267 Commits (f572286bf92c72c62776b26d69a24a8775ff2d39)

Author	SHA1	Message	Date
Brad Fitzpatrick	aa15a63651	derp: add new concurrent server benchmark ... In prep for reducing mutex contention on Server.mu. Updates #3560 Change-Id: Ie95e7c6dc9f4b64b6f79b3b2338f8cd86c688d98 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	2 months ago
Brad Fitzpatrick	cec779e771	util/slicesx: add FirstElementEqual and LastElementEqual ... And update a few callers as examples of motivation. (there are a couple others, but these are the ones where it's prettier) Updates #cleanup Change-Id: Ic8c5cb7af0a59c6e790a599136b591ebe16d38eb Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	2 months ago
Brad Fitzpatrick	910462a8e0	derp: unify server's clientSet interface into concrete type ... 73280595a8 for #2751 added a "clientSet" interface to distinguish the two cases of a client being singly connected (the common case) vs tolerating multiple connections from the client at once. At the time (three years ago) it was kinda an experiment and we didn't know whether it'd stop the reconnect floods we saw from certain clients. It did. So this promotes it to a be first-class thing a bit, removing the interface. The old tests from 73280595a were invaluable in ensuring correctness while writing this change (they failed a bunch). But the real motivation for this change is that it'll permit a future optimization to add flow tracking for stats & performance where we don't contend on Server.mu for each packet sent via DERP. Instead, each client can track its active flows and hold on to a *clientSet and ask the clientSet per packet what the active client is via one atomic load rather than a mutex. And if the atomic load returns nil, we'll know we need to ask the server to see if they died and reconnected and got a new clientSet. But that's all coming later. Updates #3560 Change-Id: I9ccda3e5381226563b5ec171ceeacf5c210e1faf Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	2 months ago
Kyle Carberry	6c852fa817	go.{mod,sum}: migrate from nhooyr.io/websocket to github.com/coder/websocket ... Coder has just adopted nhooyr/websocket which unfortunately changes the import path. `github.com/coder/coder` imports `tailscale.com/net/wsconn` which was still pointing to `nhooyr.io/websocket`, but this change updates it. See https://coder.com/blog/websocket Updates #13154 Change-Id: I3dec6512472b14eae337ae22c5bcc1e3758888d5 Signed-off-by: Kyle Carberry <kyle@carberry.com>	3 months ago
Brad Fitzpatrick	6ca078c46e	cmd/derper: move 204 handler from package main to derphttp ... Updates #13038 Change-Id: I28a8284dbe49371cae0e9098205c7c5f17225b40 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	4 months ago
Brad Fitzpatrick	f77821fd63	derp/derphttp: determine whether a region connect was to non-ideal node ... ... and then do approximately nothing with that information, other than a big TODO. This is mostly me relearning this code and leaving breadcrumbs for others in the future. Updates #12724 Signed-off-by: Brad Fitzpatrick <brad@danga.com>	4 months ago
Brad Fitzpatrick	c6af5bbfe8	all: add test for package comments, fix, add comments as needed ... Updates #cleanup Change-Id: Ic4304e909d2131a95a38b26911f49e7b1729aaef Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	5 months ago
James Tucker	114d1caf55	derp/xdp: retain the link so that the fd is not closed ... BPF links require that the owning FD remains open, this FD is embedded into the RawLink returned by the attach function and must live for the duration of the server. Updates ENG-4274 Signed-off-by: James Tucker <james@tailscale.com>	5 months ago
Brad Fitzpatrick	210264f942	cmd/derper: clarify that derper and tailscaled need to be in sync ... Fixes #12617 Change-Id: Ifc87b7d9cf699635087afb57febd01fb9a6d11b7 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	5 months ago
