509 Commits (de2af54ffcf22e995c9e09f32a7ef14c6906a5ac)

Author	SHA1	Message	Date
Brad Fitzpatrick	44c6909c92	control/controlclient: move watchdog out of mapSession ... In prep for making mapSession's lifetime not be 1:1 with a single HTTP response's lifetime, this moves the inactivity timer watchdog out of mapSession and into the caller that owns the streaming HTTP response. (This is admittedly closer to how it was prior to the mapSession type existing, but that was before we connected some dots which were impossible to even see before the mapSession type broke the code up.) Updates #7175 Change-Id: Ia108dac84a4953db41cbd30e73b1de4a2a676c11 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	c87d58063a	control/controlclient: move lastPrintMap field from Direct to mapSession ... It was a really a mutable field owned by mapSession that we didn't move in earlier commits. Once moved, it's then possible to de-func-ify the code and turn it into a regular method rather than an installed optional hook. Noticed while working to move map session lifetimes out of Direct.sendMapRequest's single-HTTP-connection scope. Updates #7175 Updates #cleanup Change-Id: I6446b15793953d88d1cabf94b5943bb3ccac3ad9 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Adrian Dewhurst	5347e6a292	control/controlclient: support certstore without cgo ... We no longer build Windows releases with cgo enabled, which automatically turned off certstore support. Rather than re-enabling cgo, we updated our fork of the certstore package to no longer require cgo. This updates the package, cleans up how the feature is configured, and removes the cgo build tag requirement. Fixes tailscale/corp#14797 Fixes tailscale/coral#118 Change-Id: Iaea34340761c0431d759370532c16a48c0913374 Signed-off-by: Adrian Dewhurst <adrian@tailscale.com>	1 year ago
Claire Wang	754fb9a8a8	tailcfg: add tailnet field to register request (#9675) ... Updates tailscale/corp#10967 Signed-off-by: Claire Wang <claire@tailscale.com>	1 year ago
Maisem Ali	3655fb3ba0	control/controlclient: fix deadlock in shutdown ... Fixes a deadlock observed in a different repo. Regressed in 5b3f5eabb5. Updates tailscale/corp#14950 Updates tailscale/corp#14515 Updates tailscale/corp#14139 Updates tailscale/corp#13175 Signed-off-by: Maisem Ali <maisem@tailscale.com>	1 year ago
Brad Fitzpatrick	425cf9aa9d	tailcfg, all: use []netip.AddrPort instead of []string for Endpoints ... It's JSON wire compatible. Updates #cleanup Change-Id: Ifa5c17768fec35b305b06d75eb5f0611c8a135a6 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	246e0ccdca	tsnet: add a test for restarting a tsnet server, fix Windows ... Thanks to @qur and @eric for debugging! Fixes #6973 Change-Id: Ib2cf8f030cf595cc73dd061c72e78ac19f5fae5d Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	5b3f5eabb5	control/controlclient: fix leaked http2 goroutines on shutdown ... If a noise dial was happening concurrently with shutdown, the http2 goroutines could leak. Updates tailscale/corp#14950 Updates tailscale/corp#14515 Updates tailscale/corp#14139 Updates tailscale/corp#13175 Signed-off-by: Brad Fitzpatrick <brad@danga.com>	1 year ago
Claire Wang	e3d6236606	winutil: refactor methods to get values from registry to also return (#9536) ... errors Updates tailscale/corp#14879 Signed-off-by: Claire Wang <claire@tailscale.com>	1 year ago
Andrew Dunham	530aaa52f1	net/dns: retry forwarder requests over TCP ... We weren't correctly retrying truncated requests to an upstream DNS server with TCP. Instead, we'd return a truncated request to the user, even if the user was querying us over TCP and thus able to handle a large response. Also, add an envknob and controlknob to allow users/us to disable this behaviour if it turns out to be buggy (✨ DNS ✨). Updates #9264 Signed-off-by: Andrew Dunham <andrew@du.nham.ca> Change-Id: Ifb04b563839a9614c0ba03e9c564e8924c1a2bfd	1 year ago
