Bigger changes coming later, but this should improve things a bit in
the meantime.
Rationale:
* 2 minutes -> 45 seconds: 2 minutes was overkill and never considered
phones/battery at the time. It was totally arbitrary. 45 seconds is
also arbitrary but is less than 2 minutes.
* heartbeat from 2 seconds to 3 seconds: in practice this meant two
packets per second (2 pings and 2 pongs every 2 seconds) because the
other side was also pinging us every 2 seconds on their own.
That's just overkill. (see #540 too)
So in the worst case before: when we sent a single packet (say: a DNS
packet), we ended up sending 61 packets over 2 minutes: the 1 DNS
query and then then 60 disco pings (2 minutes / 2 seconds) & received
the same (1 DNS response + 60 pongs). Now it's 15. In 1.22 we plan to
remove this whole timer-based heartbeat mechanism entirely.
The 5 seconds to 6.5 seconds change is just stretching out that
interval so you can still miss two heartbeats (other 3 + 3 seconds
would be greater than 5 seconds). This means that if your peer moves
without telling you, you can have a path out for 6.5 seconds
now instead of 5 seconds before disco finds a new one. That will also
improve in 1.22 when we start doing UDP+DERP at the same time
when confidence starts to go down on a UDP path.
Updates #3363
Change-Id: Ic2314bbdaf42edcdd7103014b775db9cf4facb47
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
Treat UDP send EPERM errors as a lost UDP packet, not something super
fatal. That's just the Linux firewall preventing it from going out.
And add a leaf package net/neterror for that (and future) policy that
all three packages can share, with tests.
Updates #3619
Change-Id: Ibdb838c43ee9efe70f4f25f7fc7fdf4607ba9c1d
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
Only if the source address isn't on the currently active interface or
a ping of the DERP server fails.
Updates #3619
Change-Id: I6bf06503cff4d781f518b437c8744ac29577acc8
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
We only tracked the transport type (UDP vs DERP), not what they were.
Change-Id: Ia4430c1c53afd4634e2d9893d96751a885d77955
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
And it updates the build tag style on a couple files.
Change-Id: I84478d822c8de3f84b56fa1176c99d2ea5083237
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
It's been a bunch of releases now since the TailscaleIPs slice
replacement was added.
Change-Id: I3bd80e1466b3d9e4a4ac5bedba8b4d3d3e430a03
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
endpoint.discoKey is protected by endpoint.mu.
endpoint.sendDiscoMessage was reading it without holding the lock.
This showed up in a CI failure and is readily reproducible locally.
The fix is in two parts.
First, for Conn.enqueueCallMeMaybe, eliminate the one-line helper method endpoint.sendDiscoMessage; call Conn.sendDiscoMessage directly.
This makes it more natural to read endpoint.discoKey in a context
in which endpoint.mu is already held.
Second, for endpoint.sendDiscoPing, explicitly pass the disco key
as an argument. Again, this makes it easier to read endpoint.discoKey
in a context in which endpoint.mu is already held.
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
I believe that this should eliminate the flakiness.
If GitHub CI manages to be even slower that can be believed
(and I can believe a lot at this point),
then we should roll this back and make some more invasive changes.
Updates #654Fixes#3247 (I hope)
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
We can do the "maybe delete" check unilaterally:
In the case of an insert, both oldDiscoKey
and ep.discoKey will be the zero value.
And since we don't use pi again, we can skip
giving it a name, which makes scoping clearer.
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
wgengine/wgcfg: introduce wgcfg.NewDevice helper to disable roaming
at all call sites (one real plus several tests).
Fixestailscale/corp#3016.
Signed-off-by: David Anderson <danderson@tailscale.com>
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
And annotate magicsock as a start.
And add localapi and debug handlers with the Prometheus-format
exporter.
Updates #3307
Change-Id: I47c5d535fe54424741df143d052760387248f8d3
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
There are a few remaining uses of testing.AllocsPerRun:
Two in which we only log the number of allocations,
and one in which dynamically calculate the allocations
target based on a different AllocsPerRun run.
This also allows us to tighten the "no allocs"
test in wgengine/filter.
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
And the derper change to add a CORS endpoint for latency measurement.
And a little magicsock change to cut down some log spam on js/wasm.
Updates #3157
Change-Id: I5fd9e6f5098c815116ddc8ac90cbcd0602098a48
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
Be DERP-only for now. (WebRTC can come later :))
Updates #3157
Change-Id: I56ebb3d914e37e8f4ab651306fd705b817ca381c
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
Now that peerMap tracks the set of nodes for a DiscoKey.
Updates #3088
Change-Id: I927bf2bdfd2b8126475f6b6acc44bc799fcb489f
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
Continuation of 2aa5df7ac1, remove nil
check because it can never be nil. (It previously was able to be nil.)
