Instead of just falling back to making a TCP query to an upstream DNS
server when the UDP query returns a truncated query, also start a TCP
query in parallel with the UDP query after a given race timeout. This
ensures that if the upstream DNS server does not reply over UDP (or if
the response packet is blocked, or there's an error), we can still make
queries if the server replies to TCP queries.
This also adds a new package, util/race, to contain the logic required for
racing two different functions and returning the first non-error answer.
Updates tailscale/corp#14809
Signed-off-by: Andrew Dunham <andrew@du.nham.ca>
Change-Id: I4311702016c1093b1beaa31b135da1def6d86316
Implements the ability for the address-rewriting code to support rewriting IPv6 addresses.
Specifically, UpdateSrcAddr & UpdateDstAddr.
Signed-off-by: Tom DNetto <tom@tailscale.com>
Updates https://github.com/tailscale/corp/issues/11202
go-billy is held back at v5.4.1 in order to avoid a newly introduced
subdependency that is not compatible with plan9.
Updates #8043
Signed-off-by: James Tucker <james@tailscale.com>
It had exactly one user: netstack. Just have LocalBackend notify
netstack when here's a new netmap instead, simplifying the bloated
Engine interface that has grown a bunch of non-Engine-y things.
(plenty of rando stuff remains after this, but it's a start)
Updates #cleanup
Change-Id: I45e10ab48119e962fc4967a95167656e35b141d8
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
Previously two tsnet nodes in the same process couldn't have disjoint
sets of controlknob settings from control as both would overwrite each
other's global variables.
This plumbs a new controlknobs.Knobs type around everywhere and hangs
the knobs sent by control on that instead.
Updates #9351
Change-Id: I75338646d36813ed971b4ffad6f9a8b41ec91560
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
All platforms use it at this point, including iOS which was the
original hold out for memory reasons. No more reason to make it
optional.
Updates #9332
Change-Id: I743fbc2f370921a852fbcebf4eb9821e2bdd3086
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
This PR removes the per request logging to the CLI as the CLI
will not be displaying those logs initially.
Updates #8489
Signed-off-by: Marwan Sulaiman <marwan@tailscale.com>
This PR adds a SessionID field to the ipn.Notify struct so that
ipn buses can identify a session and register deferred clean up
code in the future. The first use case this is for is to be able to
tie foreground serve configs to a specific watch session and ensure
its clean up when a connection is closed.
Updates #8489
Signed-off-by: Marwan Sulaiman <marwan@tailscale.com>
We use it a number of places in different repos. Might as well make
one. Another use is coming.
Updates #cleanup
Change-Id: Ib7ce38de0db35af998171edee81ca875102349a4
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
I noticed that failed tests were leaving aroudn stray tailscaled processes
on macOS at least.
To repro, add this to tstest/integration:
func TestFailInFewSeconds(t *testing.T) {
t.Parallel()
time.Sleep(3 * time.Second)
os.Exit(1)
t.Fatal("boom")
}
Those three seconds let the other parallel tests (with all their
tailscaled child processes) start up and start running their tests,
but then we violently os.Exit(1) the test driver and all the children
were kept alive (and were spinning away, using all available CPU in
gvisor scheduler code, which is a separate scary issue)
Updates #cleanup
Change-Id: I9c891ed1a1ec639fb2afec2808c04dbb8a460e0e
Co-authored-by: Maisem Ali <maisem@tailscale.com>
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
As a fallback to package managers, allow updating tailscale that was
self-installed in some way. There are some tricky bits around updating
the systemd unit (should we stick to local binary paths or to the ones
in tailscaled.service?), so leaving that out for now.
Updates #6995
Signed-off-by: Andrew Lytvynov <awly@tailscale.com>
They were entirely redundant and 1:1 with the status field
so this turns them into methods instead.
Updates #cleanup
Updates #1909
Change-Id: I7d939750749edf7dae4c97566bbeb99f2f75adbc
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
I'm not saying it works, but it compiles.
Updates #5794
Change-Id: I2f3c99732e67fe57a05edb25b758d083417f083e
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
The Windows Security Center is a component that manages the registration of
security products on a Windows system. Only products that have obtained a
special cert from Microsoft may register themselves using the WSC API.
Practically speaking, most vendors do in fact sign up for the program as it
enhances their legitimacy.
From our perspective, this is useful because it gives us a high-signal
source of information to query for the security products installed on the
system. I've tied this query into the osdiag package and is run during
bugreports.
It uses COM bindings that were automatically generated by my prototype
metadata processor, however that program still has a few bugs, so I had
to make a few manual tweaks. I dropped those binding into an internal
package because (for the moment, at least) they are effectively
purpose-built for the osdiag use case.
We also update the wingoes dependency to pick up BSTR.
Fixes#10646
Signed-off-by: Aaron Klotz <aaron@tailscale.com>
Adds ability to start Funnel in the foreground and stream incoming
connections. When foreground process is stopped, Funnel is turned
back off for the port.