Brad Fitzpatrick	3b32d6c679	wgengine/magicsock, controlclient, net/dns: reduce some logspam ... Updates #cleanup Change-Id: I78b0697a01e94baa33f3de474b591e616fa5e6af Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Val	65dc711c76	control,tailcfg,wgengine/magicsock: add nodeAttr to enable/disable peer MTU ... Add a nodeAttr to enable/disable peer path MTU discovery. Updates #311 Signed-off-by: Val <valerie@tailscale.com>	1 year ago
Tom DNetto	c08cf2a9c6	all: declare & plumb IPv6 masquerade address for peer ... This PR plumbs through awareness of an IPv6 SNAT/masquerade address from the wire protocol through to the low-level (tstun / wgengine). This PR is the first in two PRs for implementing IPv6 NAT support to/from peers. A subsequent PR will implement the data-plane changes to implement IPv6 NAT - this is just plumbing. Signed-off-by: Tom DNetto <tom@tailscale.com> Updates ENG-991	1 year ago
Maisem Ali	19a9d9037f	tailcfg: add NodeCapMap ... Like PeerCapMap, add a field to `tailcfg.Node` which provides a map of Capability to raw JSON messages which are deferred to be parsed later by the application code which cares about the specific capabilities. This effectively allows us to prototype new behavior without having to commit to a schema in tailcfg, and it also opens up the possibilities to develop custom behavior in tsnet applications w/o having to plumb through application specific data in the MapResponse. Updates #4217 Signed-off-by: Maisem Ali <maisem@tailscale.com>	1 year ago
Brad Fitzpatrick	0d991249e1	types/netmap: remove NetworkMap.{Addresses,MachineStatus} ... And convert all callers over to the methods that check SelfNode. Now we don't have multiple ways to express things in tests (setting fields on SelfNode vs NetworkMap, sometimes inconsistently) and don't have multiple ways to check those two fields (often only checking one or the other). Updates #9443 Change-Id: I2d7ba1cf6556142d219fae2be6f484f528756e3c Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Maisem Ali	a61caea911	tailcfg: define a type for NodeCapability ... Instead of untyped string, add a type to identify these. Updates #cleanup Signed-off-by: Maisem Ali <maisem@tailscale.com>	1 year ago
James Tucker	e7727db553	tailcfg: add DNS address list for IsWireGuardOnly nodes ... Tailscale exit nodes provide DNS service over the peer API, however IsWireGuardOnly nodes do not have a peer API, and instead need client DNS parameters passed in their node description. For Mullvad nodes this will contain the in network 10.64.0.1 address. Updates #9377 Signed-off-by: James Tucker <james@tailscale.com>	1 year ago
Brad Fitzpatrick	9203916a4a	control/controlknobs: move more controlknobs code from controlclient ... Updates #cleanup Change-Id: I2b8b6ac97589270f307bfb20e33674894ce873b5 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	3af051ea27	control/controlclient, types/netmap: start plumbing delta netmap updates ... Currently only the top four most popular changes: endpoints, DERP home, online, and LastSeen. Updates #1909 Change-Id: I03152da176b2b95232b56acabfb55dcdfaa16b79 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	7c1ed38ab3	ipn/ipnlocal: fix missing controlknobs.Knobs plumbing ... I missed connecting some controlknobs.Knobs pieces in 4e91cf20a8 resulting in that breaking control knobs entirely. Whoops. The fix in ipn/ipnlocal (where it makes a new controlclient) but to atone, I also added integration tests. Those integration tests use a new "tailscale debug control-knobs" which by itself might be useful for future debugging. Updates #9351 Change-Id: Id9c89c8637746d879d5da67b9ac4e0d2367a3f0d Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	42072683d6	control/controlknobs: move ForceBackgroundSTUN to controlknobs.Knobs ... This is both more efficient (because the knobs' bool is only updated whenever Node is changed, rarely) and also gets us one step closer to removing a case of storing a netmap.NetworkMap in magicsock. (eventually we want to phase out much of the use of that type internally) Updates #1909 Change-Id: I37e81789f94133175064fdc09984e4f3a431f1a1 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	4e91cf20a8	control/controlknobs, all: add plumbed Knobs type, not global variables ... 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>	1 year ago