Change-Id: I59cd9ad611dbdcbfba680ed9b22e841b00c9d5e6
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
This adds new fields (currently unused) to discoInfo to track what the
last verified (unambiguous) NodeKey a DiscoKey last mapped to, and
when.
Then on CallMeMaybe, Pong and on most Pings, we update the mapping
from DiscoKey to the current NodeKey for that DiscoKey.
Updates #3088
Change-Id: Idc4261972084dec71cf8ec7f9861fb9178eb0a4d
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
This lets clients quickly (sub-millisecond within a local LAN) map
from an ambiguous disco key to a node key without waiting for a
CallMeMaybe (over relatively high latency DERP).
Updates #3088
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
https://github.com/tailscale/tailscale/pull/3014 added a
rebind on STUN failure, which means there can now be a
tailscale.com/wgengine/magicsock.(*RebindingUDPConn).ReadFromNetaddr
in progress at the end of the test waiting for a STUN
response which will never arrive.
This causes a test flake due to the resource leak in those
cases where the Conn decided to rebind. For whatever reason,
it mostly flakes with Windows.
If the Conn is closed, don't Rebind after a send error.
Signed-off-by: Denton Gentry <dgentry@tailscale.com>
Renames only; continuation of earlier 8049063d35
These kept confusing me while working on #3088
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
The one remaining caller of peerMap.endpointForDiscoKey was making the
improper assumption that there's exactly 1 node with a given DiscoKey
in the network. That was the cause of #3088.
Now that all the other callers have been updated to not use
endpointForDiscoKey, there's no need to try to keep maintaining that
prone-to-misuse index.
Updates #3088
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
A DiscoKey maps 1:n to endpoints. When we get a disco pong, we don't
necessarily know which endpoint sent it to us. Ask them all. There
will only usually be 1 (and in rare circumstances 2). So it's easier
to ask all two rather than building new maps from the random ping TxID
to its endpoint.
Updates #3088
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
We can reply to a ping without knowing which exact node it's from. As
long as it's in our netmap, it's safe to reply. If there's more than
one node with that discokey, it doesn't matter who we're relpying to.
Updates #3088
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
As more prep for removing the false assumption that you're able to
map from DiscoKey to a single peer, move the lastPingFrom and lastPingTime
fields from the endpoint type to a new discoInfo type, effectively upgrading
the old sharedDiscoKey map (which only held a *[32]byte nacl precomputed key
as its value) to discoInfo which then includes that naclbox key.
Then start plumbing it into handlePing in prep for removing the need
for handlePing to take an endpoint parameter.
Updates #3088
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
The pass just after in this method handles cleaning up sharedDiscoKey.
No need to do it wrong (assuming DiscoKey => 1 node) earlier.
Updates #3088
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
It's not valid to assume that a discokey is globally unique.
This removes the first two of the four callers.
Updates #3088
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
On iOS (and possibly other platforms), sometimes our UDP socket would
get stuck in a state where it was bound to an invalid interface (or no
interface) after a network reconfiguration. We can detect this by
actually checking the error codes from sending our STUN packets.
If we completely fail to send any STUN packets, we know something is
very broken. So on the next STUN attempt, let's rebind the UDP socket
to try to correct any problems.
This fixes a problem where iOS would sometimes get stuck using DERP
instead of direct connections until the backend was restarted.
Fixes#2994
Signed-off-by: Avery Pennarun <apenwarr@tailscale.com>
AFAICT this was always present, the log read mid-execution was never safe.
But it seems like the recent magicsock refactoring made the race much
more likely.
Signed-off-by: David Anderson <danderson@tailscale.com>
And add health check errors to ipnstate.Status (tailscale status --json).
Updates #2746
Updates #2775
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
Over time, other magicsock refactors have made Start effectively a
no-op, except that some other functions choose to panic if called
before Start.
Signed-off-by: David Anderson <danderson@tailscale.com>
We were returning an error almost, but not quite like errConnClosed in
a single codepath, which could still trip the panic on reconfig in the
test logic.
Signed-off-by: David Anderson <danderson@tailscale.com>
Our prod code doesn't eagerly handshake, because our disco layer enables
on-demand handshaking. Configuring both peers to eagerly handshake leads
to WireGuard handshake races that make TestTwoDevicePing flaky.
Signed-off-by: David Anderson <danderson@tailscale.com>
It only existed to override one test-only behavior with a
different test-only behavior, in both cases working around
an annoying feature of our CI environments. Instead, handle
that weirdness entirely in the test code, with a tweaked
TestOnlyPacketListener that gets injected.
Signed-off-by: David Anderson <danderson@tailscale.com>
The docstring said it was meant for use in tests, but it's specifically a
special codepath that is _only_ used in tests, so make the claim stronger.