Exampe usage:
```
TAILSCALE_FUNNEL_V2=on tailscale funnel 8080
```
Updates #8489
Signed-off-by: Marwan Sulaiman <marwan@tailscale.com>
If a node is flapping or otherwise generating lots of STUN endpoints, we
can end up caching a ton of useless values and sending them to peers.
Instead, let's apply a fixed per-Addr limit of endpoints that we cache,
so that we're only sending peers up to the N most recent.
Updates tailscale/corp#13890
Signed-off-by: Andrew Dunham <andrew@du.nham.ca>
Change-Id: I8079a05b44220c46da55016c0e5fc96dd2135ef8
This removes the unsafe/linkname and only uses the standard library.
It's a bit slower, for now, but https://go.dev/cl/518336 should get us
back.
On darwin/arm64, without https://go.dev/cl/518336
pkg: tailscale.com/tstime/mono
│ before │ after │
│ sec/op │ sec/op vs base │
MonoNow-8 16.20n ± 0% 19.75n ± 0% +21.92% (p=0.000 n=10)
TimeNow-8 39.46n ± 0% 39.40n ± 0% -0.16% (p=0.002 n=10)
geomean 25.28n 27.89n +10.33%
And with it,
MonoNow-8 16.34n ± 1% 16.93n ± 0% +3.67% (p=0.001 n=10)
TimeNow-8 39.55n ± 15% 38.46n ± 1% -2.76% (p=0.000 n=10)
geomean 25.42n 25.52n +0.41%
Updates #8839
Updates tailscale/go#70
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
* We update wingoes to pick up new version information functionality
(See pe/version.go in the https://github.com/dblohm7/wingoes repo);
* We move the existing LogSupportInfo code (including necessary syscall
stubs) out of util/winutil into a new package, util/osdiag, and implement
the public LogSupportInfo function may be implemented for other platforms
as needed;
* We add a new reason argument to LogSupportInfo and wire that into
localapi's bugreport implementation;
* We add module information to the Windows implementation of LogSupportInfo
when reason indicates a bugreport. We enumerate all loaded modules in our
process, and for each one we gather debug, authenticode signature, and
version information.
Fixes#7802
Signed-off-by: Aaron Klotz <aaron@tailscale.com>
The util/linuxfw/iptables.go had a bunch of code that wasn't yet used
(in prep for future work) but because of its imports, ended up
initializing code deep within gvisor that panicked on init on arm64
systems not using 4KB pages.
This deletes the unused code to delete the imports and remove the
panic. We can then cherry-pick this back to the branch and restore it
later in a different way.
A new test makes sure we don't regress in the future by depending on
the panicking package in question.
Fixes#8658
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
This allows sending logs from the "logpolicy" package (and associated
callees) to something other than the log package. The behaviour for
tailscaled remains the same, passing in log.Printf
Updates #8249
Signed-off-by: Andrew Dunham <andrew@du.nham.ca>
Change-Id: Ie1d43b75fa7281933d9225bffd388462c08a5f31
When performing a fallback DNS query, run the recursive resolver in a
separate goroutine and compare the results returned by the recursive
resolver with the results we get from "regular" bootstrap DNS. This will
allow us to gather data about whether the recursive DNS resolver works
better, worse, or about the same as "regular" bootstrap DNS.
Updates #5853
Signed-off-by: Andrew Dunham <andrew@du.nham.ca>
Change-Id: Ifa0b0cc9eeb0dccd6f7a3d91675fe44b3b34bd48
This change is introducing new netfilterRunner interface and moving iptables manipulation to a lower leveled iptables runner.
For #391
Signed-off-by: KevinLiang10 <kevinliang@tailscale.com>
In order to improve our ability to understand the state of policies and
registry settings when troubleshooting, we enumerate all values in all subkeys.
x/sys/windows does not already offer this, so we need to call RegEnumValue
directly.
For now we're just logging this during startup, however in a future PR I plan to
also trigger this code during a bugreport. I also want to log more than just
registry.
Fixes#8141
Signed-off-by: Aaron Klotz <aaron@tailscale.com>
The retry logic was pathological in the following ways:
* If we restarted the logging service, any pending uploads
would be placed in a retry-loop where it depended on backoff.Backoff,
which was too aggresive. It would retry failures within milliseconds,
taking at least 10 retries to hit a delay of 1 second.
* In the event where a logstream was rate limited,
the aggressive retry logic would severely exacerbate the problem
since each retry would also log an error message.
It is by chance that the rate of log error spam
does not happen to exceed the rate limit itself.
We modify the retry logic in the following ways:
* We now respect the "Retry-After" header sent by the logging service.
* Lacking a "Retry-After" header, we retry after a hard-coded period of
30 to 60 seconds. This avoids the thundering-herd effect when all nodes
try reconnecting to the logging service at the same time after a restart.
* We do not treat a status 400 as having been uploaded.
This is simply not the behavior of the logging service.