Brad Fitzpatrick	9a86aa5732	all: depend on zstd unconditionally, remove plumbing to make it optional ... 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>	1 year ago
Maisem Ali	f6a203fe23	control/controlclient: check c.closed in waitUnpause ... We would only check if the client was paused, but not if the client was closed. This meant that a call to Shutdown may block forever/leak goroutines Updates #cleanup Signed-off-by: Maisem Ali <maisem@tailscale.com>	1 year ago
Maisem Ali	d06a75dcd0	ipn/ipnlocal: fix deadlock in resetControlClientLocked ... resetControlClientLocked is called while b.mu was held and would call cc.Shutdown which would wait for the observer queue to drain. However, there may be active callbacks from cc already waiting for b.mu resulting in a deadlock. This makes it so that resetControlClientLocked does not call Shutdown, and instead just returns the value. It also makes it so that any status received from previous cc are ignored. Updates tailscale/corp#12827 Signed-off-by: Maisem Ali <maisem@tailscale.com>	1 year ago
Joe Tsai	c6fadd6d71	all: implement AppendText alongside MarshalText (#9207) ... This eventually allows encoding packages that may respect the proposed encoding.TextAppender interface. The performance gains from this is between 10-30%. Updates tailscale/corp#14379 Signed-off-by: Joe Tsai <joetsai@digital-static.net>	1 year ago
Brad Fitzpatrick	003e4aff71	control/controlclient: clean up various things in prep for state overhaul ... We want the overall state (used only for tests) to be computed from the individual states of each component, rather than moving the state around by hand in dozens of places. In working towards that, we found a lot of things to clean up. Updates #cleanup Change-Id: Ieaaae5355dfae789a8ec7a56ce212f1d7e3a92db Co-authored-by: Maisem Ali <maisem@tailscale.com> Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	9cbec4519b	control/controlclient: serialize Observer calls ... Don't just start goroutines and hope for them to be ordered. Fixes potential regression from earlier 7074a40c0. Updates #cleanup Change-Id: I501a6f3e4e8e6306b958bccdc1e47869991c31f7 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	f7b7ccf835	control/controlclient, ipn/ipnlocal: unplumb a bool true literal opt ... Updates #cleanup Change-Id: I664f280a2e06b9875942458afcaf6be42a5e462a Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Maisem Ali	959362a1f4	ipn/ipnlocal,control/controlclient: make Logout more sync ... We already removed the async API, make it more sync and remove the FinishLogout state too. This also makes the callback be synchronous again as the previous attempt was trying to work around the logout callback resulting in a client shutdown getting blocked forever. Updates #3833 Signed-off-by: Maisem Ali <maisem@tailscale.com>	1 year ago
Maisem Ali	7074a40c06	control/controlclient: run SetControlClientStatus in goroutine ... We have cases where the SetControlClientStatus would result in a Shutdown call back into the auto client that would block forever. The right thing to do here is to fix the LocalBackend state machine but thats a different dumpster fire that we are slowly making progress towards. This makes it so that the SetControlClientStatus happens in a different goroutine so that calls back into the auto client do not block. Also add a few missing mu.Unlocks in LocalBackend.Start. Updates #9181 Signed-off-by: Maisem Ali <maisem@tailscale.com>	1 year ago
