Merge pull request #28 from matrix-org/erikj-perf

Database performance improvements.
This commit is contained in:
Mark Haines 2015-01-06 13:33:40 +00:00
commit 5e23a19204
5 changed files with 116 additions and 109 deletions

View File

@ -245,14 +245,12 @@ class RoomMemberHandler(BaseHandler):
self.distributor.declare("user_left_room") self.distributor.declare("user_left_room")
@defer.inlineCallbacks @defer.inlineCallbacks
def get_room_members(self, room_id, membership=Membership.JOIN): def get_room_members(self, room_id):
hs = self.hs hs = self.hs
memberships = yield self.store.get_room_members( users = yield self.store.get_users_in_room(room_id)
room_id=room_id, membership=membership
)
defer.returnValue([hs.parse_userid(m.user_id) for m in memberships]) defer.returnValue([hs.parse_userid(u) for u in users])
@defer.inlineCallbacks @defer.inlineCallbacks
def fetch_room_distributions_into(self, room_id, localusers=None, def fetch_room_distributions_into(self, room_id, localusers=None,
@ -531,11 +529,10 @@ class RoomListHandler(BaseHandler):
def get_public_room_list(self): def get_public_room_list(self):
chunk = yield self.store.get_rooms(is_public=True) chunk = yield self.store.get_rooms(is_public=True)
for room in chunk: for room in chunk:
joined_members = yield self.store.get_room_members( joined_users = yield self.store.get_users_in_room(
room_id=room["room_id"], room_id=room["room_id"],
membership=Membership.JOIN
) )
room["num_joined_members"] = len(joined_members) room["num_joined_members"] = len(joined_users)
# FIXME (erikj): START is no longer a valid value # FIXME (erikj): START is no longer a valid value
defer.returnValue({"start": "START", "end": "END", "chunk": chunk}) defer.returnValue({"start": "START", "end": "END", "chunk": chunk})

View File

@ -434,23 +434,29 @@ class SQLBaseStore(object):
return self.runInteraction("_simple_max_id", func) return self.runInteraction("_simple_max_id", func)
def _get_events(self, event_ids): def _get_events(self, event_ids, check_redacted=True,
get_prev_content=False):
return self.runInteraction( return self.runInteraction(
"_get_events", self._get_events_txn, event_ids "_get_events", self._get_events_txn, event_ids,
check_redacted=check_redacted, get_prev_content=get_prev_content,
) )
def _get_events_txn(self, txn, event_ids): def _get_events_txn(self, txn, event_ids, check_redacted=True,
events = [] get_prev_content=False):
for e_id in event_ids: if not event_ids:
ev = self._get_event_txn(txn, e_id) return []
if ev: return [
events.append(ev) self._get_event_txn(
txn, event_id,
return events check_redacted=check_redacted,
get_prev_content=get_prev_content
)
for event_id in event_ids
]
def _get_event_txn(self, txn, event_id, check_redacted=True, def _get_event_txn(self, txn, event_id, check_redacted=True,
get_prev_content=True): get_prev_content=False):
sql = ( sql = (
"SELECT internal_metadata, json, r.event_id FROM event_json as e " "SELECT internal_metadata, json, r.event_id FROM event_json as e "
"LEFT JOIN redactions as r ON e.event_id = r.redacts " "LEFT JOIN redactions as r ON e.event_id = r.redacts "
@ -467,6 +473,14 @@ class SQLBaseStore(object):
internal_metadata, js, redacted = res internal_metadata, js, redacted = res
return self._get_event_from_row_txn(
txn, internal_metadata, js, redacted,
check_redacted=check_redacted,
get_prev_content=get_prev_content,
)
def _get_event_from_row_txn(self, txn, internal_metadata, js, redacted,
check_redacted=True, get_prev_content=False):
d = json.loads(js) d = json.loads(js)
internal_metadata = json.loads(internal_metadata) internal_metadata = json.loads(internal_metadata)