Updates #tailscale/corp#11213
Signed-off-by: Joe Tsai <joetsai@digital-static.net>
Signed-off-by: Chenyang Gao <gps949@outlook.com>
in commit 6e96744, the tsd system type has been added.
Which will cause the daemon will crash on some OSs (Windows, darwin and so on).
The root cause is that on those OSs, handleSubnetsInNetstack() will return true and set the conf.Router with a wrapper.
Later in NewUserspaceEngine() it will do subsystem set and found that early set router mismatch to current value, then panic.
This is part of an effort to clean up tailscaled initialization between
tailscaled, tailscaled Windows service, tsnet, and the mac GUI.
Updates #8036
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
This holds back gvisor, kubernetes, goreleaser, and esbuild, which all
had breaking API changes.
Updates #8043
Updates #7381
Updates #8042 (updates u-root which adds deps)
Change-Id: I889759bea057cd3963037d41f608c99eb7466a5b
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
This change introduces address selection for wireguard only endpoints.
If a endpoint has not been used before, an address is randomly selected
to be used based on information we know about, such as if they are able
to use IPv4 or IPv6. When an address is initially selected, we also
initiate a new ICMP ping to the endpoints addresses to determine which
endpoint offers the best latency. This information is then used to
update which endpoint we should be using based on the best possible
route. If the latency is the same for a IPv4 and an IPv6 address, IPv6
will be used.
Updates #7826
Signed-off-by: Charlotte Brandhorst-Satzkorn <charlotte@tailscale.com>
On some platforms (notably macOS and iOS) we look up the default
interface to bind outgoing connections to. This is both duplicated
work and results in logspam when the default interface is not available
(i.e. when a phone has no connectivity, we log an error and thus cause
more things that we will try to upload and fail).
Fixed by passing around a netmon.Monitor to more places, so that we can
use its cached interface state.
Fixes#7850
Updates #7621
Signed-off-by: Mihai Parparita <mihai@tailscale.com>
We're using it in more and more places, and it's not really specific to
our use of Wireguard (and does more just link/interface monitoring).
Also removes the separate interface we had for it in sockstats -- it's
a small enough package (we already pull in all of its dependencies
via other paths) that it's not worth the extra complexity.
Updates #7621
Updates #7850
Signed-off-by: Mihai Parparita <mihai@tailscale.com>
Redoes the approach from #5550 and #7539 to explicitly pass in the logf
function, instead of having global state that can be overridden.
Signed-off-by: Mihai Parparita <mihai@tailscale.com>
This splits Prometheus metric handlers exposed by tsweb into two
modules:
- `varz.Handler` exposes Prometheus metrics generated by our expvar
converter;
- `promvarz.Handler` combines our expvar-converted metrics and native
Prometheus metrics.
By default, tsweb will use the promvarz handler, however users can keep
using only the expvar converter. Specifically, `tailscaled` now uses
`varz.Handler` explicitly, which avoids a dependency on the
(heavyweight) Prometheus client.
Updates https://github.com/tailscale/corp/issues/10205
Signed-off-by: Anton Tolchanov <anton@tailscale.com>
The handler will expose built-in process and Go metrics by default,
which currently duplicate some of the expvar-proxied metrics
(`goroutines` vs `go_goroutines`, `memstats` vs `go_memstats`), but as
long as their names are different, Prometheus server will just scrape
both.
This will change /debug/varz behaviour for most tsweb binaries, but
notably not for control, which configures a `tsweb.VarzHandler`
[explicitly](a5b5d5167f/cmd/tailcontrol/tailcontrol.go (L779))
Updates https://github.com/tailscale/corp/issues/10205
Signed-off-by: Anton Tolchanov <anton@tailscale.com>
I realized that a lot of the problems that we're seeing around migration and
LocalBackend state can be avoided if we drive Windows pref migration entirely
from within tailscaled. By doing it this way, tailscaled can automatically
perform the migration as soon as the connection with the client frontend is
established.
Since tailscaled is already running as LocalSystem, it already has access to
the user's local AppData directory. The profile manager already knows which
user is connected, so we simply need to resolve the user's prefs file and read
it from there.
Of course, to properly migrate this information we need to also check system
policies. I moved a bunch of policy resolution code out of the GUI and into
a new package in util/winutil/policy.
Updates #7626
Signed-off-by: Aaron Klotz <aaron@tailscale.com>
This adds the util/sysresources package, which currently only contains a
function to return the total memory size of the current system.
Then, we modify magicsock to scale the number of buffered DERP messages
based on the system's available memory, ensuring that we never use a
value lower than the previous constant of 32.
Signed-off-by: Andrew Dunham <andrew@du.nham.ca>
Change-Id: Ib763c877de4d0d4ee88869078e7d512f6a3a148d
When running a SOCKS or HTTP proxy, configure the tshttpproxy package to
drop those addresses from any HTTP_PROXY or HTTPS_PROXY environment
variables.
