Commit Graph

44 Commits

Author SHA1 Message Date
Erik Johnston
91caa5b430 Fix off by one error in SRV result shuffling 2019-08-27 13:56:42 +01:00
Erik Johnston
fbb758a7ce Fixup comments 2019-08-23 15:37:20 +01:00
Erik Johnston
e70f0081da Fix logcontexts 2019-08-23 15:37:20 +01:00
Erik Johnston
1e4b4d85e7 Merge branch 'develop' of github.com:matrix-org/synapse into erikj/reliable_lookups 2019-08-22 13:41:57 +01:00
Erik Johnston
74f016d343 Remove now unused pick_server_from_list 2019-08-20 12:37:08 +01:00
Erik Johnston
1f9df1cc7b Fixup _sort_server_list to be slightly more efficient
Also document that we are using the algorithm described in RFC2782 and
ensure we handle zero weight correctly.
2019-08-20 12:36:11 +01:00
Erik Johnston
1dec31560e Change jitter to be a factor rather than absolute value 2019-08-20 11:46:00 +01:00
Erik Johnston
861d663c15 Fixup changelog and remove debug logging 2019-08-16 13:15:26 +01:00
Erik Johnston
f299c5414c Refactor MatrixFederationAgent to retry SRV.
This refactors MatrixFederationAgent to move the SRV lookup into the
endpoint code, this has two benefits:
	1. Its easier to retry different host/ports in the same way as
	   HostnameEndpoint.
	2. We avoid SRV lookups if we have a free connection in the pool
2019-08-15 15:43:22 +01:00
Erik Johnston
e6e136decc Retry well known on fail.
If we have recently seen a valid well-known for a domain we want to
retry on (non-final) errors a few times, to handle temporary blips in
networking/etc.
2019-08-15 09:28:58 +01:00
Erik Johnston
17e1e80726 Retry well-known lookup before expiry.
This gives a bit of a grace period where we can attempt to refetch a
remote `well-known`, while still using the cached result if that fails.