Signed-off-by: David Anderson <danderson@tailscale.com>
Instead of using the legacy codepath, teach discoEndpoint to handle
peers that have a home DERP, but no disco key. We can still communicate
with them, but only over DERP.
Signed-off-by: David Anderson <danderson@tailscale.com>
magicsock makes multiple calls to Now per packet.
Move to mono.Now. Changing some of the calls to
use package mono has a cascading effect,
causing non-per-packet call sites to also switch.
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
logBufWriter had no serialization.
It just so happens that none of its users currently ever log concurrently.
Make it safe for concurrent use.
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
The DERPTestPort int meant two things before: which port to use, and
whether to disable TLS verification. Users would like to set the port
without disabling TLS, so break it into two options.
Updates #1264
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
After allowing for custom DERP maps, it's convenient to be able to see their latency in
netcheck. This adds a query to the local tailscaled for the current DERPMap.
Updates #1264
Signed-off-by: julianknodt <julianknodt@gmail.com>
Pull in the latest version of wireguard-windows.
Switch to upstream wireguard-go.
This requires reverting all of our import paths.
Unfortunately, this has to happen at the same time.
The wireguard-go change is very low risk,
as that commit matches our fork almost exactly.
(The only changes are import paths, CI files, and a go.mod entry.)
So if there are issues as a result of this commit,
the first place to look is wireguard-windows changes.
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
magicsock.Conn.ParseEndpoint requires a peer's public key,
disco key, and legacy ip/ports in order to do its job.
We currently accomplish that by:
* adding the public key in our wireguard-go fork
* encoding the disco key as magic hostname
* using a bespoke comma-separated encoding
It's a bit messy.
Instead, switch to something simpler: use a json-encoded struct
containing exactly the information we need, in the form we use it.
Our wireguard-go fork still adds the public key to the
address when it passes it to ParseEndpoint, but now the code
compensating for that is just a couple of simple, well-commented lines.
Once this commit is in, we can remove that part of the fork
and remove the compensating code.
Signed-off-by: Josh Bleecher Snyder <josharian@gmail.com>
Fields rename only.
Part of the general effort to make our code agnostic about endpoint formatting.
It's just a name, but it will soon be a misleading one; be more generic.
Do this as a separate commit because it generates a lot of whitespace changes.
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
Upstream wireguard-go renamed the interface method
from CreateEndpoint to ParseEndpoint.
I missed some comments. Fix them.
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
Legacy endpoints (addrSet) currently reconstruct their dst string when requested.
Instead, store the dst string we were given to begin with.
In addition to being simpler and cheaper, this makes less code
aware of how to interpret endpoint strings.
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
For historical reasons, we ended up with two near-duplicate
copies of curve25519 key types, one in the wireguard-go module
(wgcfg) and one in the tailscale module (types/wgkey).
Then we moved wgcfg to the tailscale module.
We can now remove the wgcfg key type in favor of wgkey.
Signed-off-by: Josh Bleecher Snyder <josharian@gmail.com>
One of the consequences of the bind refactoring in 6f23087175
is that attempting to bind an IPv6 socket will always
result in c.pconn6.pconn being non-nil.
If the bind fails, it'll be set to a placeholder packet conn
that blocks forever.
As a result, we can always run ReceiveIPv6 and health check it.
This removes IPv4/IPv6 asymmetry and also will allow health checks
to detect any IPv6 receive func failures.
Signed-off-by: Josh Bleecher Snyder <josharian@gmail.com>
It must be an IP address; enforce that at the type level.
Suggested-by: Brad Fitzpatrick <bradfitz@tailscale.com>
Signed-off-by: Josh Bleecher Snyder <josharian@gmail.com>
We had two separate code paths for the initial UDP listener bind
and any subsequent rebinds.
IPv6 got left out of the rebind code.
Rather than duplicate it there, unify the two code paths.
Then improve the resulting code:
* Rebind had nested listen attempts to try the user-specified port first,
and then fall back to :0 if that failed. Convert that into a loop.
* Initial bind tried only the user-specified port.
Rebind tried the user-specified port and 0.
But there are actually three ports of interest:
The one the user specified, the most recent port in use, and 0.
We now try all three in order, as appropriate.
* In the extremely rare case in which binding to port 0 fails,
use a dummy net.PacketConn whose reads block until close.
This will keep the wireguard-go receive func goroutine alive.
As a pleasant side-effect of this, if we decide that
we need to resuscitate #1796, it will now be much easier.
Fixes#1799
Co-authored-by: David Anderson <danderson@tailscale.com>
Signed-off-by: Josh Bleecher Snyder <josharian@gmail.com>
Assume it'll stay at 0 forever, so hard-code it
and delete code conditional on it being non-0.
Signed-off-by: Josh Bleecher Snyder <josharian@gmail.com>
It was set to context.Background by all callers, for the same reasons.
Set it locally instead, to simplify call sites.
