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
|
# -*- coding: utf-8 -*-
# Copyright 2019 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 json
import logging
import random
import time
import attr
from twisted.internet import defer
from twisted.web.client import RedirectAgent, readBody
from twisted.web.http import stringToDatetime
from synapse.logging.context import make_deferred_yieldable
from synapse.util import Clock
from synapse.util.caches.ttlcache import TTLCache
from synapse.util.metrics import Measure
# period to cache .well-known results for by default
WELL_KNOWN_DEFAULT_CACHE_PERIOD = 24 * 3600
# jitter to add to the .well-known default cache ttl
WELL_KNOWN_DEFAULT_CACHE_PERIOD_JITTER = 10 * 60
# period to cache failure to fetch .well-known for
WELL_KNOWN_INVALID_CACHE_PERIOD = 1 * 3600
# cap for .well-known cache period
WELL_KNOWN_MAX_CACHE_PERIOD = 48 * 3600
# lower bound for .well-known cache period
WELL_KNOWN_MIN_CACHE_PERIOD = 5 * 60
# Attempt to refetch a cached well-known N% of the TTL before it expires.
# e.g. if set to 0.2 and we have a cached entry with a TTL of 5mins, then
# we'll start trying to refetch 1 minute before it expires.
WELL_KNOWN_GRACE_PERIOD_FACTOR = 0.2
logger = logging.getLogger(__name__)
_well_known_cache = TTLCache("well-known")
@attr.s(slots=True, frozen=True)
class WellKnownLookupResult(object):
delegated_server = attr.ib()
class WellKnownResolver(object):
"""Handles well-known lookups for matrix servers.
"""
def __init__(self, reactor, agent, well_known_cache=None):
self._reactor = reactor
self._clock = Clock(reactor)
if well_known_cache is None:
well_known_cache = _well_known_cache
self._well_known_cache = well_known_cache
self._well_known_agent = RedirectAgent(agent)
@defer.inlineCallbacks
def get_well_known(self, server_name):
"""Attempt to fetch and parse a .well-known file for the given server
Args:
server_name (bytes): name of the server, from the requested url
Returns:
Deferred[WellKnownLookupResult]: The result of the lookup
"""
try:
prev_result, expiry, ttl = self._well_known_cache.get_with_expiry(
server_name
)
now = self._clock.time()
if now < expiry - WELL_KNOWN_GRACE_PERIOD_FACTOR * ttl:
return WellKnownLookupResult(delegated_server=prev_result)
except KeyError:
prev_result = None
# TODO: should we linearise so that we don't end up doing two .well-known
# requests for the same server in parallel?
try:
with Measure(self._clock, "get_well_known"):
result, cache_period = yield self._do_get_well_known(server_name)
except _FetchWellKnownFailure as e:
if prev_result and e.temporary:
# This is a temporary failure and we have a still valid cached
# result, so lets return that. Hopefully the next time we ask
# the remote will be back up again.
return WellKnownLookupResult(delegated_server=prev_result)
result = None
# add some randomness to the TTL to avoid a stampeding herd every hour
# after startup
cache_period = WELL_KNOWN_INVALID_CACHE_PERIOD
cache_period += random.uniform(0, WELL_KNOWN_DEFAULT_CACHE_PERIOD_JITTER)
if cache_period > 0:
self._well_known_cache.set(server_name, result, cache_period)
return WellKnownLookupResult(delegated_server=result)
@defer.inlineCallbacks
def _do_get_well_known(self, server_name):
"""Actually fetch and parse a .well-known, without checking the cache
Args:
server_name (bytes): name of the server, from the requested url
Raises:
_FetchWellKnownFailure if we fail to lookup a result
Returns:
Deferred[Tuple[bytes,int]]: The lookup result and cache period.
"""
uri = b"https://%s/.well-known/matrix/server" % (server_name,)
uri_str = uri.decode("ascii")
logger.info("Fetching %s", uri_str)
# We do this in two steps to differentiate between possibly transient
# errors (e.g. can't connect to host, 503 response) and more permenant
# errors (such as getting a 404 response).
try:
response = yield make_deferred_yieldable(
self._well_known_agent.request(b"GET", uri)
)
body = yield make_deferred_yieldable(readBody(response))
if 500 <= response.code < 600:
raise Exception("Non-200 response %s" % (response.code,))
except Exception as e:
logger.info("Error fetching %s: %s", uri_str, e)
raise _FetchWellKnownFailure(temporary=True)
try:
if response.code != 200:
raise Exception("Non-200 response %s" % (response.code,))
parsed_body = json.loads(body.decode("utf-8"))
logger.info("Response from .well-known: %s", parsed_body)
result = parsed_body["m.server"].encode("ascii")
except Exception as e:
logger.info("Error fetching %s: %s", uri_str, e)
raise _FetchWellKnownFailure(temporary=False)
cache_period = _cache_period_from_headers(
response.headers, time_now=self._reactor.seconds
)
if cache_period is None:
cache_period = WELL_KNOWN_DEFAULT_CACHE_PERIOD
# add some randomness to the TTL to avoid a stampeding herd every 24 hours
# after startup
cache_period += random.uniform(0, WELL_KNOWN_DEFAULT_CACHE_PERIOD_JITTER)
else:
cache_period = min(cache_period, WELL_KNOWN_MAX_CACHE_PERIOD)
cache_period = max(cache_period, WELL_KNOWN_MIN_CACHE_PERIOD)
return (result, cache_period)
def _cache_period_from_headers(headers, time_now=time.time):
cache_controls = _parse_cache_control(headers)
if b"no-store" in cache_controls:
return 0
if b"max-age" in cache_controls:
try:
max_age = int(cache_controls[b"max-age"])
return max_age
except ValueError:
pass
expires = headers.getRawHeaders(b"expires")
if expires is not None:
try:
expires_date = stringToDatetime(expires[-1])
return expires_date - time_now()
except ValueError:
# RFC7234 says 'A cache recipient MUST interpret invalid date formats,
# especially the value "0", as representing a time in the past (i.e.,
# "already expired").
return 0
return None
def _parse_cache_control(headers):
cache_controls = {}
for hdr in headers.getRawHeaders(b"cache-control", []):
for directive in hdr.split(b","):
splits = [x.strip() for x in directive.split(b"=", 1)]
k = splits[0].lower()
v = splits[1] if len(splits) > 1 else None
cache_controls[k] = v
return cache_controls
@attr.s()
class _FetchWellKnownFailure(Exception):
# True if we didn't get a non-5xx HTTP response, i.e. this may or may not be
# a temporary failure.
temporary = attr.ib()
|