Fixes#7407
Signed-off-by: Andrew Dunham <andrew@du.nham.ca>
Change-Id: I6cd7cad7a609c639780484bad521c7514841764b
This adds support to make exit nodes and subnet routers work
when in scenarios where NAT is required.
It also updates the NATConfig to be generated from a `wgcfg.Config` as
that handles merging prefs with the netmap, so it has the required information
about whether an exit node is already configured and whether routes are accepted.
Updates tailscale/corp#8020
Signed-off-by: Maisem Ali <maisem@tailscale.com>
Since users can run tailscaled in a variety of ways (root, non-root,
non-root with process capabilities on Linux), this check will print the
current process permissions to the log to aid in debugging.
Signed-off-by: Andrew Dunham <andrew@du.nham.ca>
Change-Id: Ida93a206123f98271a0c664775d0baba98b330c7
* wgengine/magicsock: add envknob to send CallMeMaybe to non-existent peer
For testing older client version responses to the PeerGone packet format change.
Updates #4326
Signed-off-by: Val <valerie@tailscale.com>
* derp: remove dead sclient struct member replaceLimiter
Leftover from an previous solution to the duplicate client problem.
Updates #2751
Signed-off-by: Val <valerie@tailscale.com>
* derp, derp/derphttp, wgengine/magicsock: add new PeerGone message type Not Here
Extend the PeerGone message type by adding a reason byte. Send a
PeerGone "Not Here" message when an endpoint sends a disco message to
a peer that this server has no record of.
Fixes#4326
Signed-off-by: Val <valerie@tailscale.com>
---------
Signed-off-by: Val <valerie@tailscale.com>
We were checking against the wrong directory, instead if we
have a custom store configured just use that.
Fixes#7588Fixes#7665
Signed-off-by: Maisem Ali <maisem@tailscale.com>
This change focuses on the backend log ID, which is the mostly commonly
used in the client. Tests which don't seem to make use of the log ID
just use the zero value.
Signed-off-by: Will Norris <will@tailscale.com>
This reverts commit 6eca47b16c and fixes forward.
Previously the first ever streaming MapRequest that a client sent would also
set ReadOnly to true as it didn't have any endpoints and expected/relied on the
map poll to restart as soon as it got endpoints. However with 48f6c1eba4,
we would no longer restart MapRequests as frequently as we used to, so control
would only ever get the first streaming MapRequest which had ReadOnly=true.
Control would treat this as an uninteresting request and would not send it
any further netmaps, while the client would happily stay in the map poll forever
while litemap updates happened in parallel.
This makes it so that we never set `ReadOnly=true` when we are doing a streaming
MapRequest. This is no longer necessary either as most endpoint discovery happens
over disco anyway.
Co-authored-by: Andrew Dunham <andrew@du.nham.ca>
Signed-off-by: Maisem Ali <maisem@tailscale.com>
This change adds a ringbuffer to each magicsock endpoint that keeps a
fixed set of "changes"–debug information about what updates have been
made to that endpoint.
Additionally, this adds a LocalAPI endpoint and associated
"debug peer-status" CLI subcommand to fetch the set of changes for a given
IP or hostname.
Updates tailscale/corp#9364
Signed-off-by: Andrew Dunham <andrew@du.nham.ca>
Change-Id: I34f726a71bddd0dfa36ec05ebafffb24f6e0516a
Now that we're using rand.Shuffle in a few locations, create a generic
shuffle function and use it instead. While we're at it, move the
interleaveSlices function to the same package for use.
Signed-off-by: Andrew Dunham <andrew@du.nham.ca>
Change-Id: I0b00920e5b3eea846b6cedc30bd34d978a049fd3
The debug flag on tailscaled isn't available in the macOS App Store
build, since we don't have a tailscaled binary; move it to the
'tailscale debug' CLI that is available on all platforms instead,
accessed over LocalAPI.
Updates #7377
Signed-off-by: Andrew Dunham <andrew@du.nham.ca>
Change-Id: I47bffe4461e036fab577c2e51e173f4003592ff7
Given recent changes in corp, I originally thought we could remove all of the
syso files, but then I realized that we still need them so that binaries built
purely from OSS (without going through corp) will still receive a manifest.
We can remove the arm32 one though, since we don't support 32-bit ARM on Windows.
Updates https://github.com/tailscale/corp/issues/9576
Signed-off-by: Aaron Klotz <aaron@tailscale.com>
Uses the hooks added by tailscale/go#45 to instrument the reads and
writes on the major code paths that do network I/O in the client. The
convention is to use "<package>.<type>:<label>" as the annotation for
the responsible code path.
Enabled on iOS, macOS and Android only, since mobile platforms are the
ones we're most interested in, and we are less sensitive to any
throughput degradation due to the per-I/O callback overhead (macOS is
also enabled for ease of testing during development).
For now just exposed as counters on a /v0/sockstats PeerAPI endpoint.
We also keep track of the current interface so that we can break out
the stats by interface.