Signed-off-by: Josh Bleecher Snyder <josharian@gmail.com>
The old implementation knew too much about how wireguard-go worked.
As a result, it missed genuine problems that occurred due to unrelated bugs.
This fourth attempt to fix the health checks takes a black box approach.
A receive func is healthy if one (or both) of these conditions holds:
* It is currently running and blocked.
* It has been executed recently.
The second condition is required because receive functions
are not continuously executing. wireguard-go calls them and then
processes their results before calling them again.
There is a theoretical false positive if wireguard-go go takes
longer than one minute to process the results of a receive func execution.
If that happens, we have other problems.
Updates #1790
Signed-off-by: Josh Bleecher Snyder <josharian@gmail.com>
They were not doing their job.
They need yet another conceptual re-think.
Start by clearing the decks.
Signed-off-by: Josh Bleecher Snyder <josharian@gmail.com>
The existing implementation was completely, embarrassingly conceptually broken.
We aren't able to see whether wireguard-go's receive function goroutines
are running or not. All we can do is model that based on what we have done.
This commit fixes that model.
Fixes#1781
Signed-off-by: Josh Bleecher Snyder <josharian@gmail.com>
Avery reported a sub-ms health transition from "receiveIPv4 not running" to "ok".
To avoid these transient false-positives, be more precise about
the expected lifetime of receive funcs. The problematic case is one in which
they were started but exited prior to a call to connBind.Close.
Explicitly represent started vs running state, taking care with the order of updates.
Signed-off-by: Josh Bleecher Snyder <josharian@gmail.com>
We were accidentally logging oldPort -> oldPort.
Log oldPort as well as c.port; if we failed to get the preferred port
in a previous rebind, oldPort might differ from c.port.
Signed-off-by: Josh Bleecher Snyder <josharian@gmail.com>
Track endpoints internally with a new tailcfg.Endpoint type that
includes a typed netaddr.IPPort (instead of just a string) and
includes a type for how that endpoint was discovered (STUN, local,
etc).
Use []tailcfg.Endpoint instead of []string internally.
At the last second, send it to the control server as the existing
[]string for endpoints, but also include a new parallel
MapRequest.EndpointType []tailcfg.EndpointType, so the control server
can start filtering out less-important endpoint changes from
new-enough clients. Notably, STUN-discovered endpoints can be filtered
out from 1.6+ clients, as they can discover them amongst each other
via CallMeMaybe disco exchanges started over DERP. And STUN endpoints
change a lot, causing a lot of MapResposne updates. But portmapped
endpoints are worth keeping for now, as they they work right away
without requiring the firewall traversal extra RTT dance.
End result will be less control->client bandwidth. (despite negligible
increase in client->control bandwidth)
Updates tailscale/corp#1543
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
It existed to work around the frequent opening and closing
of the conn.Bind done by wireguard-go.
The preceding commit removed that behavior,
so we can simply close the connections
when we are done with them.
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
We don't use the port that wireguard-go passes to us (via magicsock.connBind.Open).
We ignore it entirely and use the port we selected.
When we tell wireguard-go that we're changing the listen_port,
it calls connBind.Close and then connBind.Open.
And in the meantime, it stops calling the receive functions,
which means that we stop receiving and processing UDP and DERP packets.
And that is Very Bad.
That was never a problem prior to b3ceca1dd7,
because we passed the SkipBindUpdate flag to our wireguard-go fork,
which told wireguard-go not to re-bind on listen_port changes.
That commit eliminated the SkipBindUpdate flag.
We could write a bunch of code to work around the gap.
We could add background readers that process UDP and DERP packets when wireguard-go isn't.
But it's simpler to never create the conditions in which wireguard-go rebinds.
The other scenario in which wireguard-go re-binds is device.Down.
Conveniently, we never call device.Down. We go from device.Up to device.Close,
and the latter only when we're shutting down a magicsock.Conn completely.
Rubber-ducked-by: Avery Pennarun <apenwarr@tailscale.com>
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
Upstream wireguard-go has changed its receive model.
NewDevice now accepts a conn.Bind interface.
The conn.Bind is stateless; magicsock.Conns are stateful.
To work around this, we add a connBind type that supports
cheap teardown and bring-up, backed by a Conn.
The new conn.Bind allows us to specify a set of receive functions,
rather than having to shoehorn everything into ReceiveIPv4 and ReceiveIPv6.
This lets us plumbing DERP messages directly into wireguard-go,
instead of having to mux them via ReceiveIPv4.
One consequence of the new conn.Bind layer is that
closing the wireguard-go device is now indistinguishable
from the routine bring-up and tear-down normally experienced
by a conn.Bind. We thus have to explicitly close the magicsock.Conn
when the close the wireguard-go device.
One downside of this change is that we are reliant on wireguard-go
to call receiveDERP to process DERP messages. This is fine for now,
but is perhaps something we should fix in the future.