Brad Fitzpatrick	3485e4bf5a	derp: make RunConnectionLoop funcs take Messages, support PeerPresentFlags ... PeerPresentFlags was added in 5ffb2668ef but wasn't plumbed through to the RunConnectionLoop. Rather than add yet another parameter (as IP:port was added earlier), pass in the raw PeerPresentMessage and PeerGoneMessage struct values, which are the same things, plus two fields: PeerGoneReasonType for gone and the PeerPresentFlags from 5ffb2668ef. Updates tailscale/corp#17816 Change-Id: Ib19d9f95353651ada90656071fc3656cf58b7987 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	5 months ago
Brad Fitzpatrick	d91e5c25ce	derp: redo, simplify how mesh update writes are queued/written ... I couldn't convince myself the old way was safe and couldn't lose writes. And it seemed too complicated. Updates tailscale/corp#21104 Change-Id: I17ba7c7d6fd83458a311ac671146a1f6a458a5c1 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	5 months ago
Brad Fitzpatrick	ded7734c36	derp: account for increased size of peerPresent messages in mesh updates ... sendMeshUpdates tries to write as much as possible without blocking, being careful to check the bufio.Writer.Available size before writes. Except that regressed in 6c791f7d60 which made those messages larger, which meants we were doing network I/O with the Server mutex held. Updates tailscale/corp#13945 Change-Id: Ic327071d2e37de262931b9b390cae32084811919 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	5 months ago
Brad Fitzpatrick	5ffb2668ef	derp: add PeerPresentFlags bitmask to Watch messages ... Updates tailscale/corp#17816 Change-Id: Ib5baf6c981a6a4c279f8bbfef02048cfbfb3323b Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	5 months ago
Brad Fitzpatrick	25eeafde23	derp: don't verify mesh peers when --verify-clients is set ... Updates tailscale/corp#20654 Change-Id: I33c7ca3c7a3c4e492797b73c66eefb699376402c Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	5 months ago
Brad Fitzpatrick	4b39b6f7ce	derp: fix fmt verb for nodekeys ... It was hex-ifying the String() form of key.NodePublic, which was already hex. I noticed in some logs: "client 6e6f64656b65793a353537353..." And thought that 6x6x6x6x looked strange. It's "nodekey:" in hex. Updates tailscale/corp#20844 Change-Id: Ib9f2d63b37e324420b86efaa680668a9b807e465 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	5 months ago
Jordan Whited	a93173b56a	cmd/xdpderper,derp/xdp: implement mode that drops STUN packets (#12527) ... This is useful during maintenance as a method for shedding home client load. Updates tailscale/corp#20689 Signed-off-by: Jordan Whited <jordan@tailscale.com>	5 months ago
James Tucker	87c5ad4c2c	derp: add a verifyClients check to the consistency check ... Only implemented for the local tailscaled variant for now. Updates tailscale/corp#20844 Signed-off-by: James Tucker <james@tailscale.com>	5 months ago
Jordan Whited	315f3d5df1	derp/xdp: fix handling of zero value UDP checksums (#12510) ... validate_udp_checksum was previously indeterminate (not zero) at declaration, and IPv4 zero value UDP checksum packets were being passed to the kernel. Updates tailscale/corp#20689 Signed-off-by: Jordan Whited <jordan@tailscale.com>	5 months ago
Jordan Whited	65888d95c9	derp/xdp,cmd/xdpderper: initial skeleton (#12390) ... This commit introduces a userspace program for managing an experimental eBPF XDP STUN server program. derp/xdp contains the eBPF pseudo-C along with a Go pkg for loading it and exporting its metrics. cmd/xdpderper is a package main user of derp/xdp. Updates tailscale/corp#20689 Signed-off-by: Jordan Whited <jordan@tailscale.com>	5 months ago
Brad Fitzpatrick	6908fb0de3	ipn/localapi,client/tailscale,cmd/derper: add WhoIs lookup by nodekey, use in derper ... Fixes #12465 Change-Id: I9b7c87315a3d2b2ecae2b8db9e94b4f5a1eef74a Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	5 months ago
Maisem Ali	4a8cb1d9f3	all: use math/rand/v2 more ... Updates #11058 Signed-off-by: Maisem Ali <maisem@tailscale.com>	6 months ago
