Commit Graph
221 Commits
Author SHA1 Message Date
Kristoffer Dalby e3f233aa55 mapper: send nothing while any stream awaits its initial map
Sent to the ready streams alone, a change counted as delivered and
never reached the new stream; the worker retries it once the map is out.
2026-10-30 14:55:37 +00:00
Kristoffer Dalby 3201077628 mapper: cap concurrent map sessions per node
A node chose how many map streams it held, each pinning a session.
The fifth replaces the oldest; same shape as #3468's mapper commit.
2026-10-30 14:55:37 +00:00
Kristoffer Dalby 11ae125a9f change: coalesce policy recomputes that carry an origin node
The recompute render ignores the origin and always carries the node's
own entry, so a burst of tag writes cost every node one per write.
2026-10-30 14:55:37 +00:00
Kristoffer Dalby 5c5fc4feb7 mapper: show an expired peer online, without endpoints or DERP
A connected expired node shows online to its peers. Neither its map
entry nor endpoint, DERP or disco patches carry the paths to reach it.
2026-10-30 14:55:37 +00:00
Kristoffer Dalby 6aed0d5f07 mapper: trim the Hostinfo sent to peers and to the node itself
As on Tailscale SaaS, the node's own entry keeps Hostname, RequestTags,
RoutableIPs and Services; peers get Hostname, OS, Services without
peerapi-dns-proxy, SSH host keys and AppConnector, and resend on those.
2026-10-30 14:55:36 +00:00
Kristoffer Dalby 706155d1f1 mapper: keep an expired node authorised in its own entry
Tailscale SaaS sends an expired node MachineAuthorized and no Expired
flag; its past KeyExpiry already moves the client to NeedsLogin.
2026-10-30 14:55:36 +00:00
Kristoffer Dalby 4e433f7a04 mapper: send a router all of its own routes
Its AllowedIPs held only the primaries the policy let it reach. It now
holds every approved route, and PrimaryRoutes the ones it serves.
2026-10-30 14:55:36 +00:00
Kristoffer Dalby 8967c7e91c mapper: count only the batcher's goroutines in lifecycle tests
runtime.NumGoroutine deltas moved when goroutines left by earlier
tests exited mid-test. Label the batcher's goroutines and count those.
2026-10-30 14:55:36 +00:00
Kristoffer Dalby 3c99f6bade policy/v2, mapper: add nextdns attr captures
SaaS accepts malformed nextdns profiles and keeps them in the CapMap;
its resolver skips them. DoH addresses compare per node: path, query,
ID.
2026-10-30 14:55:34 +00:00
arkie 8cf43952af types: emit SSH environment capability by default
Fixes #3559
2026-10-29 20:51:36 +01:00
Kristoffer Dalby e3c1bfd68e mapper: skip peers whose key expiry cannot be encoded
json.Marshal refuses a year outside [0,9999], so one stored expiry
failed every map with that node. TailNode and key patches send UTC, the
builder drops what fails and the boot scan names the node.

Fixes ANT-2026-QMCAFJ1Q
2026-10-29 14:00:02 +01:00
Kristoffer Dalby eed5c8c80a mapper: count folded patches when bounding pending
A fold keeps every patch it merges, so counting changes hid the
backlog: the cap never fired and each fold recopied the patches.
2026-10-23 18:23:53 +02:00
Kristoffer Dalby daad12e964 mapper: retry slow sessions from current state
A full session channel after the send window evicted live but busy clients.
Requeue and regenerate; evict after no progress for connectionStallTimeout.
Retire blocked reconnect siblings once the newest stream accepts the frame.

