We pass the file as an io.Reader to http.Post under the hood as request
body. Post, helpfully, detects that the body is an io.Closer and closes
it. So when we try to explicitly close it again, we get "file already
closed" error.
The Close there is not load-bearing, we have a defer for it anyway.
Remove the explicit close and error check.
Updates #cleanup
Signed-off-by: Andrew Lytvynov <awly@tailscale.com>
Also uses `http.HandlerFunc` to pass the handler into `csrfProtect`
so we can get rid of the extraneous `api` struct.
Updates tailscale/corp#13775
Signed-off-by: Sonia Appasamy <sonia@tailscale.com>
Fly apps all set X-Forwarded-For, which breaks debug access even
with a preshared key otherwise.
Updates tailscale/corp#3601
Signed-off-by: David Anderson <danderson@tailscale.com>
It's very common for OOM crashes on Windows to be caused by lack of page
file space (the NT kernel does not overcommit). Since Windows automatically
manages page file space by default, unless the machine is out of disk space,
this is typically caused by manual page file configurations that are too
small.
This patch obtains the current page file size, the amount of free page file
space, and also determines whether the page file is automatically or manually
managed.
Fixes#9090
Signed-off-by: Aaron Klotz <aaron@tailscale.com>
Previously, the operator would only monitor Services and create
a Tailscale StatefulSet which acted as a L3 proxy which proxied
traffic inbound to the Tailscale IP onto the services ClusterIP.
This extends that functionality to also monitor Ingress resources
where the `ingressClassName=tailscale` and similarly creates a
Tailscale StatefulSet, acting as a L7 proxy instead.
Users can override the desired hostname by setting:
```
- tls
hosts:
- "foo"
```
Hostnames specified under `rules` are ignored as we only create a single
host. This is emitted as an event for users to see.
Fixes#7895
Signed-off-by: Maisem Ali <maisem@tailscale.com>
This watches the provided path for a JSON encoded ipn.ServeConfig.
Everytime the file changes, or the nodes FQDN changes it reapplies
the ServeConfig.
At boot time, it nils out any previous ServeConfig just like tsnet does.
As the ServeConfig requires pre-existing knowledge of the nodes FQDN to do
SNI matching, it introduces a special `${TS_CERT_DOMAIN}` value in the JSON
file which is replaced with the known CertDomain before it is applied.
Updates #502
Updates #7895
Signed-off-by: Maisem Ali <maisem@tailscale.com>
Previously users would have to unexpose/expose the service in order to
change Hostname/TargetIP. This now applies those changes by causing a
StatefulSet rollout now that a61a9ab087 is in.
Updates #502
Signed-off-by: Maisem Ali <maisem@tailscale.com>
1. Add TCP port forwarding.
For example: ./sniproxy -forwards=tcp/22/github.com
will forward SSH to github.
% ssh -i ~/.ssh/id_ecdsa.pem -T git@github.com
Hi GitHubUser! You've successfully authenticated, but GitHub does not
provide shell access.
% ssh -i ~/.ssh/id_ecdsa.pem -T git@100.65.x.y
Hi GitHubUser! You've successfully authenticated, but GitHub does not
provide shell access.
2. Additionally export clientmetrics as prometheus metrics for local
scraping over the tailnet: http://sniproxy-hostname:8080/debug/varz
Updates https://github.com/tailscale/tailscale/issues/1748
Signed-off-by: Denton Gentry <dgentry@tailscale.com>
I'm not saying it works, but it compiles.
Updates #5794
Change-Id: I2f3c99732e67fe57a05edb25b758d083417f083e
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
Adding a root key that signs the current signing key on
pkgs.tailscale.com. This key is here purely for development and should
be replaced before 1.50 release.
Updates #8760
Signed-off-by: Andrew Lytvynov <awly@tailscale.com>
Adds a cached self node to the web client Server struct, which will
be used from the web client api to verify that request came from the
node's own machine (i.e. came from the web client frontend). We'll
be using when we switch the web client api over to acting as a proxy
to the localapi, to protect against DNS rebinding attacks.
Updates tailscale/corp#13775
Signed-off-by: Sonia Appasamy <sonia@tailscale.com>
Add `dist.Signer` hook which can arbitrarily sign linux/synology
artifacts. Plumb it through in `cmd/dist` and remove existing tarball
signing key. Distsign signing will happen on a remote machine, not using
a local key.
Updates #755
Updates #8760
Signed-off-by: Andrew Lytvynov <awly@tailscale.com>
Synology and QNAP both run the web client as a CGI script. The old web
client didn't care too much about requests paths, since there was only a
single GET and POST handler. The new client serves assets on different
paths, so now we need to care.