Updates tailscale/corp#9230
Updates #3363
Signed-off-by: Mihai Parparita <mihai@tailscale.com>
The log ID types were moved to a separate package so that
code that only depend on log ID types do not need to link
in the logic for the logtail client itself.
Not all code need the logtail client.
Signed-off-by: Joe Tsai <joetsai@digital-static.net>
This updates all source files to use a new standard header for copyright
and license declaration. Notably, copyright no longer includes a date,
and we now use the standard SPDX-License-Identifier header.
This commit was done almost entirely mechanically with perl, and then
some minimal manual fixes.
Updates #6865
Signed-off-by: Will Norris <will@tailscale.com>
They changed a type in their SDK which meant others using the AWS APIs
in their Go programs (with newer AWS modules in their caller go.mod)
and then depending on Tailscale (for e.g. tsnet) then couldn't compile
ipn/store/awsstore.
Thanks to @thisisaaronland for bringing this up.
Fixes#7019
Change-Id: I8d2919183dabd6045a96120bb52940a9bb27193b
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
With a42a594bb3, iOS uses netstack and
hence there are no longer any platforms which use the legacy MagicDNS path. As such, we remove it.
We also normalize the limit for max in-flight DNS queries on iOS (it was 64, now its 256 as per other platforms).
It was 64 for the sake of being cautious about memory, but now we have 50Mb (iOS-15 and greater) instead of 15Mb
so we have the spare headroom.
Signed-off-by: Tom DNetto <tom@tailscale.com>
The macOS client was forgetting to call netstack.Impl.SetLocalBackend.
Change the API so that it can't be started without one, eliminating this
class of bug. Then update all the callers.
Updates #6764
Change-Id: I2b3a4f31fdfd9fdbbbbfe25a42db0c505373562f
Signed-off-by: Claire Wang <claire@tailscale.com>
Co-authored-by: Brad Fitzpatrick <bradfitz@tailscale.com>
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
We've been doing a hard kill of the subprocess, which is only safe as long as
both the cli and gui are not running and the subprocess has had the opportunity
to clean up DNS settings etc. If unattended mode is turned on, this is definitely
unsafe.
I changed babysitProc to close the subprocess's stdin to make it shut down, and
then I plumbed a cancel function into the stdin reader on the subprocess side.
Fixes https://github.com/tailscale/corp/issues/5621
Signed-off-by: Aaron Klotz <aaron@tailscale.com>
This is temporary while we work to upstream performance work in
https://github.com/WireGuard/wireguard-go/pull/64. A replace directive
is less ideal as it breaks dependent code without duplication of the
directive.
Signed-off-by: Jordan Whited <jordan@tailscale.com>
This commit updates the wireguard-go dependency and implements the
necessary changes to the tun.Device and conn.Bind implementations to
support passing vectors of packets in tailscaled. This significantly
improves throughput performance on Linux.
Updates #414
Signed-off-by: Jordan Whited <jordan@tailscale.com>
Signed-off-by: James Tucker <james@tailscale.com>
Co-authored-by: James Tucker <james@tailscale.com>
WinTun is installed lazily by tailscaled while it is running as LocalSystem.
Based upon what we're seeing in bug reports and support requests, removing
WinTun as a lesser user may fail under certain Windows versions, even when that
user is an Administrator.
By adding a user-defined command code to tailscaled, we can ask the service to
do the removal on our behalf while it is still running as LocalSystem.
* The uninstall code is basically the same as it is in corp;
* The command code will be sent as a service control request and is protected by
the SERVICE_USER_DEFINED_CONTROL access right, which requires Administrator.
I'll be adding follow-up patches in corp to engage this functionality.
Updates https://github.com/tailscale/tailscale/issues/6433
Signed-off-by: Aaron Klotz <aaron@tailscale.com>
This handles the case where the inner *os.PathError is wrapped in
another error type, and additionally will redact errors of type
*os.LinkError. Finally, add tests to verify that redaction works.
Signed-off-by: Andrew Dunham <andrew@du.nham.ca>
Change-Id: Ie83424ff6c85cdb29fb48b641330c495107aab7c
renamed from `useNetstack` to `onlyNetstack` which is 1 letter more but
more descriptive because we always have netstack enabled and `useNetstack`
doesn't convey what it is supposed to be used for. e.g. we always use
netstack for Tailscale SSH.
Also renamed shouldWrapNetstack to handleSubnetsInNetstack as it was only used
to configure subnet routing via netstack.
Updates tailscale/corp#8020
Signed-off-by: Maisem Ali <maisem@tailscale.com>
This adds an envknob to make testing async startup more reproducible.
We want the Windows GUI to behave well when wintun is not (or it's
doing its initial slow driver installation), but during testing it's often
too fast to see that it's working. This lets it be slowed down.
Updates #6522
Change-Id: I6ae19f46e270ea679cbaea32a53888efcf2943a7
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
Previously, tstun.Wrapper and magicsock.Conn managed their
own statistics data structure and relied on an external call to
Extract to extract (and reset) the statistics.