Brad Fitzpatrick	86dc0af5ae	control/controlclient: rename Auto cancel methods, add missing Lock variant ... Then use the Locked variants in Shutdown while we already hold the lock. Updates #cleanup Change-Id: I367d53e6be6f37f783c8f43fc9c4d498d0adf501 Co-authored-by: Maisem Ali <maisem@tailscale.com> Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	04e1ce0034	control/controlclient: remove unused StartLogout ... Updates #cleanup Co-authored-by: Maisem Ali <maisem@tailscale.com> Change-Id: I9d052fdbee787f1e8c872124e4bee61c7f04d142 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	c621141746	control/controlclient: cancel map poll when logging out ... Don't depend on the server to do it. Updates #cleanup Change-Id: I8ff40b02aa877155a71fd4db58cbecb872241ac8 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	313a129fe5	control/controlclient: use slices package more ... Updates #cleanup Change-Id: Ic17384266dc59bc4e710efdda311d6e0719529da Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	f5bfdefa00	control/controlclient: de-pointer Status.PersistView, document more ... Updates #cleanup Updates #1909 Change-Id: I31d91e120e3b299508de2136021eab3b34131a44 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	7053e19562	control/controlclient: delete Status.Log{in,out}Finished ... 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>	1 year ago
Brad Fitzpatrick	9ce1f5c7d2	control/controlclient: unexport Status.state, add test-only accessor ... Updates #cleanup Updates #1909 Change-Id: I38dcde6fa0de0f58ede4529992cee2e36de33dd6 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	14320290c3	control/controlclient: merge, simplify two health check calls ... I'm trying to remove some stuff from the netmap update path. Updates #1909 Change-Id: Iad2c728dda160cd52f33ef9cf0b75b4940e0ce64 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	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	6b882a1511	control/controlclient: clean up a few little things ... De-pointer a *time.Time type, move it after the mutex which guard is, rename two test-only methods with our conventional "ForTest" suffix. Updates #cleanup Change-Id: I4f4d1acd9c2de33d9c3cb6465d7349ed051aa9f9 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	55bb7314f2	control/controlclient: replace a status func with Observer interface ... For now the method has only one interface (the same as the func it's replacing) but it will grow, eventually with the goal to remove the controlclient.Status type for most purposes. Updates #1909 Change-Id: I715c8bf95e3f5943055a94e76af98d988558a2f2 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	590c693b96	types/logger: add AsJSON ... Printing out JSON representation things in log output is pretty common. Updates #cleanup Change-Id: Ife2d2e321a18e6e1185efa8b699a23061ac5e5a4 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	a79b1d23b8	control/controlclient: convert PeersChanged nodes to patches internally ... So even if the server doesn't support sending patches (neither the Tailscale control server nor Headscale yet do), this makes the client convert a changed node to its diff so the diffs can be processed individually in a follow-up change. This lets us make progress on #1909 without adding a dependency on finishing the server-side part, and also means other control servers will get the same upcoming optimizations. And add some clientmetrics while here. Updates #1909 Change-Id: I9533bcb8bba5227e17389f0b10dff71f33ee54ec Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	947def7688	types/netmap: remove redundant Netmap.Hostinfo ... It was in SelfNode.Hostinfo anyway. The redundant copy was just costing us an allocation per netmap (a Hostinfo.Clone). Updates #1909 Change-Id: Ifac568aa5f8054d9419828489442a0f4559bc099 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	db017d3b12	control/controlclient: remove quadratic allocs in mapSession ... The mapSession code