3c655a4b84
Default route ready to test on Mac.
10 years ago
b90e66f7c7
ManagedRoute, which applies C++ RAII to injected routes. Move RoutingTable to attic.
10 years ago
4446dbde5e
Big refactor in service code to prep for plumbing through route management.
10 years ago
536feb632c
Support yet another old version of http-parser for CentOS/EPEL.
10 years ago
4da9d4e53a
Fix build against Debian version of libhttp-parser -- its old but it will work.
10 years ago
587b1e05d1
Replace cluster-geo subprocess with in-memory loaded CSV of GeoIP data. This is faster, cheaper, more reliable. We use https://db-ip.com/ but others would work too.
10 years ago
63b62ebf4c
Boost local interface check interval a bit to speed direct link provisioning on startup.
10 years ago
569de22481
Make port mapper names even more unique, and trial bind before binding TCP now which should work on Linux and others.
10 years ago
c7df61a3c3
dev branch now works again on Linux
10 years ago
504738b183
Binder for Windows, and use Binder to get local interface addresses to advertise.
10 years ago
32cd2a02c9
OneService now binds all ports to specific local interfaces instead of wildcard and rebinds on changes. (default route)
10 years ago
4a109658ab
More refactoring and prep for explicit interface bindings (for default route support).
10 years ago
0e26917c72
Tweak metric and clean out attic a bit.
10 years ago
1fe251d0a0
stupid bug is stupid
10 years ago
09fc9e4d0c
indent fix
10 years ago
fe3a84a422
Fix problems with previous commit.
10 years ago
ba56a5b9d1
Another NAT-t improvement:
...
Many NATs revert to symmetric behavior from friendlier modes if they cannot
preserve ports. This can occur if there is, for example, more than one ZT
device behind the NAT using port 9993.
Others (Airport Extreme?) seem to have bugs in which they completely freak
out if more than one device tries to do a lot of mappings using the same
internal local port.
Mostly to fix the latter case and somewhat to fix the former, we introduce
a secondary port. ZeroTier now binds 9993 (or whatever port you specify)
plus another port computed deterministically from your ZeroTier address.
This port is used for new links 1/4 of the time.
This mostly addresses the second problem above and partly helps to address
the first.
If uPnP/NAT-PMP is enabled we also still open a tertiary port because some
routers freak out if NAT-t is attempted using the same port as uPnP.
All of this is IPv4 only of course. IPv6 is sane.
10 years ago
bcaf42e075
Transfer speed increate + stability fixes
10 years ago
4e4fd51117
boring doc stuff
10 years ago
d6f0f1a82a
Use network user ptr in lookup for Ethernet frame handling to eliminate map lookup.
10 years ago
83ef98a9dc
Add a network-associated user ptr in API.
10 years ago
b3e3d4cacc
Instead of using binary packet comparison, add a callback to the API to explicitly check whether paths should be used. Check in with this callback (if present) when learning new paths or sending initial packets.
10 years ago
eea634a432
Refactor to eliminate copypasta.
10 years ago
ab19e19f00
Fix a bug that we visually found in Windows code -- it was not advertising uPnP addresses?!?
10 years ago
436c1fac1d
Selectively move over changes from "edge" to "dev" excluding netcon.
10 years ago
3b22ecc80c
Make hex device address in UPnP name the right length.
10 years ago
425d311ce8
Merge of netcon with edge and unification of make processes.
10 years ago
944fdfb65e
Low-impact sanity check against GitHub issue #247 -- will likely prevent other weird recursions too.
10 years ago
40a4ba6e39
indentation
10 years ago
6768521330
Ton of uPnP work and adding NAT-PMP support, still testing...
10 years ago
038476c412
Fix for GitHub issue #253 .
...
Make 'ip' detection more robust (some distros don't ship the binary in /sbin or /usr/sbin).
Add error handling for cases where an ip add/remove fails.
10 years ago
54f68280bd
Minor Windows build fixes. Builds on VS2012 again.
10 years ago
451b8aa7b2
Query both root and network controller for multicast last resort GATHER.
10 years ago
2cc50bdb10
Try bringing back TTL escalation -- may help with Docker (IP-MASQ) type NAT
11 years ago
35c4e28f31
Mark geo-redirected paths as suboptimal and do not report that we have a peer if all we have is one of these. Also a few other small fixes.
11 years ago
83c6b7ab3d
Use bigger UDP buffer on bigger x64 machines. Keep old value elsewhere since these are likely to be smaller boxes.
11 years ago
3e22fee842
Integrate deferred background processing into main OneService implementation.
11 years ago
a994573a43
Eliminate some more dead code. We may do path trust, but not like that.
11 years ago
16bc3e0398
Factor out RemotePath subclass of Path -- no longer needed, just cruft.
11 years ago
7295fcfa86
Merge Phy<> from netcon.
11 years ago
0ffbd05c0e
--wtf; prevent roots from TCP fallback
11 years ago
29b966894c
(1) Fix bug in geo-ip service that prevented cache lookup, (2) fix problem in SelfAwareness (will need to test ALL versions in the wild with this), and (3) add more TRACE instrumentation to Cluster.
11 years ago
7711eba297
More cluster wiring...
11 years ago
cae58f43f1
More World stuff, and mkworld.
11 years ago
7394ec6f6a
Prep in controller code to run tests.
11 years ago
5076c49210
Peer serialization and related changes.
11 years ago
72e7e36a5b
No reason to randomly pick uPnP secondary port. In fact it would likely cause problems on restarts and uPnP rule bloat.
11 years ago
a7409850d6
Get trim() out of core where it is not needed.
11 years ago
cc10d4f310
Speed up multicast group checking/registration.
11 years ago
f69454ec98
(1) Make ZT_ naming convention consistent (get rid of ZT1_), (2) Make local interface a full sockaddr_storage instead of an int identifier, which turns out to be better for multi-homing and other uses.
11 years ago
Adam Ierymenko
Joseph Henry
Adrian Frühwirth
Adam Ierymenko