As we now partition proxies by NAT type, our stats are more useful if they
capture how many proxies of each type we have, and information on
whether we have enough proxies of the right NAT type for our clients.
This change adds proxy counts by NAT type and binned counts of denied clients by NAT type.
The client and proxy use the net/http default transport to make round
trip connecitons to the broker. These by default don't time out and can
wait indefinitely for the broker to respond if the broker hangs and
doesn't terminate the connection.
This will allow browser-based proxies that are unable to determine their
NAT type to conservatively label themselves as restricted NATs if they
fail to work with clients that have restricted NATs.
Now when proxies poll, they provide their NAT type to the broker. This
introduces a new snowflake heap of just restricted snowflakes that the
broker can pull from if the client has a known, unrestricted NAT. All
other clients will pull from a heap of snowflakes with unrestricted or
unknown NAT topologies.
Snowflake clients will now attempt NAT discovery using the provided STUN
servers and report their NAT type to the Snowflake broker for matching.
The three possibilities for NAT types are:
- unknown (the client was unable to determine their NAT type),
- restricted (the client has a restrictive NAT and can only be paired
with unrestricted NATs)
- unrestricted (the client can be paired with any other NAT).
The underlying smux layer sends a keep-alive ping every 10 seconds. This
modification will allow for one dropped/delayed ping before discarding
the snowflake
It was sticking out in the context of other log messages.
2020/04/30 22:39:10 WebRTC: DataChannel created.
2020/04/30 22:39:20 establishDataChannel: timeout waiting for DataChannel.OnOpen
2020/04/30 22:39:20 WebRTC: closing PeerConnection
2020/04/30 22:39:20 WebRTC: Closing
2020/04/30 22:39:20 WebRTC: WebRTC: Could not establish DataChannel Retrying in 10s...
Now callers cannot call Write without there being a DataChannel to write
to. This lets us remove the internal buffer and checks for transport ==
nil.
Don't set internal fields like writePipe, transport, and pc to nil when
closing; just close them and let them return errors if further calls are
made on them.
There's now a constant DataChannelTimeout that's separate from
SnowflakeTimeout (the latter is what checkForStaleness uses). Now we can
set DataChannel timeout to a lower value, to quickly dispose of
unconnectable proxies, while still keeping the threshold for detecting
the failure of a once-working proxy at 30 seconds.
https://bugs.torproject.org/33897
Formerly, preparePeerConnection set up a callback that sent into a
channel, and exchangeSDP waited until it could receive from the channel.
We can move the channel entirely into preparePeerConnection (having it
not return until the callback has been called) and that way remove some
shared state.
Do it as a side effect of NewWebRTCPeer.
Remove WebRTCPeer tests as they currently require invasively modifying
internal fields at different stages of construction.
The other interfaces in client/lib/interfaces.go exist for the purpose
of running tests, but not Snowflake. Existing code would not have worked
with other types anyway, because it does unchecked .(*WebRTCPeer)
conversions.
A short write will result in a non-nil error. It's an io.PipeWriter
anyway, which blocks until all the data has been read or the read end is
closed, in which case it returns io.ErrClosedPipe if not some other
error.
This allows multiple SOCKS connections to share the available proxies,
and in particular prevents a SOCKS connection from being starved of a
proxy when the maximum proxy capacity is less then the number of the
number of SOCKS connections.
This is option 4 from https://bugs.torproject.org/33519.
The difficulty here is that the whole point of turbotunnel sessions is
that they are not necessarily tied to a single WebSocket connection, nor
even a single client IP address. We use a heuristic: whenever a
WebSocket connection starts that has a new ClientID, we store a mapping
from that ClientID to the IP address attached to the WebSocket
connection in a lookup table. Later, when enough packets have arrived to
establish a turbotunnel session, we recover the ClientID associated with
the session (which kcp-go has stored in the RemoteAddr field), and look
it up in the table to get an IP address. We introduce a new data type,
clientIDMap, to store the clientID-to-IP mapping during the short time
between when a WebSocket connection starts and handleSession receives a
fully fledged KCP session.
The client opts into turbotunnel mode by sending a magic token at the
beginning of each WebSocket connection (before sending even the
ClientID). The token is just a random byte string I generated. The
server peeks at the token and, if it matches, uses turbotunnel mode.
Otherwise, it unreads the token and continues in the old
one-session-per-WebSocket mode.
Formerly we waiting until *both* directions finished. What this meant in
practice is that when the remote connection ended, copyLoop would become
useless but would continue blocking its caller until something else
finally closed the socks connection.