was previously quadratic: N clients in a netmap send updates proportional to N and then for each, we do N units of work. This removes most of that "N units of work" per update. There's still a netmap-sized slice allocation per update (that's #8963), but that's it. Bit more efficient now, especially with larger netmaps: │ before │ after │ │ sec/op │ sec/op vs base │ MapSessionDelta/size_10-8 47.935µ ± 3% 1.232µ ± 2% -97.43% (p=0.000 n=10) MapSessionDelta/size_100-8 79.950µ ± 3% 1.642µ ± 2% -97.95% (p=0.000 n=10) MapSessionDelta/size_1000-8 355.747µ ± 10% 4.400µ ± 1% -98.76% (p=0.000 n=10) MapSessionDelta/size_10000-8 3079.71µ ± 3% 27.89µ ± 3% -99.09% (p=0.000 n=10) geomean 254.6µ 3.969µ -98.44% │ before │ after │ │ B/op │ B/op vs base │ MapSessionDelta/size_10-8 9.651Ki ± 0% 2.395Ki ± 0% -75.19% (p=0.000 n=10) MapSessionDelta/size_100-8 83.097Ki ± 0% 3.192Ki ± 0% -96.16% (p=0.000 n=10) MapSessionDelta/size_1000-8 800.25Ki ± 0% 10.32Ki ± 0% -98.71% (p=0.000 n=10) MapSessionDelta/size_10000-8 7896.04Ki ± 0% 82.32Ki ± 0% -98.96% (p=0.000 n=10) geomean 266.8Ki 8.977Ki -96.64% │ before │ after │ │ allocs/op │ allocs/op vs base │ MapSessionDelta/size_10-8 72.00 ± 0% 20.00 ± 0% -72.22% (p=0.000 n=10) MapSessionDelta/size_100-8 523.00 ± 0% 20.00 ± 0% -96.18% (p=0.000 n=10) MapSessionDelta/size_1000-8 5024.00 ± 0% 20.00 ± 0% -99.60% (p=0.000 n=10) MapSessionDelta/size_10000-8 50024.00 ± 0% 20.00 ± 0% -99.96% (p=0.000 n=10) geomean 1.754k 20.00 -98.86% Updates #1909 Change-Id: I41ee29358a5521ed762216a76d4cc5b0d16e46ac Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	cb4a61f951	control/controlclient: don't clone self node on each NetworkMap ... Drop in the bucket, but have to start somewhere. Real wins will come once this is done for peers. │ before │ after │ │ B/op │ B/op vs base │ MapSessionDelta/size_10-8 10.213Ki ± ∞ ¹ 9.650Ki ± ∞ ¹ -5.51% (p=0.008 n=5) MapSessionDelta/size_100-8 83.64Ki ± ∞ ¹ 83.08Ki ± ∞ ¹ -0.67% (p=0.008 n=5) MapSessionDelta/size_1000-8 800.8Ki ± ∞ ¹ 800.3Ki ± ∞ ¹ -0.07% (p=0.008 n=5) MapSessionDelta/size_10000-8 7.712Mi ± ∞ ¹ 7.711Mi ± ∞ ¹ -0.01% (p=0.008 n=5) geomean 271.1Ki 266.8Ki -1.59% │ before │ after │ │ allocs/op │ allocs/op vs base │ MapSessionDelta/size_10-8 73.00 ± ∞ ¹ 72.00 ± ∞ ¹ -1.37% (p=0.008 n=5) MapSessionDelta/size_100-8 524.0 ± ∞ ¹ 523.0 ± ∞ ¹ -0.19% (p=0.008 n=5) MapSessionDelta/size_1000-8 5.025k ± ∞ ¹ 5.024k ± ∞ ¹ -0.02% (p=0.008 n=5) MapSessionDelta/size_10000-8 50.02k ± ∞ ¹ 50.02k ± ∞ ¹ -0.00% (p=0.040 n=5) geomean 1.761k 1.754k -0.40% Updates #1909 Change-Id: Ie19dea3371de251d64d4373dd00422f53c2675ea Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	84b94b3146	types/netmap, all: make NetworkMap.SelfNode a tailcfg.NodeView ... Updates #1909 Change-Id: I8c470cbc147129a652c1d58eac9b790691b87606 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	b5ff68a968	control/controlclient: flesh out mapSession to break up gigantic method ... Now mapSession has a bunch more fields and methods, rather than being just one massive func with a ton of local variables. So far there are no major new optimizations, though. It should behave the same as before. This has been done with an eye towards testability (so tests can set all the callback funcs as needed, or not, without a huge Direct client or long-running HTTP requests), but this change doesn't add new tests yet. That will follow in the changes which flesh out the NetmapUpdater interface. Updates #1909 Change-Id: Iad4e7442d5bbbe2614bd4b1dc4b02e27504898df Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago
Brad Fitzpatrick	165f0116f1	types/netmap: move some mutations earlier, remove, document some fields ... And optimize the Persist setting a bit, allocating later and only mutating fields when there's been a Node change. Updates #1909 Change-Id: Iaddfd9e88ef76e1d18e8d0a41926eb44d0955312 Signed-off-by: Brad Fitzpatrick <bradfitz@tailscale.com>	1 year ago