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
The code is not obviously better or worse, but this makes the little warning
triangle in my editor go away, and the distraction removal is worth it.
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
It can end up executing an a new goroutine,
at which point instead of immediately stopping test execution, it hangs.
Since this is unexpected anyway, panic instead.
As a bonus, it makes call sites nicer and removes a kludge comment.
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
The tstun packagen contains both constructors for generic tun
Devices, and a wrapper that provides additional functionality.
Signed-off-by: David Anderson <danderson@tailscale.com>
e.g.
$ tailscale ping 1.1.1.1
exit node found but not enabled
$ tailscale ping 10.2.200.2
node "tsbfvlan2" found, but not using its 10.2.200.0/24 route
$ sudo tailscale up --accept-routes
$ tailscale ping 10.2.200.2
pong from tsbfvlan2 (100.124.196.94) via 10.2.200.34:41641 in 1ms
$ tailscale ping mon.ts.tailscale.com
pong from monitoring (100.88.178.64) via DERP(sfo) in 83ms
pong from monitoring (100.88.178.64) via DERP(sfo) in 21ms
pong from monitoring (100.88.178.64) via [2604:a880:4:d1::37:d001]:41641 in 22ms
This necessarily moves code up from magicsock to wgengine, so we can
look at the actual wireguard config.
Fixes#1564
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
No server support yet, but we want Tailscale 1.6 clients to be able to respond
to them when the server can do it.
Updates #1310
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
There was a logical race where Conn.Rebind could acquire the
RebindingUDPConn mutex, close the connection, fail to rebind, release
the mutex, and then because the mutex was no longer held, ReceiveIPv4
wouldn't retry reads that failed with net.ErrClosed, letting that
error back to wireguard-go, which would then stop running that receive
IP goroutine.
Instead, keep the RebindingUDPConn mutex held for the entirety of the
replacement in all cases.
Updates tailscale/corp#1289
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
Also change the type to netaddr.IP while here, because it made sorting
easier.
Updates tailscale/corp#1397
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
UIs need to see the full unedited netmap in order to know what exit nodes they
can offer to the user.
Signed-off-by: David Anderson <danderson@tailscale.com>
* move probing out of netcheck into new net/portmapper package
* use PCP ANNOUNCE op codes for PCP discovery, rather than causing
short-lived (sub-second) side effects with a 1-second-expiring map +
delete.
* track when we heard things from the router so we can be less wasteful
in querying the router's port mapping services in the future
* use portmapper from magicsock to map a public port
Fixes#1298Fixes#1080Fixes#1001
Updates #864
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
When a handshake race occurs, a queued data packet can get lost.
TestTwoDevicePing expected that the very first data packet would arrive.
This caused occasional flakes.
Change TestTwoDevicePing to repeatedly re-send packets
and succeed when one of them makes it through.
This is acceptable (vs making WireGuard not drop the packets)
because this only affects communication with extremely old clients.
And those extremely old clients will eventually connect,
because the kernel will retry sends on timeout.
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
We modified the standard net package to not allocate a *net.UDPAddr
during a call to (*net.UDPConn).ReadFromUDP if the caller's use
of the *net.UDPAddr does not cause it to escape.
That is https://golang.org/cl/291390.
This is the companion change to magicsock.
There are two changes required.
First, call ReadFromUDP instead of ReadFrom, if possible.
ReadFrom returns a net.Addr, which is an interface, which always allocates.
Second, reduce the lifetime of the returned *net.UDPAddr.
We do this by immediately converting it into a netaddr.IPPort.
We left the existing RebindingUDPConn.ReadFrom method in place,
as it is required to satisfy the net.PacketConn interface.
With the upstream change and both of these fixes in place,
we have removed one large allocation per packet received.
name old time/op new time/op delta
ReceiveFrom-8 16.7µs ± 5% 16.4µs ± 8% ~ (p=0.310 n=5+5)
name old alloc/op new alloc/op delta
ReceiveFrom-8 112B ± 0% 64B ± 0% -42.86% (p=0.008 n=5+5)
name old allocs/op new allocs/op delta
ReceiveFrom-8 3.00 ± 0% 2.00 ± 0% -33.33% (p=0.008 n=5+5)
Co-authored-by: Sonia Appasamy <sonia@tailscale.com>
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
addrSet maintained duplicate lists of netaddr.IPPorts and net.UDPAddrs.
Unify to use the netaddr type only.
This makes (*Conn).ReceiveIPvN a bit uglier,
but that'll be cleaned up in a subsequent commit.
This is preparatory work to remove an allocation from ReceiveIPv4.
Co-authored-by: Sonia Appasamy <sonia@tailscale.com>
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
I based my estimation of the required timeout based on locally
observed behavior. But CI machines are worse than my local machine.
16s was enough to reduce flakiness but not eliminate it. Bump it up again.