First, enforce that the CGI script is always accessed from its full
path, including a trailing slash (e.g. /cgi-bin/tailscale/index.cgi/).
Then, strip that prefix off before passing the request along to the main
serve handler. This allows for properly serving both static files and
the API handler in a CGI environment. Also add a CGIPath option to allow
other CGI environments to specify a custom path.
Finally, update vite and one "api/data" call to no longer assume that we
are always serving at the root path of "/".
Updates tailscale/corp#13775
Signed-off-by: Will Norris <will@tailscale.com>
Add separate server methods for synology and qnap, and enforce
authentication and authorization checks before calling into the actual
serving handlers. This allows us to remove all of the auth logic from
those handlers, since all requests will already be authenticated by that
point.
Also simplify the Synology token redirect handler by using fetch.
Remove the SynologyUser from nodeData, since it was never used in the
frontend anyway.
Updates tailscale/corp#13775
Signed-off-by: Will Norris <will@tailscale.com>
This commit doesn't change any of the logic, but just organizes the code
a little to prepare for future changes.
Updates tailscale/corp#13775
Signed-off-by: Will Norris <will@tailscale.com>
Previously we would not reapply changes to TS_HOSTNAME etc when
then the container restarted and TS_AUTH_ONCE was enabled.
This splits those into two steps login and set, allowing us to
only rerun the set step on restarts.
Updates #502
Signed-off-by: Maisem Ali <maisem@tailscale.com>
Now we have all the commands to generate the key hierarchy and verify
that signing keys were signed correctly:
```
$ ./tool/go run ./cmd/dist gen-key --priv-path root-priv.pem --pub-path root-pub.pem --root
wrote private key to root-priv.pem
wrote public key to root-pub.pem
$ ./tool/go run ./cmd/dist gen-key --priv-path signing-priv.pem --pub-path signing-pub.pem --signing
wrote private key to signing-priv.pem
wrote public key to signing-pub.pem
$ ./tool/go run ./cmd/dist sign-key --root-priv-path root-priv.pem --sign-pub-path signing-pub.pem
wrote signature to signature.bin
$ ./tool/go run ./cmd/dist verify-key-signature --root-pub-path root-pub.pem --sign-pub-path signing-pub.pem --sig-path signature.bin
signature ok
```
Updates #8760
Signed-off-by: Andrew Lytvynov <awly@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>
Refresh node data when user switches to the web client browser tab.
This helps clean up the auth flow where they're sent to another tab
to authenticate then return to the original tab, where the data
should be refreshed to pick up the login updates.
Updates tailscale/corp#13775
Signed-off-by: Sonia Appasamy <sonia@tailscale.com>
Indicate to the web client when it is running in CGI mode, and if it is
then cache the csrf key between requests.
Updates tailscale/corp#13775
Signed-off-by: Will Norris <will@tailscale.com>
This adds a workflow_dispatch input to the update-flakehub workflow that
allows the user to specify an existing tag to publish to FlakeHub. This
is useful for publishing a version of a package that has already been
tagged in the repository.
Updates #9008
Signed-off-by: Shayne Sweeney <shayne@tailscale.com>
Can write "wasm" instead of js || wasi1p, since there's only two:
$ go tool dist list | grep wasm
js/wasm
wasip1/wasm
Plus, if GOOS=wasip2 is added later, we're already set.
Updates #5794
Change-Id: Ifcfb187c3775c17c9141bc721512dc4577ac4434
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
Some OS-specific funcs were defined in init. Another used build tags
and required all other OSes to stub it out. Another one could just be in
the portable file.
Simplify it a bit, removing a file and some stubs in the process.
Updates #5794
Change-Id: I51df8772cc60a9335ac4c1dc0ab59b8a0d236961
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
We have a fancy package for doing TLS cert validation even if the machine
doesn't have TLS certs (for LetsEncrypt only) but the CLI's netcheck command
wasn't using it.
Also, update the tlsdial's outdated package docs while here.
Updates #cleanup
Change-Id: I74b3cb645d07af4d8ae230fb39a60c809ec129ad
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
This simplifies some netmon code in prep for other changes.
It breaks up Monitor.debounce into a helper method so locking is
easier to read and things unindent, and then it simplifies the polling
netmon implementation to remove the redundant stuff that the caller
(the Monitor.debounce loop) was already basically doing.
Updates #9040
Change-Id: Idcfb45201d00ae64017042a7bdee6ef86ad37a9f
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
Under normal circumstances, you would typically want to keep the default
behavior of requiring secure cookies. In the case of the Tailscale web
client, we are regularly serving on localhost (where secure cookies
don't really matter), and/or we are behind a reverse proxy running on a
network appliance like a NAS or Home Assistant. In those cases, those
devices are regularly accessed over local IP addresses without https
configured, so would not work with secure cookies.
