = args, kwargs

        setup_profiler(binary, host)
        self.rpcserver: Optional['messaging.rpc.RPCServer'] = None
        self.backend_rpcserver: Optional['messaging.rpc.RPCServer'] = None
        self.cluster_rpcserver: Optional['messaging.rpc.RPCServer'] = None

    def start(self) -> None:
        version_string = version.version_string()
        LOG.info('Starting %(topic)s node (version %(version_string)s)',
                 {'topic': self.topic, 'version_string': version_string})
        self.model_disconnected = False

        if self.coordination:
            coordination.COORDINATOR.start()

        # NOTE(yikun): When re-spawning child process, we should set the class
        # attribute back using the origin service_id, otherwise,
        # the Service.service_id will be inherited from the parent process,
        # and will be recorded as the last started service id by mistaken.
        Service.service_id = self.origin_service_id

        self.manager.init_host(added_to_cluster=self.added_to_cluster,
                               service_id=Service.service_id)

        LOG.debug("Creating RPC server for service %s", self.topic)

        ctxt = context.get_admin_context()
        endpoints = [self.manager]
        endpoints.extend(self.manager.additional_endpoints)
        obj_version_cap = objects.Service.get_minimum_obj_version(ctxt)
        LOG.debug("Pinning object versions for RPC server serializer to %s",
                  obj_version_cap)
        serializer = objects_base.CinderObjectSerializer(obj_version_cap)

        target = messaging.Target(topic=self.topic, server=self.host)
        self.rpcserver = rpc.get_server(target, endpoints, serializer)
        self.rpcserver.start()

        # NOTE(dulek): Kids, don't do that at home. We're relying here on
        # oslo.messaging implementation details to keep backward compatibility
        # with pre-Ocata services. This will not matter once we drop
        # compatibility with them.
        if self.topic == constants.VOLUME_TOPIC:
            target = messaging.Target(
                topic='%(topic)s.%(host)s' % {'topic': self.topic,
                                              'host': self.host},
                server=volume_utils.extract_host(self.host, 'host'))
            self.backend_rpcserver = rpc.get_server(target, endpoints,
                                                    serializer)
            self.backend_rpcserver.start()

        if self.cluster:
            LOG.info('Starting %(topic)s cluster %(cluster)s (version '
                     '%(version)s)',
                     {'topic': self.topic, 'version': version_string,
                      'cluster': self.cluster})
            target = messaging.Target(
                topic='%s.%s' % (self.topic, self.cluster),
                server=volume_utils.extract_host(self.cluster, 'host'))
            serializer = objects_base.CinderObjectSerializer(obj_version_cap)
            self.cluster_rpcserver = rpc.get_server(target, endpoints,
                                                    serializer)
            self.cluster_rpcserver.start()

        self.manager.init_host_with_rpc()

        if self.report_interval:
            self.tg.add_timer(self.report_interval, self.report_state,
                              initial_delay=self.report_interval)

        if self.periodic_interval:
            initial_delay: Optional[int]
            if self.periodic_fuzzy_delay:
                initial_delay = random.randint(0, self.periodic_fuzzy_delay)
            else:
                initial_delay = None
            self.tg.add_timer(self.periodic_interval, self.periodic_tasks,
                              initial_delay=initial_delay)

    def basic_config_check(self) -> None:
        """Perform basic config checks before starting service."""
        # Make sure report interval is less than service down time
        if self.report_interval:
            if CONF.service_down_time <= self.report_interval:
                new_down_time = int(self.report_interval * 2.5)
                LOG.warning(
                    "Report interval must be less than service down "
                    "time. Current config service_down_time: "
                    "%(service_down_time)s, report_interval for this: "
                    "service is: %(report_interval)s. Setting global "
                    "service_down_time to: %(new_down_time)s",
                    {'service_down_time': CONF.service_down_time,
                     'report_interval': self.report_interval,
                     'new_down_time': new_down_time})
                CONF.set_override('service_down_time', new_down_time)