This makes it difficult to ensure a maximum size on the statistics
as the caller has no introspection into whether the number
of unique connections is getting too large.
Invert the control flow such that a *connstats.Statistics
is registered with tstun.Wrapper and magicsock.Conn.
Methods on non-nil *connstats.Statistics are called for every packet.
This allows the implementation of connstats.Statistics (in the future)
to better control when it needs to flush to ensure
bounds on maximum sizes.
The value registered into tstun.Wrapper and magicsock.Conn could
be an interface, but that has two performance detriments:
1. Method calls on interface values are more expensive since
they must go through a virtual method dispatch.
2. The implementation would need a sync.Mutex to protect the
statistics value instead of using an atomic.Pointer.
Given that methods on constats.Statistics are called for every packet,
we want reduce the CPU cost on this hot path.
Signed-off-by: Joe Tsai <joetsai@digital-static.net>
Many packages reference the logtail ID types,
but unfortunately pull in the transitive dependencies of logtail.
Fix this problem by putting the log ID types in its own package
with minimal dependencies.
Signed-off-by: Joe Tsai <joetsai@digital-static.net>
tailscaled on Windows had two entirely separate start-up paths for running
as a service vs in the foreground. It's been causing problems for ages.
This unifies the two paths, making them be the same as the path used
for every other platform.
Also, it uses the new async LocalBackend support in ipnserver.Server
so the Server can start serving HTTP immediately, even if tun takes
awhile to come up.
Updates #6535
Change-Id: Icc8c4f96d4887b54a024d7ac15ad11096b5a58cf
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
We use this pattern in a number of places (in this repo and elsewhere)
and I was about to add a fourth to this repo which was crossing the line.
Add this type instead so they're all the same.
Also, we have another Set type (SliceSet, which tracks its keys in
order) in another repo we can move to this package later.
Change-Id: Ibbdcdba5443fae9b6956f63990bdb9e9443cefa9
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
This is step 1 of de-special-casing of Windows and letting the
LocalAPI HTTP server start serving immediately, even while the rest of
the world (notably the Engine and its TUN device) are being created,
which can take a few to dozens of seconds on Windows.
With this change, the ipnserver.New function changes to not take an
Engine and to return immediately, not returning an error, and let its
Run run immediately. If its ServeHTTP is called when it doesn't yet
have a LocalBackend, it returns an error. A TODO in there shows where
a future handler will serve status before an engine is available.
Future changes will:
* delete a bunch of tailscaled_windows.go code and use this new API
* add the ipnserver.Server ServerHTTP handler to await the engine
being available
* use that handler in the Windows GUI client
Updates #6522
Change-Id: Iae94e68c235e850b112a72ea24ad0e0959b568ee
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
We'll eventually remove it entirely, but for now move get it out of ipnserver
where it's distracting and move it to its sole caller.
Updates #6522
Change-Id: I9c6f6a91bf9a8e3c5ea997952b7c08c81723d447
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
It's only used by Windows. No need for it to be in ipn/ipnserver,
which we're trying to trim down.
Change-Id: Idf923ac8b6cdae8b5338ec26c16fb8b5ea548071
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
Follow-up to #6467 and #6506.
LocalBackend knows the server-mode state, so move more auth checking
there, removing some bookkeeping from ipnserver.Server.
Updates #6417
Updates tailscale/corp#8051
Change-Id: Ic5d14a077bf0dccc92a3621bd2646bab2cc5b837
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
This patch removes the crappy, half-backed COM initialization used by `go-ole`
and replaces that with the `StartRuntime` function from `wingoes`, a library I
have started which, among other things, initializes COM properly.
In particular, we should always be initializing COM to use the multithreaded
apartment. Every single OS thread in the process becomes implicitly initialized
as part of the MTA, so we do not need to concern ourselves as to whether or not
any particular OS thread has initialized COM. Furthermore, we no longer need to
lock the OS thread when calling methods on COM interfaces.
Single-threaded apartments are designed solely for working with Win32 threads
that have a message pump; any other use of the STA is invalid.
Fixes https://github.com/tailscale/tailscale/issues/3137
Signed-off-by: Aaron Klotz <aaron@tailscale.com>
Centralize the fake GOOS stuff, start to use it more. To be used more
in the future.
Change-Id: Iabacfbeaf5fca0b53bf4d5dbcdc0367f05a205f9
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
We're trying to gut 90% of the ipnserver package. A lot will get
deleted, some will move to LocalBackend, and a lot is being moved into
this new ipn/ipnauth package which will be leaf-y and testable.
This is a baby step towards moving some stuff to ipnauth.
Update #6417
Updates tailscale/corp#8051
Change-Id: I28bc2126764f46597d92a2d72565009dc6927ee0
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
This uses a go:generate statement to create a bunch of .syso files that
contain a Windows resource file. We check these in since they're less
than 1KiB each, and are only included on Windows.