Hopefully this will make the well-known resolution a bit more torelant
of failures, rather than it immediately treating failures as "no result"
and caching that for an hour.
2019-08-13 16:20:38 +01:00
Erik Johnston
f218705d2a Make default well known cache global again. 2019-08-13 10:06:51 +01:00
Erik Johnston
107ad133fc Move well known lookup into a separate clas 2019-08-07 15:36:38 +01:00
Erik Johnston
af9f1c0764 Add a lower bound for TTL on well known results.
It costs both us and the remote server for us to fetch the well known
for every single request we send, so we add a minimum cache period. This
is set to 5m so that we still honour the basic premise of "refetch
frequently".
2019-08-06 17:01:23 +01:00
Erik Johnston
a9bcae9f50 Share SSL options for well-known requests 2019-07-31 10:39:24 +01:00
Amber Brown
4806651744
Replace returnValue with return (#5736) 2019-07-23 23:00:55 +10:00
Amber Brown
463b072b12
Move logging utilities out of the side drawer of util/ and into logging/ (#5606) 2019-07-04 00:07:04 +10:00
Amber Brown
32e7c9e7f2
Run Black. (#5482) 2019-06-20 19:32:02 +10:00
Andrew Morgan
6824ddd93d Config option for verifying federation certificates (MSC 1711) (#4967) 2019-04-25 14:22:49 +01:00
Andrew Morgan
ac61b45a75 Minor docstring fixes for MatrixFederationAgent (#4765) 2019-02-28 16:24:01 +00:00
Richard van der Hoff
fa794980ec
Merge pull request #4544 from matrix-org/rav/skip_invalid_well_known
Treat an invalid .well-known the same as an absent one
2019-02-01 13:18:36 +00:00
Richard van der Hoff
e9779a6f8f Fix b'ab' noise in logs 2019-02-01 12:34:31 +00:00
Richard van der Hoff
8a21b03fba Treat an invalid .well-known the same as an absent one
... basically, carry on and fall back to SRV etc.
2019-02-01 11:37:31 +00:00
Richard van der Hoff
3c8a41140e Cache failures to parse .well-known
Also add a Measure block around the .well-known fetch
2019-02-01 00:37:52 +00:00
Richard van der Hoff
24d59c7568 better logging for federation connections 2019-01-31 23:18:20 +00:00
Richard van der Hoff
d428b46346 Update federation routing logic to check .well-known before SRV 2019-01-31 23:14:18 +00:00
Richard van der Hoff
a79034aedf
Merge pull request #4521 from matrix-org/rav/fed_routing/cleanups
Tiny .well-known fixes
2019-01-30 11:47:24 +00:00
Richard van der Hoff
c7b24ac3d0
Follow redirects on .well-known (#4520) 2019-01-30 11:43:33 +00:00
Richard van der Hoff
09a1a6b55e fix exception text 2019-01-30 10:58:52 +00:00
Richard van der Hoff
928c50b59a Also jitter the invalid cache period 2019-01-30 10:58:52 +00:00
Richard van der Hoff
bc5f6e1797
Add a caching layer to .well-known responses (#4516) 2019-01-30 10:55:25 +00:00
Richard van der Hoff
cc2d650ef7
Relax requirement for a content-type on .well-known (#4511) 2019-01-29 16:49:17 +00:00
Richard van der Hoff
99e36d5e24
Implement MSC1708 (.well-known lookups for server routing) (#4489) 2019-01-29 13:53:02 +00:00
Richard van der Hoff
3bd0f1a4a3 docstrings for _RoutingResult 2019-01-28 12:43:09 +00:00
Richard van der Hoff
0fd5b3b53e Handle IP literals explicitly
We don't want to be doing .well-known lookups on these guys.
2019-01-28 10:34:30 +00:00
Richard van der Hoff
51958df766 MatrixFederationAgent: factor out routing logic
This is going to get too big and unmanageable.
2019-01-28 10:34:30 +00:00
Richard van der Hoff
d840019192
Fix idna and ipv6 literal handling in MatrixFederationAgent (#4487)
Turns out that the library does a better job of parsing URIs than our
reinvented wheel. Who knew.

There are two things going on here. The first is that, unlike
parse_server_name, URI.fromBytes will strip off square brackets from IPv6
literals, which means that it is valid input to ClientTLSOptionsFactory and
HostnameEndpoint.

The second is that we stay in `bytes` throughout (except for the argument to
ClientTLSOptionsFactory), which avoids the weirdness of (sometimes) ending up
with idna-encoded values being held in `unicode` variables. TBH it probably
would have been ok but it made the tests fragile.
2019-01-28 09:56:59 +00:00
Richard van der Hoff
8520bc3109
Fix Host header sent by MatrixFederationAgent (#4468)
Move the Host header logic down here so that (a) it is used if we reuse the
agent elsewhere, and (b) we can mess about with it with .well-known.
2019-01-25 12:38:16 +00:00
Richard van der Hoff
afd69a0920 Look up the right SRV record 2019-01-24 13:31:43 +00:00
Richard van der Hoff
7021784d46 put resolve_service in an object
this makes it easier to stub things out for tests.
2019-01-22 20:35:12 +00:00
Richard van der Hoff
53a327b4d5 Require that service_name be a byte string
it is only ever a bytes now, so let's enforce that.
2019-01-22 20:35:12 +00:00
Richard van der Hoff
fe212bbe4a Kill off matrix_federation_endpoint
this thing is now redundant.
2019-01-22 20:34:35 +00:00
Richard van der Hoff
44be7513bf MatrixFederationAgent
Pull the magic that is currently in matrix_federation_endpoint and friends into
an agent-like thing
2019-01-22 20:34:35 +00:00
Richard van der Hoff
33a55289cb
Refactor and bugfix for resove_service (#4427) 2019-01-22 10:59:27 +00:00