    def _ensure_cluster_exists(self,
                               context: context.RequestContext,
                               service: 'Service') -> None:
        if self.cluster:
            try:
                cluster = objects.Cluster.get_by_id(context, None,
                                                    name=self.cluster,
                                                    binary=self.binary)
                # If the cluster already exists, then the service replication
                # fields must match those of the cluster unless the service
                # is in error status.
                error_states = (fields.ReplicationStatus.ERROR,
                                fields.ReplicationStatus.FAILOVER_ERROR)
                if service.replication_status not in error_states:
                    for attr in ('replication_status', 'active_backend_id',
                                 'frozen'):
                        if getattr(service, attr) != getattr(cluster, attr):
                            setattr(service, attr, getattr(cluster, attr))

            except exception.ClusterNotFound:
                # Since the cluster didn't exist, we copy replication fields
                # from the service.
                cluster = objects.Cluster(
                    context=context,
                    name=self.cluster,
                    binary=self.binary,
                    disabled=service.disabled,
                    replication_status=service.replication_status,
                    active_backend_id=service.active_backend_id,
                    frozen=service.frozen)
                try:
                    cluster.create()

                # Race condition occurred and another service created the
                # cluster, so we can continue as it already exists.
                except exception.ClusterExists:
                    pass

    def _create_service_ref(self,
                            context: context.RequestContext,
                            rpc_version: Optional[str] = None) -> None:
        kwargs = {
            'host': self.host,
            'binary': self.binary,
            'topic': self.topic,
            'report_count': 0,
            'availability_zone': self.availability_zone,
            'rpc_current_version': rpc_version or self.manager.RPC_API_VERSION,
            'object_current_version': objects_base.OBJ_VERSIONS.get_current(),
        }
        kwargs['cluster_name'] = self.cluster
        service_ref = objects.Service(context=context, **kwargs)
        service_ref.create()
        Service.service_id = service_ref.id
        self.origin_service_id = service_ref.id
        self._ensure_cluster_exists(context, service_ref)
        # If we have updated the service_ref with replication data from
        # the cluster it will be saved.
        service_ref.save()

    def __getattr__(self, key: str):
        manager = self.__dict__.get('manager', None)
        return getattr(manager, key)

    @classmethod
    def create(cls,
               host: Optional[str] = None,
               binary: Optional[str] = None,
               topic: Optional[str] = None,
               manager: Optional[str] = None,
               report_interval: Optional[int] = None,
               periodic_interval: Optional[int] = None,
               periodic_fuzzy_delay: Optional[int] = None,
               service_name: Optional[str] = None,
               coordination: bool = False,
               cluster: Optional[str] = None,
               **kwargs) -> 'Service':
        """Instantiates class and passes back application object.

        :param host: defaults to CONF.host
        :param binary: defaults to basename of executable
        :param topic: defaults to bin_name - 'cinder-' part
        :param manager: defaults to CONF.<topic>_manager
        :param report_interval: defaults to CONF.report_interval
        :param periodic_interval: defaults to CONF.periodic_interval
        :param periodic_fuzzy_delay: defaults to CONF.periodic_fuzzy_delay
        :param cluster: Defaults to None, as only some services will have it

        """
        if not host:
            host = CONF.host
        if not binary:
            binary = os.path.basename(inspect.stack()[-1][1])
        if not topic:
            topic = binary
        if not manager:
            subtopic = topic.rpartition('cinder-')[2]
            manager = CONF.get('%s_manager' % subtopic, None)
        if report_interval is None:
            report_interval = CONF.report_interval
        if periodic_interval is None:
            periodic_interval = CONF.periodic_interval
        if periodic_fuzzy_delay is None:
            periodic_fuzzy_delay = CONF.periodic_fuzzy_delay

        assert host is not None
        assert manager is not None
        service_obj = cls(host, binary, topic, manager,
                          report_interval=report_interval,
                          periodic_interval=periodic_interval,
                          periodic_fuzzy_delay=periodic_fuzzy_delay,
                          service_name=service_name,
                          coordination=coordination,
                          cluster=cluster, **kwargs)

