Use event streams to calculate presence

Primarily this fixes a bug in the handling of remote users joining a
room where the server sent out the presence for all local users in the
room to all servers in the room.

We also change to using the state delta stream, rather than the
distributor, as it will make it easier to split processing out of the
master process (as well as being more flexible).

Finally, when sending presence states to newly joined servers we filter
out old presence states to reduce the number sent. Initially we filter
out states that are offline and have a last active more than a week ago,
though this can be changed down the line.

Fixes #3962
This commit is contained in:
Erik Johnston 2019-03-26 12:45:22 +00:00
parent bbd244c7b2
commit 197fae1639
3 changed files with 219 additions and 32 deletions

View file

@ -31,9 +31,11 @@ from prometheus_client import Counter
from twisted.internet import defer
from synapse.api.constants import PresenceState
import synapse.metrics
from synapse.api.constants import EventTypes, Membership, PresenceState
from synapse.api.errors import SynapseError
from synapse.metrics import LaterGauge
from synapse.metrics.background_process_metrics import run_as_background_process
from synapse.storage.presence import UserPresenceState
from synapse.types import UserID, get_domain_from_id
from synapse.util.async_helpers import Linearizer
@ -98,6 +100,7 @@ class PresenceHandler(object):
self.hs = hs
self.is_mine = hs.is_mine
self.is_mine_id = hs.is_mine_id
self.server_name = hs.hostname
self.clock = hs.get_clock()
self.store = hs.get_datastore()
self.wheel_timer = WheelTimer()
@ -132,9 +135,6 @@ class PresenceHandler(object):
)
)
distributor = hs.get_distributor()
distributor.observe("user_joined_room", self.user_joined_room)
active_presence = self.store.take_presence_startup_info()
# A dictionary of the current state of users. This is prefilled with
@ -220,6 +220,15 @@ class PresenceHandler(object):
LaterGauge("synapse_handlers_presence_wheel_timer_size", "", [],
lambda: len(self.wheel_timer))
# Used to handle sending of presence to newly joined users/servers
if hs.config.use_presence:
self.notifier.add_replication_callback(self.notify_new_event)
# Presence is best effort and quickly heals itself, so lets just always
# stream from the current state when we restart.
self._event_pos = self.store.get_current_events_token()
self._event_processing = False
@defer.inlineCallbacks
def _on_shutdown(self):
"""Gets called when shutting down. This lets us persist any updates that
@ -750,31 +759,6 @@ class PresenceHandler(object):
yield self._update_states([prev_state.copy_and_replace(**new_fields)])
@defer.inlineCallbacks
def user_joined_room(self, user, room_id):
"""Called (via the distributor) when a user joins a room. This funciton
sends presence updates to servers, either:
1. the joining user is a local user and we send their presence to
all servers in the room.
2. the joining user is a remote user and so we send presence for all
local users in the room.
"""
# We only need to send presence to servers that don't have it yet. We
# don't need to send to local clients here, as that is done as part
# of the event stream/sync.
# TODO: Only send to servers not already in the room.
if self.is_mine(user):
state = yield self.current_state_for_user(user.to_string())
self._push_to_remotes([state])
else:
user_ids = yield self.store.get_users_in_room(room_id)
user_ids = list(filter(self.is_mine_id, user_ids))
states = yield self.current_state_for_users(user_ids)
self._push_to_remotes(list(states.values()))
@defer.inlineCallbacks
def get_presence_list(self, observer_user, accepted=None):
"""Returns the presence for all users in their presence list.
@ -945,6 +929,123 @@ class PresenceHandler(object):
rows = yield self.store.get_all_presence_updates(last_id, current_id)
defer.returnValue(rows)
def notify_new_event(self):
"""Called when new events have happened. Handles users and servers
joining rooms and require being sent presence.
"""
@defer.inlineCallbacks
def _process_presence():
if self._event_processing:
return
self._event_processing = True
try:
yield self._unsafe_process()
finally:
self._event_processing = False
run_as_background_process("presence.notify_new_event", _process_presence)
@defer.inlineCallbacks
def _unsafe_process(self):
# Loop round handling deltas until we're up to date
while True:
with Measure(self.clock, "presence_delta"):
deltas = yield self.store.get_current_state_deltas(self._event_pos)
if not deltas:
return
yield self._handle_state_delta(deltas)
self._event_pos = deltas[-1]["stream_id"]
# Expose current event processing position to prometheus
synapse.metrics.event_processing_positions.labels("presence").set(
self._event_pos
)
@defer.inlineCallbacks
def _handle_state_delta(self, deltas):
"""Process current state deltas to find new joins that need to be
handled.
"""
for delta in deltas:
typ = delta["type"]
state_key = delta["state_key"]
room_id = delta["room_id"]
event_id = delta["event_id"]
prev_event_id = delta["prev_event_id"]
logger.debug("Handling: %r %r, %s", typ, state_key, event_id)
if typ != EventTypes.Member:
continue
event = yield self.store.get_event(event_id)
if event.content.get("membership") != Membership.JOIN:
# We only care about joins
continue
if prev_event_id:
prev_event = yield self.store.get_event(prev_event_id)
if prev_event.content.get("membership") == Membership.JOIN:
# Ignore changes to join events.
continue
if self.is_mine_id(state_key):
# If this is a local user then we need to send their presence
# out to hosts in the room (who don't already have it)
# TODO: We should be able to filter the hosts down to those that
# haven't previously seen the user
state = yield self.current_state_for_user(state_key)
hosts = yield self.state.get_current_hosts_in_room(room_id)
# Filter out ourselves.
hosts = set(host for host in hosts if host != self.server_name)
self.federation.send_presence_to_destinations(
states=[state],
destinations=hosts,
)
else:
# A remote user has joined the room, so we need to:
# 1. Check if this is a new server in the room
# 2. If so send any presence they don't already have for
# local users in the room.
# TODO: We should be able to filter the users down to those that
# the server hasn't previously seen
# TODO: Check that this is actually a new server joining the
# room.
user_ids = yield self.state.get_current_user_in_room(room_id)
user_ids = list(filter(self.is_mine_id, user_ids))
states = yield self.current_state_for_users(user_ids)
# Filter out old presence, i.e. offline presence states where
# the user hasn't been active for a week. We can change this
# depending on what we want the UX to be, but at the least we
# should filter out offline presence where the state is just the
# default state.
now = self.clock.time_msec()
states = [
state for state in states.values()
if state.state != PresenceState.OFFLINE
or now - state.last_active_ts < 7 * 24 * 60 * 60 * 1000
or state.status_msg is not None
]
if states:
self.federation.send_presence_to_destinations(
states=states,
destinations=[get_domain_from_id(state_key)],
)
def should_notify(old_state, new_state):
"""Decides if a presence state change should be sent to interested parties.