Updates #3129.
2026-10-23 18:23:53 +02:00
Kristoffer Dalby f46ea3d1df mapper: skip responses left empty by visibility filtering
A peer change hidden from a recipient still produced a frame carrying only
ControlTime, one per sender per recipient. Drop it once the response is final.
2026-10-23 18:23:53 +02:00
Kristoffer Dalby 2047ef35b8 hscontrol: wait for map session cleanup during shutdown
Batcher shutdown cuts reconnect grace short to keep shutdown prompt.
Wait for each session's Disconnect to persist LastSeen before the database closes.
2026-10-23 18:23:53 +02:00
Kristoffer Dalby 8a7921ae45 mapper: fold a node's pending backlog while its bundle is in flight
Behind workers, pending grew by every change until the next drain; under
ci-churn that held ~900 MB of queued changes.
2026-10-23 18:23:53 +02:00
Kristoffer Dalby 7a2335dc2b mapper: render each tick's peer deltas as one response
One frame per change let a join storm overflow the session channel in a
single tick, and the batcher then dropped live clients as stale.
2026-10-23 18:23:53 +02:00
Kristoffer Dalby fb015933ca mapper: drop pending changes for nodes without a connection
Disconnected nodes keep their batcher entry, and every broadcast was
rendered for them only for send to discard it.
2026-10-23 18:23:53 +02:00
Kristoffer Dalby 7414c276f6 mapper: limit capability copies to requested peers
Small peer updates should not copy capabilities for the entire tailnet.
2026-10-22 22:17:36 +02:00
Kristoffer Dalby 6489054805 mapper: enqueue broadcasts without refiltering recipients
Broadcast recipients already share the same changes.
2026-10-22 22:17:36 +02:00
Kristoffer Dalby 068711518f mapper: store user views directly in profile maps
Avoid allocating an owner pointer for every peer during profile generation.
2026-10-22 22:17:36 +02:00
Kristoffer Dalby 79ae38f0af mapper: stop shadowing predeclared cap
golangci-lint 2.15 predeclared check flags it.
2026-10-22 14:37:00 +02:00
Kristoffer Dalby eef0d1fe47 dns: add acme-dns provider
_acme-challenge is CNAMEd once to an acme-dns registration, so headscale
needs no DNS credentials. Node certificates need one registration per
name; CNAME chains are followed and writes never evict a live value.

Updates #2137
2026-10-18 20:53:18 +02:00
Kristoffer Dalby 3f6e54cbd5 mapper: send CertDomains and https cap when certificates are enabled
Both gate tailscale cert and serve HTTPS on the client; httpsEnabled
follows, as in Tailscale. No certificate when an operator extra record
sits at or below the name: a wildcard would cover it.

Updates #2137
Updates #1921
2026-10-18 20:53:18 +02:00
Kristoffer Dalby 4fd4da75f2 hscontrol: fix formatting after combining SSH policy changes
Updates #3517
2026-10-08 10:29:23 +02:00
Kristoffer Dalby 785ba22c65 mapper: send SSHPolicy only when it changes for the connection
Any non-nil SSHPolicy forces a full client netmap rebuild; the empty
policy now sent to every node made each policy change one.

Updates #3508
2026-10-08 10:29:23 +02:00
Kristoffer Dalby d7b333d2aa mapper: send an empty SSH policy when compiling it fails
Fail closed; nil kept the client's previous rules.

Updates #3508
2026-10-08 10:29:23 +02:00
Kristoffer Dalby 48046dc9c5 mapper: allow unchanged self in the full-update ping test
Updates #3518
2026-10-07 23:36:16 +02:00
Kristoffer Dalby ea6b0bf0b7 mapper: reconcile self on policy recomputes
Self renders from the same state as peers, so leaving it out of policy
responses hid an exit node's approved routes until it reconnected.

Fixes #3502
2026-10-07 23:36:16 +02:00
Kristoffer Dalby 52a7f8c89c mapper: send the logtail instruction on every full map
Full maps moved to buildFromChange and lost WithDebugConfig, so no
frame told clients to disable log upload. Enabled logtail sends nil.
2026-10-07 18:11:07 +02:00
Kristoffer Dalby d4948da301 mapper: keep pings when a full update collapses pending changes
A full renders state at drain time but cannot carry a one-shot
PingRequest; queue each ping as a ping-only frame after the full.
2026-10-07 15:25:25 +02:00
Kristoffer Dalby b4d5cdd6aa mapper: keep reconnect removals tracked until delivered
AddNode reset lastSentPeers before the queued removal went out, so a
superseding full or a disconnect before the tick lost it for good.
2026-10-07 11:09:47 +02:00
Kristoffer Dalby eeaac680be mapper: drop DNSConfig from policy responses
It forced every client into a full netmap rebuild; the resolver race it
guarded against was a client bug (tailscale/tailscale#19749).
2026-09-30 19:03:13 +02:00
Kristoffer Dalby 60d42808b8 state: keep node removal apart from policy refresh in DeleteNode
Peers get the deletion as an explicit PeersRemoved change, independent of
their sent-peers tracking.

Updates tailscale/tailscale#15660
2026-09-30 19:03:13 +02:00
Kristoffer Dalby 1bd62737b9 mapper: send removed peers as their own delta
Removals derived from a policy change, full update or reconnect rode a
response whose DNSConfig, SSHPolicy, Node or Peers force a full rebuild.