        return service_obj

    def stop(self) -> None:
        # Try to shut the connection down, but if we get any sort of
        # errors, go ahead and ignore them.. as we're shutting down anyway
        try:
            if self.rpcserver is not None:
                self.rpcserver.stop()
            if self.backend_rpcserver:
                self.backend_rpcserver.stop()
            if self.cluster_rpcserver:
                self.cluster_rpcserver.stop()
        except Exception:
            pass

        if self.coordination:
            try:
                coordination.COORDINATOR.stop()
            except Exception:
                pass
        super(Service, self).stop(graceful=True)

    def wait(self) -> None:
        if self.rpcserver:
            self.rpcserver.wait()
        if self.backend_rpcserver:
            self.backend_rpcserver.wait()
        if self.cluster_rpcserver:
            self.cluster_rpcserver.wait()
        super(Service, self).wait()

    def periodic_tasks(self, raise_on_error: bool = False) -> None:
        """Tasks to be run at a periodic interval."""
        ctxt = context.get_admin_context()
        self.manager.run_periodic_tasks(ctxt, raise_on_error=raise_on_error)

    def report_state(self) -> None:
        """Update the state of this service in the datastore."""
        if not self.manager.is_working():
            # NOTE(dulek): If manager reports a problem we're not sending
            # heartbeats - to indicate that service is actually down.
            LOG.error('Manager for service %(binary)s %(host)s is '
                      'reporting problems, not sending heartbeat. '
                      'Service will appear "down".',
                      {'binary': self.binary,
                       'host': self.host})
            return

        ctxt = context.get_admin_context()
        try:
            try:
                service_ref = objects.Service.get_by_id(ctxt,
                                                        Service.service_id)
            except exception.NotFound:
                LOG.debug('The service database object disappeared, '
                          'recreating it.')
                self._create_service_ref(ctxt)
                service_ref = objects.Service.get_by_id(ctxt,
                                                        Service.service_id)

            service_ref.report_count += 1
            if self.availability_zone != service_ref.availability_zone:
                service_ref.availability_zone = self.availability_zone

            service_ref.save()

            # TODO(termie): make this pattern be more elegant.
            if getattr(self, 'model_disconnected', False):
                self.model_disconnected = False
                LOG.error('Recovered model server connection!')

        except db_exc.DBConnectionError:
            if not getattr(self, 'model_disconnected', False):
                self.model_disconnected = True
                LOG.exception('model server went away')

        # NOTE(jsbryant) Other DB errors can happen in HA configurations.
        # such errors shouldn't kill this thread, so we handle them here.
        except db_exc.DBError:
            if not getattr(self, 'model_disconnected', False):
                self.model_disconnected = True
                LOG.exception('DBError encountered: ')

        except Exception:
            if not getattr(self, 'model_disconnected', False):
                self.model_disconnected = True
                LOG.exception('Exception encountered: ')

    def reset(self) -> None:
        self.manager.reset()
        super(Service, self).reset()


class WSGIService(service.ServiceBase):
    """Provides ability to launch API from a 'paste' configuration."""

    def __init__(self, name, loader=None):
        """Initialize, but do not start the WSGI server.

        :param name: The name of the WSGI server given to the loader.
        :param loader: Loads the WSGI application using the given name.
        :returns: None

        """
        self.name = name
        self.manager = self._get_manager()
        self.loader = loader or wsgi.Loader(CONF)
        self.app = self.loader.load_app(name)
        self.host = getattr(CONF, '%s_listen' % name, "0.0.0.0")
        self.port = getattr(CONF, '%s_listen_port' % name, 0)
        self.use_ssl = getattr(CONF, '%s_use_ssl' % name, False)
        self.workers = (getattr(CONF, '%s_workers' % name, None) or
                        processutils.get_worker_count())
        if self.workers and self.workers < 1:
            worker_name = '%s_workers' % name
            msg = (_("%(worker_name)s value of %(workers)d is invalid, "
                     "must be greater than 0.") %
                   {'worker_name': worker_name,
                    'workers': self.workers})
            raise exception.InvalidConfigurationValue(msg)
        setup_profiler(name, self.host)

        self.server = wsgi.Server(CONF,
                                  name,
                                  self.app,
                                  host=self.host,
                                  port=self.port,
                                  use_ssl=self.use_ssl)

    def _get_manager(self):
        """Initialize a Manager object appropriate for this service.

