b4b067bf12
So we need to keep track of external surface per reporter, since some NATs assign different external IPs for each external destination. Keeping just one known surface could create a race condition.
2015-05-04 18:34:30 -07:00
d3820049b8
Add reported external address to OK(HELLO) TRACE to verify SN behavior.
2015-04-30 18:25:31 -07:00
9eb7698f0e
Learn external IP addresses on OK(HELLO) too.
2015-04-30 16:40:04 -07:00
f5848972f9
Windows now builds and runs selftest correctly, and fixed a Windows (and possibly other platforms) issue in Phy<>.
2015-04-24 15:05:28 -07:00
417f56de2f
Add some TRACE instrumentation to external surface address awareness.
2015-04-17 12:19:01 -07:00
ea1859541c
More cleanup, and fix for the extremely unlikely case of identity collision.
2015-04-15 18:32:25 -07:00
6369c264e2
Rename netconf to controller and NetworkConfigMaster to NetworkController for consistency.
2015-04-15 15:12:09 -07:00
98bcc3d4b5
Disable a few noisy TRACEs, and limit how often we confirm new paths to avoid flooding.
2015-04-15 13:15:09 -07:00
068d311ecc
TRACE compile fixes, other fixes, and it basically works! It says HELLO.
2015-04-09 20:54:00 -07:00
4d5a6a25d3
Add events for packet decode errors, etc., and re-implement TRACE as an event.
2015-04-08 16:49:21 -07:00
52c3b7c34e
Implemented empirical determination of external addressing, paritioned per scope.
2015-04-07 11:56:10 -07:00
817824b88b
Some external surface awareness work, and IP scope classification.
2015-04-07 10:57:59 -07:00
a2821e9000
Add code to check external surface against reported surface from other trusted peers, and also rename ExternalSurface to SelfAwareness because lulz.
2015-04-06 20:17:21 -07:00
f4fd2d4971
Bring IncomingPacket into line with new changes.
2015-04-06 14:50:53 -07:00
1f28ce3980
Tons more refactoring: simplify Network, move explicit management of Tap out, redo COM serialization, etc.
2015-04-01 19:09:18 -07:00
b6fba5934a
RedisNetworkConfigMaster in its own folder. Also fix some hex/decimal Redis database confusion.
2015-02-24 14:17:57 -08:00
ff255a34de
Make NetworkConfigMaster a plugin to get Redis and other non-endpoint code out of node/
2015-02-24 12:28:58 -08:00
93012b0ee5
Re-incorporation: ZeroTier Networks -> ZeroTier, Inc. [Delaware]
2015-02-17 13:11:34 -08:00
b1bf3f68c3
Drop support for legacy P5 multicast, as there are fewer than 1% of these remaining on the network.
2015-02-02 16:34:01 -08:00
0b84c10ccc
Add confirmation step to new netconf, with the caveat that it will be disabled for older netconf servers to avoid race. Also add some comments.
2015-01-09 16:35:20 -05:00
afea153a0b
Build fixes.
2015-01-08 18:17:02 -05:00
64ba596e0b
C++ network config master ready to test.
2015-01-08 14:27:55 -08:00
4e95384ad6
Cleanup, add tristate to config code in Network, and happy new year!
2015-01-05 17:47:59 -08:00
56cfe1d603
Strip out old Service code, add new service message type.
2015-01-05 11:47:22 -08:00
ff539c22f9
locallyValidate() is expensive -- stop doing it on every HELLO since in most cases we already know the identity and know it is valid
2014-12-16 09:29:40 -08:00
8c64046a53
docs
2014-11-26 14:59:43 -08:00
15d3e383e6
Add ZT_SUPPORT_LEGACY_MULTICAST ifdef to enable the legacy code to all be toggled.
2014-11-25 12:46:51 -08:00
7619b0ecbd
Send multicasts in random order.
...
This should not affect most users, but on large networks it should cause service
announcements to work a lot better. This is the result of a prolonged discussion
with a user about the visibility of game servers on a large network. The old
multicast algorithm was de-facto randomized due to its distributed nature, while
the new algorithm is more deterministic. This will restore some randomization
beyond limit-overflow conditions.
It won't affect small networks at all.
2014-11-21 10:50:27 -08:00
c61e9c0ef9
Prevent "software laser" in legacy multicast support. Already hotpatched in supernodes.
2014-11-13 14:21:06 -05:00
5484cf4309
More cleanup, and fix a bug in Multicaster::gather()
2014-10-29 16:24:19 -07:00
22d8aa4dc9
Moderate efficiency improvement on multicast gather result parsing, and go ahead and keep track of total known peers.
2014-10-29 15:26:32 -07:00
0e47f13f14
Simplify locking semantics some more to address a deadlock.
2014-10-21 10:42:04 -07:00
42d644a57e
More fixes to legacy support, and to a potential issue on quit.
2014-10-14 12:37:35 -07:00
2416491cbc
Permanently retire peers.persist, but make iddb.d always enabled instead since identities are what we really want to cache.
2014-10-13 14:12:51 -07:00
8b0846d077
Delete bunch of commented out code.
2014-10-11 16:53:21 -07:00
0d017c043f
Stop persisting last announcement time since Multicaster is volatile. Also some more legacy multicast fixes.
2014-10-11 16:26:02 -07:00
c2aac69a9f
Fixes to legacy peer support.
2014-10-11 15:49:31 -07:00
e071c05f1b
Add a sanity limit to legacy multicast repeater function in supernode-mode nodes, and change netconf-master to issue multicast limit (ml) instead of old p5 stuff.
2014-10-10 14:59:07 -07:00
1774e615a1
<= MTU
2014-10-10 09:09:56 -07:00
56f8f8aa24
Return self in GATHER requests if self is a member of multicast group, and reinstate legacy support.
2014-10-09 18:32:05 -07:00
4941c8a1f3
New multicast bug fixes, TRACE improvements, and temporarily disable legacy multicast for debugging purposes.
2014-10-09 17:58:31 -07:00
d5e0f7e3e4
Reorg multicast packet, and a whole bunch of refactoring around the pushing of certificates of membership.
2014-10-09 12:42:25 -07:00
620e64c58f
Supernode propagation of legacy multicast frames was brokezored.
2014-10-09 09:03:12 -07:00
87f1b1b1e3
Bug fix in new multicast frame handler, handling of old "P5" multicast frames in new way.
2014-10-06 13:16:16 -07:00
ab22feba9a
Bump version to 1.0.0, add legacy support code to Multicaster to not send new frame to known-to-be-old peers.
2014-10-05 10:34:25 -07:00
3f7e7e8a88
Do not multicast to self.
2014-10-03 18:42:41 -07:00
1109046782
Last steps before test: parse OK(MULTICAST_GATHER) and OK(MULTICAST_FRAME)
2014-10-02 13:50:37 -07:00
49dc47ff38
Make multicast gathering a bit smarter.
2014-10-02 11:35:37 -07:00
23836d4c11
Change "encrypted" flag to full cipher suite selector. Go ahead and reserve AES256-GCM which might be added in the future.
2014-10-02 10:54:34 -07:00
e53d208ea4
Improve security posture by eliminating non-const data() accessor from Buffer.
2014-10-02 10:06:29 -07:00