Brad Fitzpatrick	f227083539	derp: add some guardrails for derpReason metrics getting out of sync ... The derp metrics got out of sync in 74eb99aed1 (2023-03). They were fixed in 0380cbc90d (2024-05). This adds some further guardrails (atop the previous fix) to make sure they don't get out of sync again. Updates #12288 Change-Id: I809061a81f8ff92f45054d0253bc13871fc71634 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	6 months ago
Spike Curtis	0380cbc90d	derp: fix dropReason metrics labels (#12288) ... Updates #2745 Updates #7552 Signed-off-by: Spike Curtis <spike@coder.com>	6 months ago
Andrew Dunham	c6d42b1093	derp: remove stats goroutine, use a timer ... Without changing behaviour, don't create a goroutine per connection that sits and sleeps, but rather use a timer that wakes up and gathers statistics on a regular basis. Fixes #12127 Signed-off-by: Andrew Dunham <andrew@du.nham.ca> Change-Id: Ibc486447e403070bdc3c2cd8ae340e7d02854f21	6 months ago
Brad Fitzpatrick	15fc6cd966	derp/derphttp: fix netcheck HTTPS probes ... The netcheck client, when no UDP is available, probes distance using HTTPS. Several problems: * It probes using /derp/latency-check. * But cmd/derper serves the handler at /derp/probe * Despite the difference, it work by accident until c8f4dfc8c0 which made netcheck's probe require a 2xx status code. * in tests, we only use derphttp.Handler, so the cmd/derper-installed mux routes aren't preesnt, so there's no probe. That breaks tests in airplane mode. netcheck.Client then reports "unexpected HTTP status 426" (Upgrade Required) This makes derp handle both /derp/probe and /derp/latency-check equivalently, and in both cmd/derper and derphttp.Handler standalone modes. I notice this when wgengine/magicsock TestActiveDiscovery was failing in airplane mode (no wifi). It still doesn't pass, but it gets further. Fixes #11989 Change-Id: I45213d4bd137e0f29aac8bd4a9ac92091065113f	7 months ago
Brad Fitzpatrick	ee3bd4dbda	derp/derphttp, net/netcheck: plumb netmon.Monitor to derp netcheck client ... Fixes #11981 Change-Id: I0e15a09f93aefb3cfddbc12d463c1c08b83e09fd Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	7 months ago
Maisem Ali	a49ed2e145	derp,ipn/ipnlocal: stop calling rand.Seed ... It's deprecated and using it gets us the old slow behavior according to https://go.dev/blog/randv2. > Having eliminated repeatability of the global output stream, Go 1.20 > was also able to make the global generator scale better in programs > that don’t call rand.Seed, replacing the Go 1 generator with a very > cheap per-thread wyrand generator already used inside the Go > runtime. This removed the global mutex and made the top-level > functions scale much better. Programs that do call rand.Seed fall > back to the mutex-protected Go 1 generator. Updates #7123 Change-Id: Ia5452e66bd16b5457d4b1c290a59294545e13291 Signed-off-by: Maisem Ali <maisem@tailscale.com>	7 months ago
Brad Fitzpatrick	3672f29a4e	net/netns, net/dns/resolver, etc: make netmon required in most places ... The goal is to move more network state accessors to netmon.Monitor where they can be cheaper/cached. But first (this change and others) we need to make sure the one netmon.Monitor is plumbed everywhere. Some notable bits: * tsdial.NewDialer is added, taking a now-required netmon * because a tsdial.Dialer always has a netmon, anything taking both a Dialer and a NetMon is now redundant; take only the Dialer and get the NetMon from that if/when needed. * netmon.NewStatic is added, primarily for tests Updates tailscale/corp#10910 Updates tailscale/corp#18960 Updates #7967 Updates #3299 Change-Id: I877f9cb87618c4eb037cee098241d18da9c01691 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	7 months ago