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
We removed the "fast retry" code from our wireguard-go fork.
As a result, pings can take longer to transit when retries are required.
Allow that.
Fixes#1277
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
The fix can make this test run unconditionally.
This moves code from 5c619882bc for
testability but doesn't fix it yet. The #1282 problem remains (when I
wrote its wake-up mechanism, I forgot there were N DERP readers
funneling into 1 UDP reader, and the code just isn't correct at all
for that case).
Also factor out some test helper code from BenchmarkReceiveFrom.
The refactoring in magicsock.go for testability should have no
behavior change.
Magicsock started dropping all traffic internally when Tailscale is
shut down, to avoid spurious wireguard logspam. This made the benchmark
not receive anything. Setting a dummy private key is sufficient to get
magicsock to pass traffic for benchmarking purposes.
Fixes#1270.
Signed-off-by: David Anderson <danderson@tailscale.com>
Use tb.Cleanup to simplify both the API and the implementation.
One behavior change: When the number of goroutines shrinks, don't log.
I've never found these logs to be useful, and they frequently add noise.
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
This is mostly code movement from the wireguard-go repo.
Most of the new wgcfg package corresponds to the wireguard-go wgcfg package.
wgengine/wgcfg/device{_test}.go was device/config{_test}.go.
There were substantive but simple changes to device_test.go to remove
internal package device references.
The API of device.Config (now wgcfg.DeviceConfig) grew an error return;
we previously logged the error and threw it away.
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
Rewrite log lines on the fly, based on the set of known peers.
This enables us to use upstream wireguard-go logging,
but maintain the Tailscale-style peer public key identifiers
that the rest of our systems (and people) expect.
Fixes#1183
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
Also, don't try to use IPv6 LinkLocalUnicast addresses for now. Like endpoints
exchanged with control, we share them but don't yet use them.
Updates #1172
c8c493f3d9 made it always say
`created=false` which scared me when I saw it, as that would've implied
things were broken much worse. Fortunately the logging was just wrong.
DstToString is used in two places in wireguard-go: Logging and uapi.
We are switching to use uapi for wireguard-go config.
To preserve existing behavior, we need the full set of addrs.
And for logging, having the full set of addrs seems useful.
(The Addrs method itself is slated for removal. When that happens,
the implementation will move to DstToString.)
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
To save CPU and wakeups, don't run the DERP cleanup timer regularly
unless there is a non-home DERP connection open.
Also eliminates the goroutine, moving to a time.AfterFunc.
Updates #1034
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
This reverts commit 08baa17d9a.
It caused deadlocks due to lock ordering violations.
It was not the right fix, and thus should simply be reverted
while we look for the right fix (if we haven't already found it
in the interim; we've fixed other logging-after-test issues).
Fixes#1161
context.cancelCtx.Done involves a mutex and isn't as cheap as I
previously assumed. Convert the donec method into a struct field and
store the channel value once. Our one magicsock.Conn gets one pointer
larger, but it cuts ~1% of the CPU time of the ReceiveFrom benchmark
and removes a bubble from the --svg output :)
22507adf54 stopped relying on
our fork of wireguard-go's UpdateDst callback.
As a result, we can unwind that code,
and the extra return value of ReceiveIPv{4,6}.
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
TwoDevicePing is explicitly testing the behavior of the legacy codepath, everything
else is happy to assume that code no longer exists.
Signed-off-by: David Anderson <danderson@tailscale.com>
Previously, this benchmark relied on behavior of the legacy
receive codepath, which I changed in 22507adf. With this
change, the benchmark instead relies on the new active discovery
path.
Signed-off-by: David Anderson <danderson@tailscale.com>
This prevents us from continuing to do unnecessary work
(including logging) after the connection has closed.
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
This eliminates a dependency on wgcfg.Endpoint,
as part of the effort to eliminate our wireguard-go fork.
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
This makes connectivity between ancient and new tailscale nodes slightly
worse in some cases, but only in cases where the ancient version would
likely have failed to get connectivity anyway.
Signed-off-by: David Anderson <danderson@tailscale.com>
In sendDiscoMessage there is a check of whether the connection is
closed, which is not being reliably exercised by other tests.
This shows up in code coverage reports, the lines of code in
sendDiscoMessage are alternately added and subtracted from
code coverage.
Add a test to specifically exercise and verify this code path.
Signed-off-by: Denton Gentry <dgentry@tailscale.com>
In derpWriteChanOfAddr when we call derphttp.NewRegionClient(),
there is a check of whether the connection is already errored and
if so it returns before grabbing the lock. The lock might already
be held and would be a deadlock.
This corner case is not being reliably exercised by other tests.
This shows up in code coverage reports, the lines of code in
derpWriteChanOfAddr are alternately added and subtracted from
code coverage.