Fixes#6429
Signed-off-by: Andrew Dunham <andrew@du.nham.ca>
Change-Id: I0512c3c0b2ab9d8d8509cf2037b88b81affcb81f
Run an inotify goroutine and watch if another program takes over
/etc/inotify.conf. Log if so.
For now this only logs. In the future I want to wire it up into the
health system to warn (visible in "tailscale status", etc) about the
situation, with a short URL to more info about how you should really
be using systemd-resolved if you want programs to not fight over your
DNS files on Linux.
Updates #4254 etc etc
Change-Id: I86ad9125717d266d0e3822d4d847d88da6a0daaa
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
Example output:
# Health check:
# - Some peers are advertising routes but --accept-routes is false
Also, move "tailscale status" health checks to the bottom, where they
won't be lost in large netmaps.
Updates #2053
Updates #6266
Change-Id: I5ae76a0cd69a452ce70063875cd7d974bfeb8f1a
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
Leave only the HTTP/auth bits in localapi.
Change-Id: I8e23fb417367f1e0e31483e2982c343ca74086ab
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
This is similar to the golang.org/x/tools/internal/fastwalk I'd
previously written but not recursive and using mem.RO.
The metrics package already had some Linux-specific directory reading
code in it. Move that out to a new general package that can be reused
by portlist too, which helps its scanning of all /proc files:
name old time/op new time/op delta
FindProcessNames-8 2.79ms ± 6% 2.45ms ± 7% -12.11% (p=0.000 n=10+10)
name old alloc/op new alloc/op delta
FindProcessNames-8 62.9kB ± 0% 33.5kB ± 0% -46.76% (p=0.000 n=9+10)
name old allocs/op new allocs/op delta
FindProcessNames-8 2.25k ± 0% 0.38k ± 0% -82.98% (p=0.000 n=9+10)
Change-Id: I75db393032c328f12d95c39f71c9742c375f207a
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
The //go:build syntax was introduced in Go 1.17:
https://go.dev/doc/go1.17#build-lines
gofmt has kept the +build and go:build lines in sync since
then, but enough time has passed. Time to remove them.
Done with:
perl -i -npe 's,^// \+build.*\n,,' $(git grep -l -F '+build')
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
The netlog.Message type is useful to depend on from other packages,
but doing so would transitively cause gvisor and other large packages
to be linked in.
Avoid this problem by moving all network logging types to a single package.
We also update staticcheck to take in:
003d277bcf
Signed-off-by: Joe Tsai <joetsai@digital-static.net>
Turns out using win32 instead of shelling out to child processes is a
bit faster:
name old time/op new time/op delta
GetListIncremental-4 278ms ± 2% 0ms ± 7% -99.93% (p=0.000 n=8+10)
name old alloc/op new alloc/op delta
GetListIncremental-4 238kB ± 0% 9kB ± 0% -96.12% (p=0.000 n=10+8)
name old allocs/op new allocs/op delta
GetListIncremental-4 1.19k ± 0% 0.02k ± 0% -98.49% (p=0.000 n=10+10)
Fixes#3876 (sadly)
Change-Id: I1195ac5de21a8a8b3cdace5871d263e81aa27e91
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
name old time/op new time/op delta
GetList-8 11.2ms ± 5% 11.1ms ± 3% ~ (p=0.661 n=10+9)
name old alloc/op new alloc/op delta
GetList-8 83.3kB ± 1% 67.4kB ± 1% -19.05% (p=0.000 n=10+10)
name old allocs/op new allocs/op delta
GetList-8 2.89k ± 2% 2.19k ± 1% -24.24% (p=0.000 n=10+10)
(real issue is we're calling this code as much as we are, but easy
enough to make it efficient because it'll still need to be called
sometimes in any case)
Updates #5958
Change-Id: I90c20278d73e80315a840aed1397d24faa308d93
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
We removed it in #4806 in favor of the built-in functionality from the
nhooyr.io/websocket package. However, it has an issue with deadlines
that has not been fixed yet (see nhooyr/websocket#350). Temporarily
go back to using a custom wrapper (using the fix from our fork) so that
derpers will stop closing connections too aggressively.
Updates #5921
Signed-off-by: Mihai Parparita <mihai@tailscale.com>
If the wgcfg.Config is specified with network logging arguments,
then Userspace.Reconfig starts up an asynchronous network logger,
which is shutdown either upon Userspace.Close or when Userspace.Reconfig
is called again without network logging or route arguments.
Signed-off-by: Joe Tsai <joetsai@digital-static.net>
When Tailscale is installed via Homebrew, `/usr/local/bin/tailscaled`
is a symlink to the actual binary.
Now when `tailscaled install-system-daemon` runs, it will not attempt
to overwrite that symlink if it already points to the tailscaled binary.
However, if executed binary and the link target differ, the path will
he overwritten - this can happen when a user decides to replace
Homebrew-installed tailscaled with a one compiled from source code.
Fixes#5353
Signed-off-by: Anton Tolchanov <anton@tailscale.com>
If Wrapper.StatisticsEnable is enabled,
then per-connection counters are maintained.