Brad Fitzpatrick	723c775dbb	tsd, ipnlocal, etc: add tsd.System.HealthTracker, start some plumbing ... This adds a health.Tracker to tsd.System, accessible via a new tsd.System.HealthTracker method. In the future, that new method will return a tsd.System-specific HealthTracker, so multiple tsnet.Servers in the same process are isolated. For now, though, it just always returns the temporary health.Global value. That permits incremental plumbing over a number of changes. When the second to last health.Global reference is gone, then the tsd.System.HealthTracker implementation can return a private Tracker. The primary plumbing this does is adding it to LocalBackend and its dozen and change health calls. A few misc other callers are also plumbed. Subsequent changes will flesh out other parts of the tree (magicsock, controlclient, etc). Updates #11874 Updates #4136 Change-Id: Id51e73cfc8a39110425b6dc19d18b3975eac75ce Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	7 months ago
Brad Fitzpatrick	7c1d6e35a5	all: use Go 1.22 range-over-int ... Updates #11058 Change-Id: I35e7ef9b90e83cac04ca93fd964ad00ed5b48430 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	7 months ago
Brad Fitzpatrick	b27238b654	derp/derphttp: don't block in LocalAddr method ... The derphttp.Client mutex is held during connects (for up to 10 seconds) so this LocalAddr method (blocking on said mutex) could also block for up to 10 seconds, causing a pileup upstream in magicsock/wgengine and ultimately a watchdog timeout resulting in a crash. Updates #11519 Change-Id: Idd1d94ee00966be1b901f6899d8b9492f18add0f Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	8 months ago
Brad Fitzpatrick	3e6306a782	derp/derphttp: make CONNECT Host match request-target's authority-form ... This CONNECT client doesn't match what Go's net/http.Transport does (making the two values match). This makes it match. This is all pretty unspecified but most clients & doc examples show these matching. And some proxy implementations (such as Zscaler) care. Updates tailscale/corp#18716 Change-Id: I135c5facbbcec9276faa772facbde1bb0feb2d26 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	8 months ago
Brad Fitzpatrick	10d130b845	cmd/derper, derp, tailcfg: add admission controller URL option ... So derpers can check an external URL for whether to permit access to a certain public key. Updates tailscale/corp#17693 Change-Id: I8594de58f54a08be3e2dbef8bcd1ff9b728ab297 Co-authored-by: Maisem Ali <maisem@tailscale.com> Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	9 months ago
Brad Fitzpatrick	2988c1ec52	derp: plumb context to Server.verifyClient ... Updates tailscale/corp#17693 Change-Id: If17e02c77d5ad86b820e639176da2d3e61296bae Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	9 months ago
Brad Fitzpatrick	2bd3c1474b	util/cmpx: delete now that we're using Go 1.22 ... Updates #11058 Change-Id: I09dea8e86f03ec148b715efca339eab8b1f0f644 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	10 months ago
James Tucker	7100b6e721	derp: optimize another per client field alignment ... Updates #self Signed-off-by: James Tucker <james@tailscale.com>	11 months ago
James Tucker	ee20327496	derp: remove unused per-client struct field ... Updates #self Signed-off-by: James Tucker <james@tailscale.com>	11 months ago
James Tucker	a7f65b40c5	derp: optimize field order to reduce GC cost ... See the field alignment lints for more information. Reductions are 64->24 and 64->32 respectively. Updates #self Signed-off-by: James Tucker <james@tailscale.com>	11 months ago
James Tucker	ff9c1ebb4a	derp: reduce excess goroutines blocking on broadcasts ... Observed on one busy derp node, there were 600 goroutines blocked writing to this channel, which represents not only more blocked routines than we need, but also excess wake-ups downstream as the latent goroutines writes represent no new work. Updates #self Signed-off-by: James Tucker <james@tailscale.com>	11 months ago
Andrew Lytvynov	2716250ee8	all: cleanup unused code, part 2 (#10670) ... And enable U1000 check in staticcheck. Updates #cleanup Signed-off-by: Andrew Lytvynov <awly@tailscale.com>	11 months ago
Anton Tolchanov	ef6a6e94f1	derp/derphttp: use a getter method to read server key ... To hold the mutex while accessing it. Fixes #10122 Signed-off-by: Anton Tolchanov <anton@tailscale.com>	1 year ago