Add a test to specifically exercise this code path, and verify that
it doesn't deadlock.
This is the best tradeoff I could come up with:
+ the moment code calls Err() to check if there is an error, we
grab the lock to make sure it would deadlock if it tries to grab
the lock itself.
+ if a new call to Err() is added in this code path, only the
first one will be covered and the rest will not be tested.
+ this test doesn't verify whether code is checking for Err() in
the right place, which ideally I guess it would.
Signed-off-by: Denton Gentry <dgentry@tailscale.com>
netaddr.IP no longer allocates, so don't need a cache or all its associated
code/complexity.
This totally removes groupcache/lru from the deps.
Also go mod tidy.
Not usefully functional yet (mostly a proof of concept), but getting
it submitted for some work @namansood is going to do atop this.
Updates #707
Updates #634
Updates #48
Updates #835
* show DNS name over hostname, removing domain's common MagicDNS suffix.
only show hostname if there's no DNS name.
but still show shared devices' MagicDNS FQDN.
* remove nerdy low-level details by default: endpoints, DERP relay,
public key. They're available in JSON mode still for those who need
them.
* only show endpoint or DERP relay when it's active with the goal of
making debugging easier. (so it's easier for users to understand
what's happening) The asterisks are gone.
* remove Tx/Rx numbers by default for idle peers; only show them when
there's traffic.
* include peers' owner login names
* add CLI option to not show peers (matching --self=true, --peers= also
defaults to true)
* sort by DNS/host name, not public key
* reorder columns
This is a replacement for the key-related parts
of the wireguard-go wgcfg package.
This is almost a straight copy/paste from the wgcfg package.
I have slightly changed some of the exported functions and types
to avoid stutter, added and tweaked some comments,
and removed some now-unused code.
To avoid having wireguard-go depend on this new package,
wgcfg will keep its key types.
We translate into and out of those types at the last minute.
These few remaining uses will be eliminated alongside
the rest of the wgcfg package.
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
The previous code used a lot of whole-function variables and shared
behavior that only triggered based on prior action from a single codepath.
Instead of that, move the small amounts of "shared" code into each switch
case.
Signed-off-by: David Anderson <danderson@tailscale.com>
Before, tailscaled would log every 10 seconds when the periodic noteRecvActivity
call happens. This is noisy, but worse it's misleading, because the message
suggests that the disco code is starting a lazy config run for a missing peer,
whereas in fact it's just an internal piece of keepalive logic.
With this change, we still log when going from 0->1 tunnel for the peer, but
not every 10s thereafter.
Signed-off-by: David Anderson <danderson@tailscale.com>
In tests, we force binding to localhost to avoid OS firewall warning
dialogs.
But for IPv6, we were trying (and failing) to bind to 127.0.0.1.
You'd think we'd just say "localhost", but that's apparently ill
defined. See
https://tools.ietf.org/html/draft-ietf-dnsop-let-localhost-be-localhost
and golang/go#22826. (It's bitten me in the past, but I can't
remember specific bugs.)
So use "::1" explicitly for "udp6", which makes the test quieter.
We still use the packet.* alloc-free types in the data path, but
the compilation from netaddr to packet happens within the filter
package.
Signed-off-by: David Anderson <danderson@tailscale.com>
At startup the client doesn't yet have the DERP map so can't do STUN
queries against DERP servers, so it only knows it local interface
addresses, not its STUN-mapped addresses.
We were reporting the interface-local addresses to control, getting
the DERP map, and then immediately reporting the full set of
updates. That was an extra HTTP request to control, but worse: it was
an extra broadcast from control out to all the peers in the network.
Now, skip the initial update if there are no stun results and we don't
have a DERP map.
More work remains optimizing start-up requests/map updates, but this
is a start.
Updates tailscale/corp#557
If no interfaces are up, calm down and stop spamming so much. It was
noticed as especially bad on Windows, but probably was bad
everywhere. I just have the best network conditions testing on a
Windows VM.
Updates #604
Updates #654. See that issue for a discussion of why
this timeout reduces flakiness, and what next steps are.
Signed-off-by: Josh Bleecher Snyder <josh@tailscale.com>
Rather than consider bigs jumps in last-received-from activity as a
signal to possibly reconfigure the set of wireguard peers to have
configured, instead just track the set of peers that are currently
excluded from the configuration. Easier to reason about.
Also adds a bit more logging.
This might fix an error we saw on a machine running a recent unstable
build:
2020-08-26 17:54:11.528033751 +0000 UTC: 8.6M/92.6M magicsock: [unexpected] lazy endpoint not created for [UcppE], d:42a770f678357249
2020-08-26 17:54:13.691305296 +0000 UTC: 8.7M/92.6M magicsock: DERP packet received from idle peer [UcppE]; created=false
2020-08-26 17:54:13.691383687 +0000 UTC: 8.7M/92.6M magicsock: DERP packet from unknown key: [UcppE]
If it does happen again, though, we'll have more logs.