View File

@ -123,6 +123,19 @@ class RoomMemberStore(SQLBaseStore):
else: else:
return None return None
def get_users_in_room(self, room_id):
def f(txn):
sql = (
"SELECT m.user_id FROM room_memberships as m"
" INNER JOIN current_state_events as c"
" ON m.event_id = c.event_id"
" WHERE m.membership = ? AND m.room_id = ?"
)
txn.execute(sql, (Membership.JOIN, room_id))
return [r[0] for r in txn.fetchall()]
return self.runInteraction("get_users_in_room", f)
def get_room_members(self, room_id, membership=None): def get_room_members(self, room_id, membership=None):
"""Retrieve the current room member list for a room. """Retrieve the current room member list for a room.
@ -183,20 +196,14 @@ class RoomMemberStore(SQLBaseStore):
) )
def _get_members_query_txn(self, txn, where_clause, where_values): def _get_members_query_txn(self, txn, where_clause, where_values):
del_sql = (
"SELECT event_id FROM redactions WHERE redacts = e.event_id "
"LIMIT 1"
)
sql = ( sql = (
"SELECT e.*, (%(redacted)s) AS redacted FROM events as e " "SELECT e.* FROM events as e "
"INNER JOIN room_memberships as m " "INNER JOIN room_memberships as m "
"ON e.event_id = m.event_id " "ON e.event_id = m.event_id "
"INNER JOIN current_state_events as c " "INNER JOIN current_state_events as c "
"ON m.event_id = c.event_id " "ON m.event_id = c.event_id "
"WHERE %(where)s " "WHERE %(where)s "
) % { ) % {
"redacted": del_sql,
"where": where_clause, "where": where_clause,
} }

View File

@ -15,6 +15,10 @@
from ._base import SQLBaseStore from ._base import SQLBaseStore
import logging
logger = logging.getLogger(__name__)
class StateStore(SQLBaseStore): class StateStore(SQLBaseStore):
""" Keeps track of the state at a given event. """ Keeps track of the state at a given event.
@ -62,14 +66,8 @@ class StateStore(SQLBaseStore):
keyvalues={"state_group": group}, keyvalues={"state_group": group},
retcol="event_id", retcol="event_id",
) )
state = []
for state_id in state_ids: state = self._get_events_txn(txn, state_ids)
s = self._get_events_txn(
txn,
[state_id],
)
if s:
state.extend(s)
res[group] = state res[group] = state