Updates tailscale/corp#13775
Signed-off-by: Will Norris <will@tailscale.com>
To make key management less error-prone, use different PEM block types
for root and signing keys. As a result, separate out most of the Go code
between root/signing keys too.
Updates #8760
Signed-off-by: Andrew Lytvynov <awly@tailscale.com>
This PR removes calls to ioutil library and replaces them
with their new locations in the io and os packages.
Fixes#9034
Updates #5210
Signed-off-by: Marwan Sulaiman <marwan@tailscale.com>
Open control server auth URLs in new browser tabs on web clients
so users don't loose original client URL when redirected for login.
Updates tailscale/corp#13775
Signed-off-by: Sonia Appasamy <sonia@tailscale.com>
Actually fixed in 77ff705545 but that was cherry-picked to a branch
and we don't bump capver in branches.
This tells the control plane that UPnP should be re-enabled going
forward.
Updates #8992
Change-Id: I5c4743eb52fdee94175668c368c0f712536dc26b
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
Upcoming work on incremental netmap change handling will require some
replumbing of which subsystems get notified about what. Done naively,
it could break "tailscale status --json" visibility later. To make sure
I understood the flow of all the updates I was rereading the status code
and realized parts of ipnstate.Status were being populated by the wrong
subsystems.
The engine (wireguard) and magicsock (data plane, NAT traveral) should
only populate the stuff that they uniquely know. The WireGuard bits
were fine but magicsock was populating stuff stuff that LocalBackend
could've better handled, so move it there.
Updates #1909
Change-Id: I6d1b95d19a2d1b70fbb3c875fac8ea1e169e8cb0
Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>
I forgot to move the defer out of the func, so the tsnet.Server
immediately closed after starting.
Updates #502
Signed-off-by: Maisem Ali <maisem@tailscale.com>
This ensures that `go mod vendor` includes these files, which are needed
for client builds run in corp.
Updates tailscale/corp#13775
Signed-off-by: Will Norris <will@tailscale.com>
It was jumbled doing a lot of things, this breaks it up into
the svc reconciliation and the tailscale sts reconciliation.
Prep for future commit.
Updates #502
Signed-off-by: Maisem Ali <maisem@tailscale.com>
I thought this had something to do with Synology or QNAP support, since
they both have specific authentication logic. But it turns out this was
part of the original web client added in #1621, and then refactored as
part of #2093. But with how we handle logging in now, it's never
called.
Updates tailscale/corp#13775
Signed-off-by: Will Norris <will@tailscale.com>
Add a new subcommand to generate a Ed25519 key pair for release signing.
The same command can be used to generate both root and signing keys.
Updates #8760
Signed-off-by: Andrew Lytvynov <awly@tailscale.com>
If we don't have the ICMP hint available, such as on Android, we can use
the signal of rx traffic to bias toward a particular endpoint.
We don't want to stick to a particular endpoint for a very long time
without any signals, so the sticky time is reduced to 1 second, which is
large enough to avoid excessive packet reordering in the common case,
but should be small enough that either rx provides a strong signal, or
we rotate in a user-interactive schedule to another endpoint, improving
the feel of failover to other endpoints.
Updates #8999
Co-authored-by: Charlotte Brandhorst-Satzkorn <charlotte@tailscale.com>
Signed-off-by: James Tucker <james@tailscale.com>
Signed-off-by: Charlotte Brandhorst-Satzkorn <charlotte@tailscale.com>
There are cases where we do not detect the non-viability of a route, but
we will instead observe a failure to send. In a Disco path this would
normally be handled as a side effect of Disco, which is not available to
non-Disco WireGuard nodes. In both cases, recognizing the failure as
such will result in faster convergence.
Updates #8999
Signed-off-by: James Tucker <james@tailscale.com>
LastFullPing is now used for disco or wireguard only endpoints. This
change updates the comment to make that clear.
Updates #7826
Signed-off-by: Charlotte Brandhorst-Satzkorn <charlotte@tailscale.com>
In order for the installer to restart the GUI correctly post-upgrade, we
need the GUI to be able to register its restart preferences.
This PR adds API support for doing so. I'm adding it to OSS so that it
is available should we need to do any such registrations on OSS binaries
in the future.
Updates https://github.com/tailscale/corp/issues/13998
Signed-off-by: Aaron Klotz <aaron@tailscale.com>
There are latency values stored in bestAddr and endpointState that are
no longer applicable after a connectivity change and should be cleared
out, following the documented behavior of the function.
Updates #8999
Signed-off-by: James Tucker <james@tailscale.com>