Seems to break linux CI builder. Cannot reproduce locally,
so attempting a rollback.
This reverts commit cd7bc02ab1.
Signed-off-by: David Crawshaw <crawshaw@tailscale.com>
Without this, a freshly started ipn client will be stuck in the
"Starting" state until something triggers a call to RequestStatus.
Usually a UI does this, but until then we can sit in this state
until poked by an external event, as is evidenced by our e2e tests
locking up when DERP is attached.
(This only recently became a problem when we enabled lazy handshaking
everywhere, otherwise the wireugard tunnel creation would also
trigger a RequestStatus.)
Signed-off-by: David Crawshaw <crawshaw@tailscale.com>
Consider:
Hard NAT (A) <---> Hard NAT w/ mapped port (B)
If A sends a packet to B's mapped port, A can disco ping B directly,
with low latency, without DERP.
But B couldn't establish a path back to A and needed to use DERP,
despite already logging about A's endpoint and adding a mapping to it
for other purposes (the wireguard conn.Endpoint lookup also needed
it).
This adds the tracking to discoEndpoint too so it'll be used for
finding a path back.
Fixestailscale/corp#556
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
For example:
$ tailscale ping -h
USAGE
ping <hostname-or-IP>
FLAGS
-c 10 max number of pings to send
-stop-once-direct true stop once a direct path is established
-verbose false verbose output
$ tailscale ping mon.ts.tailscale.com
pong from monitoring (100.88.178.64) via DERP(sfo) in 65ms
pong from monitoring (100.88.178.64) via DERP(sfo) in 252ms
pong from monitoring (100.88.178.64) via [2604:a880:2:d1::36:d001]:41641 in 33ms
Fixes#661
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
1) we weren't waking up a discoEndpoint that once existed and
went idle for 5 minutes and then got a disco message again.
2) userspaceEngine.noteReceiveActivity had a buggy check; fixed
and added a test
This removes the atomic bool that tried to track whether we needed to acquire
the lock on a future recursive call back into magicsock. Unfortunately that
hack doesn't work because we also had a lock ordering issue between magicsock
and userspaceEngine (see issue). This documents that too.
Fixes#644
If a node is behind a hard NAT and is using an explicit local port
number, assume they might've mapped a port and add their public IPv4
address with the local tailscaled's port number as a candidate endpoint.
Starting with fe68841dc7, some e2e tests
got flaky. Rather than debug them (they're gnarly), just revert to the old
behavior as far as those tests are concerned. The tests were somehow
using magicsock without a private key and expecting it to do ... something.
My goal with fe68841dc7 was to stop log spam
and unnecessary work I saw on the iOS app when when stopping the app.
Instead, only stop doing that work on any transition from
once-had-a-private-key to no-longer-have-a-private-key. That fixes
what I wanted to fix while still making the mysterious e2e tests
happy.
There is a race in natlab where we might start shutdown while natlab is still running
a goroutine or two to deliver packets. This adds a small grace period to try and receive
it before continuing shutdown.
Signed-off-by: David Anderson <danderson@tailscale.com>
The first packet to transit may take several seconds to do so, because
setup rates in wgengine may result in the initial WireGuard handshake
init to get dropped. So, we have to wait at least long enough for a
retransmit to correct the fault.
Signed-off-by: David Anderson <danderson@tailscale.com>
Active discovery lets us introspect the state of the network stack precisely
enough that it's unnecessary, and dropping the initial DERP packets greatly
slows down tests. Additionally, it's unrealistic since our production network
will never deliver _only_ discovery packets, it'll be all or nothing.
Signed-off-by: David Anderson <danderson@tailscale.com>
Uses natlab only, because the point of this active discovery test is going to be
that it should get through a lot of obstacles.
Signed-off-by: David Anderson <danderson@tailscale.com>
The deadlock was:
* Conn.Close was called, which acquired c.mu
* Then this goroutine scheduled:
if firstDerp {
startGate = c.derpStarted
go func() {
dc.Connect(ctx)
close(c.derpStarted)
}()
}
* The getRegion hook for that derphttp.Client then ran, which also
tries to acquire c.mu.
This change makes that hook first see if we're already in a closing
state and then it can pretend that region doesn't exist.
wireguard-go uses 3 goroutines per peer (with reasonably large stacks
& buffers).
Rather than tell wireguard-go about all our peers, only tell it about
peers we're actively communicating with. That means we need hooks into
magicsock's packet receiving path and tstun's packet sending path to
lazily create a wireguard peer on demand from the network map.
This frees up lots of memory for iOS (where we have almost nothing
left for larger domains with many users).
We should ideally do this in wireguard-go itself one day, but that'd
be a pretty big change.
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
Brad Fitzpatrick