        Use the service name to look up a Manager subclass from the
        configuration and initialize an instance. If no class name
        is configured, just return None.

        :returns: a Manager instance, or None.

        """
        fl = '%s_manager' % self.name
        if fl not in CONF:
            return None

        manager_class_name = CONF.get(fl, None)
        if not manager_class_name:
            return None

        manager_class = importutils.import_class(manager_class_name)
        return manager_class()

    def start(self) -> None:
        """Start serving this service using loaded configuration.

        Also, retrieve updated port number in case '0' was passed in, which
        indicates a random port should be used.

        :returns: None

        """
        if self.manager:
            self.manager.init_host()
        self.server.start()
        self.port = self.server.port

    def stop(self) -> None:
        """Stop serving this API.

        :returns: None

        """
        self.server.stop()

    def wait(self) -> None:
        """Wait for the service to stop serving this API.

        :returns: None

        """
        self.server.wait()

    def reset(self) -> None:
        """Reset server greenpool size to default.

        :returns: None

        """
        self.server.reset()


def process_launcher() -> service.ProcessLauncher:
    return service.ProcessLauncher(CONF, restart_method='mutate')


# NOTE(vish): the global launcher is to maintain the existing
#             functionality of calling service.serve +
#             service.wait
_launcher = None


def serve(server, workers=None):
    global _launcher
    if _launcher:
        raise RuntimeError(_('serve() can only be called once'))

    _launcher = service.launch(CONF, server, workers=workers,
                               restart_method='mutate')


def wait() -> None:
    CONF.log_opt_values(LOG, logging.DEBUG)

    try:
        _launcher.wait()  # type: ignore
    except KeyboardInterrupt:
        _launcher.stop()  # type: ignore
    rpc.cleanup()


class Launcher(object):
    def __init__(self):
        self.launch_service = serve
        self.wait = wait


def get_launcher() -> service.ProcessLauncher:
    # Note(lpetrut): ProcessLauncher uses green pipes which fail on Windows
    # due to missing support of non-blocking I/O pipes. For this reason, the
    # service must be spawned differently on Windows, using the ServiceLauncher
    # class instead.
    if os.name == 'nt':
        return Launcher()
    else:
        return process_launcher()


class WindowsProcessLauncher(object):
    def __init__(self):
        self._processutils = os_win_utilsfactory.get_processutils()

        self._workers = []
        self._worker_job_handles = []
        self._signal_handler = service.SignalHandler()
        self._add_signal_handlers()

    def add_process(self, cmd):
        LOG.info("Starting subprocess: %s", cmd)

        worker = subprocess.Popen(cmd)
        try:
            job_handle = self._processutils.kill_process_on_job_close(
                worker.pid)
        except Exception:
            LOG.exception("Could not associate child process "
                          "with a job, killing it.")
            worker.kill()
            raise

        self._worker_job_handles.append(job_handle)
        self._workers.append(worker)

    def _add_signal_handlers(self):
        self._signal_handler.add_handler('SIGINT', self._terminate)
        self._signal_handler.add_handler('SIGTERM', self._terminate)

    def _terminate(self, *args):
        # We've already assigned win32 job objects to child processes,
        # requesting them to stop once all the job handles are closed.
        # When this process dies, so will the child processes.
        LOG.info("Received request to terminate.")
        sys.exit(1)

    def wait(self):
        pids = [worker.pid for worker in self._workers]
        if pids:
            self._processutils.wait_for_multiple_processes(pids,
                                                           wait_all=True)
        # By sleeping here, we allow signal handlers to be executed.
        time.sleep(0)