summary refs log tree commit diff
path: root/synapse/logging/_remote.py
blob: 8202d0494d728e266e84c249d5828e162aa95eab (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
# Copyright 2020 The Matrix.org Foundation C.I.C.
#
# 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
import sys
import traceback
from collections import deque
from ipaddress import IPv4Address, IPv6Address, ip_address
from math import floor
from typing import Callable, Optional

import attr
from typing_extensions import Deque
from zope.interface import implementer

from twisted.application.internet import ClientService
from twisted.internet.defer import CancelledError, Deferred
from twisted.internet.endpoints import (
    HostnameEndpoint,
    TCP4ClientEndpoint,
    TCP6ClientEndpoint,
)
from twisted.internet.interfaces import IPushProducer, IStreamClientEndpoint
from twisted.internet.protocol import Factory, Protocol
from twisted.internet.tcp import Connection
from twisted.python.failure import Failure

logger = logging.getLogger(__name__)


@attr.s
@implementer(IPushProducer)
class LogProducer:
    """
    An IPushProducer that writes logs from its buffer to its transport when it
    is resumed.

    Args:
        buffer: Log buffer to read logs from.
        transport: Transport to write to.
        format: A callable to format the log record to a string.
    """

    # This is essentially ITCPTransport, but that is missing certain fields
    # (connected and registerProducer) which are part of the implementation.
    transport = attr.ib(type=Connection)
    _format = attr.ib(type=Callable[[logging.LogRecord], str])
    _buffer = attr.ib(type=deque)
    _paused = attr.ib(default=False, type=bool, init=False)

    def pauseProducing(self):
        self._paused = True

    def stopProducing(self):
        self._paused = True
        self._buffer = deque()

    def resumeProducing(self):
        # If we're already producing, nothing to do.
        self._paused = False

        # Loop until paused.
        while self._paused is False and (self._buffer and self.transport.connected):
            try:
                # Request the next record and format it.
                record = self._buffer.popleft()
                msg = self._format(record)

                # Send it as a new line over the transport.
                self.transport.write(msg.encode("utf8"))
                self.transport.write(b"\n")
            except Exception:
                # Something has gone wrong writing to the transport -- log it
                # and break out of the while.
                traceback.print_exc(file=sys.__stderr__)
                break


class RemoteHandler(logging.Handler):
    """
    An logging handler that writes logs to a TCP target.

    Args:
        host: The host of the logging target.
        port: The logging target's port.
        maximum_buffer: The maximum buffer size.
    """

    def __init__(
        self,
        host: str,
        port: int,
        maximum_buffer: int = 1000,
        level=logging.NOTSET,
        _reactor=None,
    ):
        super().__init__(level=level)
        self.host = host
        self.port = port
        self.maximum_buffer = maximum_buffer

        self._buffer: Deque[logging.LogRecord] = deque()
        self._connection_waiter: Optional[Deferred] = None
        self._producer: Optional[LogProducer] = None

        # Connect without DNS lookups if it's a direct IP.
        if _reactor is None:
            from twisted.internet import reactor

            _reactor = reactor

        try:
            ip = ip_address(self.host)
            if isinstance(ip, IPv4Address):
                endpoint: IStreamClientEndpoint = TCP4ClientEndpoint(
                    _reactor, self.host, self.port
                )
            elif isinstance(ip, IPv6Address):
                endpoint = TCP6ClientEndpoint(_reactor, self.host, self.port)
            else:
                raise ValueError("Unknown IP address provided: %s" % (self.host,))
        except ValueError:
            endpoint = HostnameEndpoint(_reactor, self.host, self.port)

        factory = Factory.forProtocol(Protocol)
        self._service = ClientService(endpoint, factory, clock=_reactor)
        self._service.startService()
        self._stopping = False
        self._connect()

    def close(self):
        self._stopping = True
        self._service.stopService()

    def _connect(self) -> None:
        """
        Triggers an attempt to connect then write to the remote if not already writing.
        """
        # Do not attempt to open multiple connections.
        if self._connection_waiter:
            return

        def fail(failure: Failure) -> None:
            # If the Deferred was cancelled (e.g. during shutdown) do not try to
            # reconnect (this will cause an infinite loop of errors).
            if failure.check(CancelledError) and self._stopping:
                return

            # For a different error, print the traceback and re-connect.
            failure.printTraceback(file=sys.__stderr__)
            self._connection_waiter = None
            self._connect()

        def writer(result: Protocol) -> None:
            # Force recognising transport as a Connection and not the more
            # generic ITransport.
            transport: Connection = result.transport  # type: ignore

            # We have a connection. If we already have a producer, and its
            # transport is the same, just trigger a resumeProducing.
            if self._producer and transport is self._producer.transport:
                self._producer.resumeProducing()
                self._connection_waiter = None
                return

            # If the producer is still producing, stop it.
            if self._producer:
                self._producer.stopProducing()

            # Make a new producer and start it.
            self._producer = LogProducer(
                buffer=self._buffer,
                transport=transport,
                format=self.format,
            )
            transport.registerProducer(self._producer, True)
            self._producer.resumeProducing()
            self._connection_waiter = None

        deferred: Deferred = self._service.whenConnected(failAfterFailures=1)
        deferred.addCallbacks(writer, fail)
        self._connection_waiter = deferred

    def _handle_pressure(self) -> None:
        """
        Handle backpressure by shedding records.

        The buffer will, in this order, until the buffer is below the maximum:
            - Shed DEBUG records.
            - Shed INFO records.
            - Shed the middle 50% of the records.
        """
        if len(self._buffer) <= self.maximum_buffer:
            return

        # Strip out DEBUGs
        self._buffer = deque(
            filter(lambda record: record.levelno > logging.DEBUG, self._buffer)
        )

        if len(self._buffer) <= self.maximum_buffer:
            return

        # Strip out INFOs
        self._buffer = deque(
            filter(lambda record: record.levelno > logging.INFO, self._buffer)
        )

        if len(self._buffer) <= self.maximum_buffer:
            return

        # Cut the middle entries out
        buffer_split = floor(self.maximum_buffer / 2)

        old_buffer = self._buffer
        self._buffer = deque()

        for _ in range(buffer_split):
            self._buffer.append(old_buffer.popleft())

        end_buffer = []
        for _ in range(buffer_split):
            end_buffer.append(old_buffer.pop())

        self._buffer.extend(reversed(end_buffer))

    def emit(self, record: logging.LogRecord) -> None:
        self._buffer.append(record)

        # Handle backpressure, if it exists.
        try:
            self._handle_pressure()
        except Exception:
            # If handling backpressure fails, clear the buffer and log the
            # exception.
            self._buffer.clear()
            logger.warning("Failed clearing backpressure")

        # Try and write immediately.
        self._connect()