Brad Fitzpatrick	b2b836214c	derp/derphttp: fix derptrack fix ... 3d7fb6c21d dropped the explicit called to (*Client).connect when its (*Client).WatchConnectionChanges got removed+refactored. This puts it back, but in RunWatchConnectionLoop, before the call to the (*Client).ServerPublicKey accessor, which is documented to return the zero value (which is what broke us) on an unconnected connection. Plus some tests. Fixes tailscale/corp#15604 Change-Id: I0f242816f5ee4ad3bb0bf0400abc961dbe9f5fc8 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Anton Tolchanov	c3313133b9	derp/derphttp: close DERP client to avoid data race in test ... Fixes #10041 Signed-off-by: Anton Tolchanov <anton@tailscale.com>	1 year ago
Anton Tolchanov	3114a1c88d	derp/derphttp: add watch reconnection tests from #9719 ... Co-authored-by: Val <valerie@tailscale.com> Signed-off-by: Anton Tolchanov <anton@tailscale.com>	1 year ago
Brad Fitzpatrick	3d7fb6c21d	derp/derphttp: fix race in mesh watcher ... The derphttp client automatically reconnects upon failure. RunWatchConnectionLoop called derphttp.Client.WatchConnectionChanges once, but that wrapper method called the underlying derp.Client.WatchConnectionChanges exactly once on derphttp.Client's currently active connection. If there's a failure, we need to re-subscribe upon all reconnections. This removes the derphttp.Client.WatchConnectionChanges method, which was basically impossible to use correctly, and changes it to be a boolean field on derphttp.Client alongside MeshKey and IsProber. Then it moves the call to the underlying derp.Client.WatchConnectionChanges to derphttp's client connection code, so it's resubscribed on any reconnect. Some paranoia is then added to make sure people hold the API right, not calling derphttp.Client.RunWatchConnectionLoop on an already-started Client without having set the bool to true. (But still auto-setting it to true if that's the first method that's been called on that derphttp.Client, as is commonly the case, and prevents existing code from breaking) Fixes tailscale/corp#9916 Supercedes tailscale/tailscale#9719 Co-authored-by: Val <valerie@tailscale.com> Co-authored-by: Irbe Krumina <irbe@tailscale.com> Co-authored-by: Anton Tolchanov <anton@tailscale.com> Signed-off-by: Brad Fitzpatrick <brad@danga.com>	1 year ago
Thomas Kosiewski	b2ae8fdf80	derp/derphttp: strip port numbers from URL hostname ... When trying to set up multiple derper instances meshing with each other, it turned out that while one can specify an alternative listening port using the -a flag, the TLS hostname gets incorrectly determined and includes the set alternative listening port as part of the hostname. Thus, the TLS hostname validation always fails when the -mesh-with values have ports. Updates #9949 Signed-off-by: Thomas Kosiewski <thomas.kosiewski@loft.sh>	1 year ago
David Anderson	8b492b4121	net/wsconn: accept a remote addr string and plumb it through ... This makes wsconn.Conns somewhat present reasonably when they are the client of an http.Request, rather than just put a placeholder in that field. Updates tailscale/corp#13777 Signed-off-by: David Anderson <danderson@tailscale.com>	1 year ago
Brad Fitzpatrick	3bce9632d9	derp/derphttp: fix data race and crash in proxy dial error path ... Named result meant error paths assigned that variable to nil. But a goroutine was concurrently using that variable. Don't use a named result for that first parameter. Then then return paths don't overwrite it. Fixes #9129 Change-Id: Ie57f99d40ca8110085097780686d9bd620aaf160 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	6c791f7d60	derp: include src IPs in mesh watch messages ... Updates tailscale/corp#13945 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	be4eb6a39e	derp, net/dns/recursive: use Go 1.21 min ... Updates #8419 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago