summary refs log tree commit diff
path: root/tests/replication/tcp/streams/_base.py
blob: 83e16cfe3d0821a671281fe846825becc3ff317e (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
# -*- coding: utf-8 -*-
# Copyright 2019 New Vector Ltd
#
# 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 logging
from typing import Any, Dict, List, Optional, Tuple

import attr

from twisted.internet.interfaces import IConsumer, IPullProducer, IReactorTime
from twisted.internet.task import LoopingCall
from twisted.web.http import HTTPChannel

from synapse.app.generic_worker import GenericWorkerServer
from synapse.http.site import SynapseRequest
from synapse.replication.slave.storage._base import BaseSlavedStore
from synapse.replication.tcp.client import ReplicationDataHandler
from synapse.replication.tcp.handler import ReplicationCommandHandler
from synapse.replication.tcp.protocol import ClientReplicationStreamProtocol
from synapse.replication.tcp.resource import ReplicationStreamProtocolFactory
from synapse.util import Clock

from tests import unittest
from tests.server import FakeTransport

logger = logging.getLogger(__name__)


class BaseStreamTestCase(unittest.HomeserverTestCase):
    """Base class for tests of the replication streams"""

    def prepare(self, reactor, clock, hs):
        # build a replication server
        server_factory = ReplicationStreamProtocolFactory(hs)
        self.streamer = hs.get_replication_streamer()
        self.server = server_factory.buildProtocol(None)

        # Make a new HomeServer object for the worker
        config = self.default_config()
        config["worker_app"] = "synapse.app.generic_worker"
        config["worker_replication_host"] = "testserv"
        config["worker_replication_http_port"] = "8765"

        self.reactor.lookups["testserv"] = "1.2.3.4"

        self.worker_hs = self.setup_test_homeserver(
            http_client=None,
            homeserverToUse=GenericWorkerServer,
            config=config,
            reactor=self.reactor,
        )

        # Since we use sqlite in memory databases we need to make sure the
        # databases objects are the same.
        self.worker_hs.get_datastore().db = hs.get_datastore().db

        self.test_handler = self._build_replication_data_handler()
        self.worker_hs.replication_data_handler = self.test_handler

        repl_handler = ReplicationCommandHandler(self.worker_hs)
        self.client = ClientReplicationStreamProtocol(
            self.worker_hs, "client", "test", clock, repl_handler,
        )

        self._client_transport = None
        self._server_transport = None

    def _build_replication_data_handler(self):
        return TestReplicationDataHandler(self.worker_hs.get_datastore())

    def reconnect(self):
        if self._client_transport:
            self.client.close()

        if self._server_transport:
            self.server.close()

        self._client_transport = FakeTransport(self.server, self.reactor)
        self.client.makeConnection(self._client_transport)

        self._server_transport = FakeTransport(self.client, self.reactor)
        self.server.makeConnection(self._server_transport)

    def disconnect(self):
        if self._client_transport:
            self._client_transport = None
            self.client.close()

        if self._server_transport:
            self._server_transport = None
            self.server.close()

    def replicate(self):
        """Tell the master side of replication that something has happened, and then
        wait for the replication to occur.
        """
        self.streamer.on_notifier_poke()
        self.pump(0.1)

    def handle_http_replication_attempt(self) -> SynapseRequest:
        """Asserts that a connection attempt was made to the master HS on the
        HTTP replication port, then proxies it to the master HS object to be
        handled.

        Returns:
            The request object received by master HS.
        """

        # We should have an outbound connection attempt.
        clients = self.reactor.tcpClients
        self.assertEqual(len(clients), 1)
        (host, port, client_factory, _timeout, _bindAddress) = clients.pop(0)
        self.assertEqual(host, "1.2.3.4")
        self.assertEqual(port, 8765)

        # Set up client side protocol
        client_protocol = client_factory.buildProtocol(None)

        request_factory = OneShotRequestFactory()

        # Set up the server side protocol
        channel = _PushHTTPChannel(self.reactor)
        channel.requestFactory = request_factory
        channel.site = self.site

        # Connect client to server and vice versa.
        client_to_server_transport = FakeTransport(
            channel, self.reactor, client_protocol
        )
        client_protocol.makeConnection(client_to_server_transport)

        server_to_client_transport = FakeTransport(
            client_protocol, self.reactor, channel
        )
        channel.makeConnection(server_to_client_transport)

