9284e4edfe
agree() must be called on our identity, the one with the secret
2017-02-01 15:22:14 -08:00
62a705af1c
Eliminate another check in cluster frontplane mode.
2017-02-01 14:35:07 -08:00
29ec7bf3a2
Add more specific check in source==self case instead of dumping it.
2017-02-01 14:18:56 -08:00
fc3f4fb988
Yeah that could never have worked (normal packets in cluster mode).
2017-02-01 14:05:13 -08:00
60ff280dcb
Another tweak to cluster I/O rules.
2017-02-01 13:52:53 -08:00
b378f5dcd7
Take 3
2017-02-01 13:20:51 -08:00
e778d45128
Still want to send WANT_PEER under two failure modes.
2017-02-01 12:51:52 -08:00
5e11cf6378
Can't armor() a packet until all flags are set.
2017-02-01 12:32:06 -08:00
5dbebc513a
Minor send path refactor to make packet I/O work on clusters if they are members of networks. Also fix a crash if compiled in cluster mode but no cluster is enabled.
2017-02-01 12:00:25 -08:00
0b3b994241
Relay policy can now be computed.
2017-01-27 14:05:09 -08:00
64774d0d4f
Replace piecemeal designation of upstreams with the concept of moons, which is simpler and easier to use and inherits all the cool live update stuff of worlds (now called planets) and global roots.
2017-01-27 13:27:52 -08:00
84732fcb12
Wire through external path lookup. Static paths should now work.
2016-11-22 14:23:13 -08:00
42ba70e79e
Replace long callback arg list with struct, and implement path whitelisting, path blacklisting, and local.conf support for roles.
2016-11-22 10:54:58 -08:00
39333c9e8e
Modify unite() to deal with a second layer of upstreams.
2016-11-17 16:59:04 -08:00
1615ef1114
Rename getBestRoot() etc.
2016-11-17 16:31:58 -08:00
d5f4d381d0
Go ahead and loop back packets whose destination is self. Some OSes require this since they aactually follow the full network path even for local IPs.
2016-10-05 10:12:06 -07:00
e1fbf7b34c
Check multicast limit on send after NDP emulation code.
2016-09-28 12:21:08 -07:00
cc4bacc199
Cleanup, and implement compression disable flag for networks.
2016-09-27 12:22:25 -07:00
d3524f3609
Refactor COM stuff a bit, and respond to COM requests a bit more readily for rapid setup. Will need to revisit later.
2016-09-20 21:21:34 -07:00
5b6d27e659
Implement relay policy, and setting multicast limit to 0 now disables multicast on the network as would be expected.
2016-09-13 14:27:18 -07:00
ef87069957
Fix gating of multicast GATHER replies since these can come from upstream, etc., and fix an issue with sending ECHO to recheck marginal paths.
2016-09-09 09:32:00 -07:00
0d4109a9f1
More refactoring to clean up code, and add a gate function to make sure we do not handle OK packets we did not expect. This hardens up a few potential edge cases around security, since such messages might be used to e.g. pollute a cache and DOS under certain conditions.
2016-09-09 08:43:58 -07:00
b5c86b6ba4
Bunch more path refactoring. Peers no longer forget paths, but do not normally use expired paths. Expired paths might still be tried if nothing else is reachable.
2016-09-07 11:13:17 -07:00
01aa469591
Remove debug line.
2016-09-02 14:26:04 -07:00
412979ba8f
Attempt to reactivate dead paths.
2016-09-02 13:55:33 -07:00
4f8253dcdb
Tweaks to path handling...
2016-09-02 13:33:56 -07:00
e8f6b4b5d3
Rest of big Path canonicalization refactor.
2016-09-02 11:51:33 -07:00
a3bdae9735
Work in progress: Path canonicalization refactor.
2016-09-01 15:43:07 -07:00
d5e6f59004
.
2016-09-01 13:45:32 -07:00
74afef8eb1
Think through and refine a few things in rules, especially edge case TEE and REDIRECT behavior and semantics.
2016-08-31 16:50:22 -07:00
c476285bd6
Harden PUSH_DIRECT_PATHS and simplify things by only doing it on receive when hops>0 and trust has been established.
2016-08-24 16:16:39 -07:00
2cdda38dc4
It basically works... at least on current controllers.
2016-08-24 15:26:18 -07:00
e1310a764a
More cleanup and removal of cruft due to obsolete network-specific relays (will be replaced with federation stuff).
2016-08-09 15:45:26 -07:00
00fd9c3a15
It builds... almost ready to test some rules engine stuff.
2016-08-08 17:33:26 -07:00
98152d974a
More cleanup and removal of DeferredPackets, will do the latter in a more elegant way.
2016-08-04 11:40:38 -07:00
91940cbcf5
Kill network preferred relays -- this feature is gone (and was seldom used anyway) in favor of federation.
2016-08-02 14:40:26 -07:00
7404eb46c4
Integration of Filter into inbound and outbound packet path.
2016-07-25 16:51:10 -07:00
765082fdb6
Trusted path support, and version bump to 1.1.9
2016-07-12 08:29:50 -07:00
20d155e630
.
2016-06-24 05:21:25 -07:00
be5996daca
Cleanup of 6plane stuff.
2016-06-24 04:54:05 -07:00
38dfebad8c
IPv6 NDP emulation flag in NetworkConfig, and implement Docker-friendly
...
(and other host friendly) IPv6 /80 magic subnetting to allow massive
multicast-free NDP emulated IPv6 networks where each host can have a
/48 worth of IPv6 IPs for internal containers, VMs, etc.
Alan Kay, thou art avenged.
https://ivanovivan.wordpress.com/2010/09/13/alan-kay-quotes/
2016-06-23 22:41:14 -07:00
8b9519f0af
Simplify a bunch of NetworkConfig stuff by eliminating accessors, also makes network controller easier to refactor.
2016-05-06 16:13:11 -07:00
b9dba97fdb
Bunch more refactoring for an even more compact NetworkConfig representation, especially rules.
2016-04-26 17:11:25 -07:00
2f18a92e20
Cleanup in numerous places, reduce network chattiness around MULTICAST_LIKE, and fix a "how was that working" latent bug causing some control traffic to take the scenic route.
2016-04-19 12:09:35 -07:00
e28838805f
Brace fix...
2016-04-12 13:17:16 -07:00
31db6f8f36
NetworkConfig refactor almost done.
2016-04-12 12:49:46 -07:00
56096be8b6
Tweak new RX queue algorithm to "expire" old entries to prevent always needing to traverse the whole queue array.
2016-04-06 16:28:40 -07:00
76f9b4c582
A few other little formatting things.
2016-03-18 15:06:45 -07:00
c589bc4c57
Indentation
2016-03-18 14:32:48 -07:00
d6a1868d0a
Refactor incoming packet (rxQueue/fragmentQueue) to eliminate variable length queues and merge queues. This is both faster and saves memory.
2016-03-18 14:16:07 -07:00