If enabled, Wrapper.StatisticsExtract must be periodically called
otherwise there is unbounded memory growth.
Signed-off-by: Joe Tsai <joetsai@digital-static.net>
Most visible when using tsnet.Server, but could have resulted in dropped
messages in a few other places too.
Fixes#5743
Signed-off-by: Mihai Parparita <mihai@tailscale.com>
I added new functions to winutil to obtain the state of a service and all
its depedencies, serialize them to JSON, and write them to a Logf.
When tstun.New returns a wrapped ERROR_DEVICE_NOT_AVAILABLE, we know that wintun
installation failed. We then log the service graph rooted at "NetSetupSvc".
We are interested in that specific service because network devices will not
install if that service is not running.
Updates https://github.com/tailscale/tailscale/issues/5531
Signed-off-by: Aaron Klotz <aaron@tailscale.com>
* and move goroutine scrubbing code to its own package for reuse
* bump capver to 45
Change-Id: I9b4dfa5af44d2ecada6cc044cd1b5674ee427575
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
Personal preference (so it's obvious it's not a bool flag), but it
also matches the --state= before it.
Bonus: stop allowing PORT to sneak in extra flags to be passed as
their own arguments, as $FOO and ${FOO} expand differently. (${FOO} is
required to concat to strings)
Change-Id: I994626a5663fe0948116b46a971e5eb2c4023216
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
ipnserver previously had support for a Windows-only environment
variable mechanism that further only worked when Windows was running
as a service, not from a console.
But we want it to work from tailscaed too, and we want it to work on
macOS and Synology. So move it to envknob, now that envknob can change
values at runtime post-init.
A future change will wire this up for more platforms, and do something
more for CLI flags like --port, which the bug was originally about.
Updates #5114
Change-Id: I9fd69a9a91bb0f308fc264d4a6c33e0cbe352d71
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
The io/ioutil package has been deprecated as of Go 1.16 [1]. This commit
replaces the existing io/ioutil functions with their new definitions in
io and os packages.
Reference: https://golang.org/doc/go1.16#ioutil
Signed-off-by: Eng Zer Jun <engzerjun@gmail.com>
As noted in #5617, our documented method of blocking log.tailscale.io
DNS no longer works due to bootstrap DNS.
Instead, provide an explicit flag (--no-logs-no-support) and/or env
variable (TS_NO_LOGS_NO_SUPPORT=true) to explicitly disable logcatcher
uploads. It also sets a bit on Hostinfo to say that the node is in that
mode so we can end any support tickets from such nodes more quickly.
This does not yet provide an easy mechanism for users on some
platforms (such as Windows, macOS, Synology) to set flags/env. On
Linux you'd used /etc/default/tailscaled typically. Making it easier
to set flags for other platforms is tracked in #5114.
Fixes#5617Fixestailscale/corp#1475
Change-Id: I72404e1789f9e56ec47f9b7021b44c025f7a373a
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
NextDNS is unique in that users create accounts and then get
user-specific DNS IPs & DoH URLs.
For DoH, the customer ID is in the URL path.
For IPv6, the IP address includes the customer ID in the lower bits.
For IPv4, there's a fragile "IP linking" mechanism to associate your
public IPv4 with an assigned NextDNS IPv4 and that tuple maps to your
customer ID.
We don't use the IP linking mechanism.
Instead, NextDNS is DoH-only. Which means using NextDNS necessarily
shunts all DNS traffic through 100.100.100.100 (programming the OS to
use 100.100.100.100 as the global resolver) because operating systems
can't usually do DoH themselves.
Once it's in Tailscale's DoH client, we then connect out to the known
NextDNS IPv4/IPv6 anycast addresses.
If the control plane sends the client a NextDNS IPv6 address, we then
map it to the corresponding NextDNS DoH with the same client ID, and
we dial that DoH server using the combination of v4/v6 anycast IPs.
Updates #2452
Change-Id: I3439d798d21d5fc9df5a2701839910f5bef85463
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
This is entirely optional (i.e. failing in this code is non-fatal) and
only enabled on Linux for now. Additionally, this new behaviour can be
disabled by setting the TS_DEBUG_DISABLE_AF_PACKET environment variable.
Updates #3824
Replaces #5474
Co-authored-by: Andrew Dunham <andrew@du.nham.ca>
Signed-off-by: David Anderson <danderson@tailscale.com>
This lets the control plane can make HTTP requests to nodes.
Then we can use this for future things rather than slapping more stuff
into MapResponse, etc.
Change-Id: Ic802078c50d33653ae1f79d1e5257e7ade4408fd
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
This works around the 2.3s delay in short name lookups when SNR is
enabled.
C:\Windows\System32\drivers\etc\hosts file. We only add known hosts that
match the search domains, and we populate the list in order of
Search Domains so that our matching algorithm mimics what Windows would
otherwise do itself if SNR was off.
Updates #1659
Signed-off-by: Maisem Ali <maisem@tailscale.com>