        # The request will now be processed by `self.site` and the response
        # streamed back.
        self.reactor.advance(0)

        # We tear down the connection so it doesn't get reused without our
        # knowledge.
        server_to_client_transport.loseConnection()
        client_to_server_transport.loseConnection()

        return request_factory.request

    def assert_request_is_get_repl_stream_updates(
        self, request: SynapseRequest, stream_name: str
    ):
        """Asserts that the given request is a HTTP replication request for
        fetching updates for given stream.
        """

        self.assertRegex(
            request.path,
            br"^/_synapse/replication/get_repl_stream_updates/%s/[^/]+$"
            % (stream_name.encode("ascii"),),
        )

        self.assertEqual(request.method, b"GET")


class TestReplicationDataHandler(ReplicationDataHandler):
    """Drop-in for ReplicationDataHandler which just collects RDATA rows"""

    def __init__(self, store: BaseSlavedStore):
        super().__init__(store)

        # streams to subscribe to: map from stream id to position
        self.stream_positions = {}  # type: Dict[str, int]

        # list of received (stream_name, token, row) tuples
        self.received_rdata_rows = []  # type: List[Tuple[str, int, Any]]

    def get_streams_to_replicate(self):
        return self.stream_positions

    async def on_rdata(self, stream_name, token, rows):
        await super().on_rdata(stream_name, token, rows)
        for r in rows:
            self.received_rdata_rows.append((stream_name, token, r))

        if (
            stream_name in self.stream_positions
            and token > self.stream_positions[stream_name]
        ):
            self.stream_positions[stream_name] = token


@attr.s()
class OneShotRequestFactory:
    """A simple request factory that generates a single `SynapseRequest` and
    stores it for future use. Can only be used once.
    """

    request = attr.ib(default=None)

    def __call__(self, *args, **kwargs):
        assert self.request is None

        self.request = SynapseRequest(*args, **kwargs)
        return self.request


class _PushHTTPChannel(HTTPChannel):
    """A HTTPChannel that wraps pull producers to push producers.

    This is a hack to get around the fact that HTTPChannel transparently wraps a
    pull producer (which is what Synapse uses to reply to requests) with
    `_PullToPush` to convert it to a push producer. Unfortunately `_PullToPush`
    uses the standard reactor rather than letting us use our test reactor, which
    makes it very hard to test.
    """

    def __init__(self, reactor: IReactorTime):
        super().__init__()
        self.reactor = reactor

        self._pull_to_push_producer = None  # type: Optional[_PullToPushProducer]

    def registerProducer(self, producer, streaming):
        # Convert pull producers to push producer.
        if not streaming:
            self._pull_to_push_producer = _PullToPushProducer(
                self.reactor, producer, self
            )
            producer = self._pull_to_push_producer

        super().registerProducer(producer, True)

    def unregisterProducer(self):
        if self._pull_to_push_producer:
            # We need to manually stop the _PullToPushProducer.
            self._pull_to_push_producer.stop()


class _PullToPushProducer:
    """A push producer that wraps a pull producer.
    """

    def __init__(
        self, reactor: IReactorTime, producer: IPullProducer, consumer: IConsumer
    ):
        self._clock = Clock(reactor)
        self._producer = producer
        self._consumer = consumer

        # While running we use a looping call with a zero delay to call
        # resumeProducing on given producer.
        self._looping_call = None  # type: Optional[LoopingCall]

        # We start writing next reactor tick.
        self._start_loop()

    def _start_loop(self):
        """Start the looping call to
        """

        if not self._looping_call:
            # Start a looping call which runs every tick.
            self._looping_call = self._clock.looping_call(self._run_once, 0)

    def stop(self):
        """Stops calling resumeProducing.
        """
        if self._looping_call:
            self._looping_call.stop()
            self._looping_call = None

    def pauseProducing(self):
        """Implements IPushProducer
        """
        self.stop()

    def resumeProducing(self):
        """Implements IPushProducer
        """
        self._start_loop()

    def stopProducing(self):
        """Implements IPushProducer
        """
        self.stop()
        self._producer.stopProducing()

    def _run_once(self):
        """Calls resumeProducing on producer once.
        """

        try:
            self._producer.resumeProducing()
        except Exception:
            logger.exception("Failed to call resumeProducing")
            try:
                self._consumer.unregisterProducer()
            except Exception:
                pass

            self.stopProducing()