View File

@ -137,7 +137,6 @@ class StreamStore(SQLBaseStore):
with_feedback=with_feedback, with_feedback=with_feedback,
) )
@defer.inlineCallbacks
@log_function @log_function
def get_room_events_stream(self, user_id, from_key, to_key, room_id, def get_room_events_stream(self, user_id, from_key, to_key, room_id,
limit=0, with_feedback=False): limit=0, with_feedback=False):
@ -157,11 +156,6 @@ class StreamStore(SQLBaseStore):
"WHERE m.user_id = ? " "WHERE m.user_id = ? "
) )
del_sql = (
"SELECT event_id FROM redactions WHERE redacts = e.event_id "
"LIMIT 1"
)
if limit: if limit:
limit = max(limit, MAX_STREAM_SIZE) limit = max(limit, MAX_STREAM_SIZE)
else: else:
@ -172,38 +166,42 @@ class StreamStore(SQLBaseStore):
to_id = _parse_stream_token(to_key) to_id = _parse_stream_token(to_key)
if from_key == to_key: if from_key == to_key:
defer.returnValue(([], to_key)) return defer.succeed(([], to_key))
return
sql = ( sql = (
"SELECT *, (%(redacted)s) AS redacted FROM events AS e WHERE " "SELECT e.event_id, e.stream_ordering FROM events AS e WHERE "
"(e.outlier = 0 AND (room_id IN (%(current)s)) OR " "(e.outlier = 0 AND (room_id IN (%(current)s)) OR "
"(event_id IN (%(invites)s))) " "(event_id IN (%(invites)s))) "
"AND e.stream_ordering > ? AND e.stream_ordering <= ? " "AND e.stream_ordering > ? AND e.stream_ordering <= ? "
"ORDER BY stream_ordering ASC LIMIT %(limit)d " "ORDER BY stream_ordering ASC LIMIT %(limit)d "
) % { ) % {
"redacted": del_sql,
"current": current_room_membership_sql, "current": current_room_membership_sql,
"invites": membership_sql, "invites": membership_sql,
"limit": limit "limit": limit
} }
rows = yield self._execute_and_decode( def f(txn):
sql, txn.execute(sql, (user_id, user_id, from_id, to_id,))
user_id, user_id, from_id, to_id
)
ret = yield self._parse_events(rows) rows = self.cursor_to_dict(txn)
ret = self._get_events_txn(
txn,
[r["event_id"] for r in rows],
get_prev_content=True
)
if rows: if rows:
key = "s%d" % max([r["stream_ordering"] for r in rows]) key = "s%d" % max([r["stream_ordering"] for r in rows])
else: else:
# Assume we didn't get anything because there was nothing to get. # Assume we didn't get anything because there was nothing to
# get.
key = to_key key = to_key
defer.returnValue((ret, key)) return ret, key
return self.runInteraction("get_room_events_stream", f)
@defer.inlineCallbacks
@log_function @log_function
def paginate_room_events(self, room_id, from_key, to_key=None, def paginate_room_events(self, room_id, from_key, to_key=None,
direction='b', limit=-1, direction='b', limit=-1,
@ -221,7 +219,9 @@ class StreamStore(SQLBaseStore):
bounds = _get_token_bound(from_key, from_comp) bounds = _get_token_bound(from_key, from_comp)
if to_key: if to_key:
bounds = "%s AND %s" % (bounds, _get_token_bound(to_key, to_comp)) bounds = "%s AND %s" % (
bounds, _get_token_bound(to_key, to_comp)
)
if int(limit) > 0: if int(limit) > 0:
args.append(int(limit)) args.append(int(limit))
@ -229,27 +229,21 @@ class StreamStore(SQLBaseStore):
else: else:
limit_str = "" limit_str = ""
del_sql = (
"SELECT event_id FROM redactions WHERE redacts = events.event_id "
"LIMIT 1"
)
sql = ( sql = (
"SELECT *, (%(redacted)s) AS redacted FROM events" "SELECT * FROM events"
" WHERE outlier = 0 AND room_id = ? AND %(bounds)s" " WHERE outlier = 0 AND room_id = ? AND %(bounds)s"
" ORDER BY topological_ordering %(order)s," " ORDER BY topological_ordering %(order)s,"
" stream_ordering %(order)s %(limit)s" " stream_ordering %(order)s %(limit)s"
) % { ) % {
"redacted": del_sql,
"bounds": bounds, "bounds": bounds,
"order": order, "order": order,
"limit": limit_str "limit": limit_str
} }
rows = yield self._execute_and_decode( def f(txn):
sql, txn.execute(sql, args)
*args
) rows = self.cursor_to_dict(txn)
if rows: if rows:
topo = rows[-1]["topological_ordering"] topo = rows[-1]["topological_ordering"]
@ -262,38 +256,31 @@ class StreamStore(SQLBaseStore):
# TODO (erikj): We should work out what to do here instead. # TODO (erikj): We should work out what to do here instead.
next_token = to_key if to_key else from_key next_token = to_key if to_key else from_key
events = yield self._parse_events(rows) events = self._get_events_txn(
txn,
defer.returnValue( [r["event_id"] for r in rows],
( get_prev_content=True
events,
next_token
)
) )
@defer.inlineCallbacks return events, next_token,
return self.runInteraction("paginate_room_events", f)
def get_recent_events_for_room(self, room_id, limit, end_token, def get_recent_events_for_room(self, room_id, limit, end_token,
with_feedback=False): with_feedback=False):
# TODO (erikj): Handle compressed feedback # TODO (erikj): Handle compressed feedback
del_sql = (
"SELECT event_id FROM redactions WHERE redacts = events.event_id "
"LIMIT 1"
)
sql = ( sql = (
"SELECT *, (%(redacted)s) AS redacted FROM events " "SELECT * FROM events "
"WHERE room_id = ? AND stream_ordering <= ? AND outlier = 0 " "WHERE room_id = ? AND stream_ordering <= ? AND outlier = 0 "
"ORDER BY topological_ordering DESC, stream_ordering DESC LIMIT ? " "ORDER BY topological_ordering DESC, stream_ordering DESC LIMIT ? "
) % {
"redacted": del_sql,
}
rows = yield self._execute_and_decode(
sql,
room_id, end_token, limit
) )
def f(txn):
txn.execute(sql, (room_id, end_token, limit,))
rows = self.cursor_to_dict(txn)
rows.reverse() # As we selected with reverse ordering rows.reverse() # As we selected with reverse ordering
if rows: if rows:
@ -305,11 +292,15 @@ class StreamStore(SQLBaseStore):
else: else:
token = (end_token, end_token) token = (end_token, end_token)
events = yield self._parse_events(rows) events = self._get_events_txn(
txn,
[r["event_id"] for r in rows],
get_prev_content=True
)
ret = (events, token) return events, token
defer.returnValue(ret) return self.runInteraction("get_recent_events_for_room", f)
def get_room_events_max_id(self): def get_room_events_max_id(self):
return self.runInteraction( return self.runInteraction(