Updates tailscale/tailscale#15660
2026-09-30 19:03:13 +02:00
Kristoffer Dalby 21f6e46fb8 state: refresh policy nodes inside the peer map build
One peer build per tag/user/IP/route write; callers detect policy moves
via NodesGeneration. Per-node caches only store results for the node
pm holds, so a mapper reading mid-build cannot pin a stale filter.
2026-09-30 17:21:02 +02:00
Kristoffer Dalby 393dd3e2d9 db: store all credentials in one SHA-256-hashed table
API keys, pre-auth keys and OAuth clients/tokens share one table and verify
path. Secrets carry 256 bits of crypto/rand entropy, so a SHA-256 digest
needs no stretching; bcrypt/argon2id rows rehash on use until 0.32.
2026-09-26 00:33:12 +02:00
Kristoffer Dalby 06fa3075da hscontrol: replace tailscale line refs with doc links
Line numbers drift on every upstream bump.
2026-09-23 14:48:45 +02:00
Kristoffer Dalby e48bc46cc6 mapper: take peer visibility from the peer map only
Fixes #3408
2026-09-23 14:48:35 +02:00
Kristoffer Dalby 4c6a2dff52 state: resolve changed peers through adjacency
ListPeers now filters named peers against the recipient's adjacency, so
a node the policy hides is never delivered.

Updates #3417
2026-09-10 13:06:13 +02:00
Kristoffer Dalby 286f1d5a12 mapper: drop empty changes before fan-out
An empty change carries no work for any recipient, so it never becomes
a pending entry. Adds headscale_mapper_changes_dropped_total.

Updates #3417
2026-09-10 13:06:13 +02:00
Kristoffer Dalby 8995d8a558 mapper: assert no map response lists the recipient as its own peer
Covers every change shape under four policy shapes, plus connect churn. The
zero-matcher shape is the gap: buildTailPeers skips ReduceNodes there, so the
peer lookup is the only self filter left.
2026-09-09 18:18:53 +02:00
Kristoffer Dalby a91c0519c2 change, mapper: distinguish deleted nodes
Updates #3410
2026-09-09 18:18:17 +02:00
Kristoffer Dalby 0b69e844f5 mapper: stop a deleted node's map session
Dropping the batcher entry left serveLongPoll streaming to a node that no
longer exists: Close ranges b.nodes and can no longer reach it, so shutdown
blocks and the client keeps polling instead of re-authenticating.

Updates #3410
2026-09-09 18:18:17 +02:00
Kristoffer Dalby 63123196cc all: inline deprecated tailcfg capability aliases
go 1.27 vet reports the //go:fix inline directives tailscale added to
the tailcfg cap aliases; applied with `go fix -inline ./...`.
2026-08-25 21:59:00 +02:00
Simon Law 3aece65014 build: update tailscale.com to v1.103.0-pre.0.20260819232550-0e84b4a3a0a0
PR tailscale/tailscale#20646 changed the representation of DERP region
IDs from plain ints to a tailcfg.DERPRegionID type. This patch
requires that version of the tailscale.com library and updates our
code to also use the new type.

Updates: tailscale/tailscale#20165
(cherry picked from commit 79695ab51b)
2026-08-25 21:59:00 +02:00
Kristoffer Dalby d7c1ecce92 mapper: correct the send timer rationale
time.After has been GC-recoverable since Go 1.23, so the leak the comment
described cannot happen and the test asserting it proved nothing.
2026-08-25 21:59:00 +02:00
Kristoffer Dalby 373c60efe3 all: use sync.WaitGroup.Go
Drops the Add/Done bookkeeping, which in batcher.go was spread across
three functions.
2026-08-25 21:59:00 +02:00
Kristoffer Dalby 08956d51a4 mapper: skip peers with invalid names instead of failing the map
A peer whose GivenName fails GetFQDN aborted the whole map for every node
that could see it. Drop and log it; SSH policy errors degrade too.

Fixes #3346
2026-07-01 15:19:08 +02:00
Kristoffer Dalby d528686f14 mapper,policy: add reconnect-storm and lock-concurrency regression tests
TestInitialMapNotStarvedByReconnectStorm reproduces the #3346 stall;
TestPolicyManagerConcurrentReads guards the RLock cache access under -race.

Updates #3346
2026-07-01 14:41:03 +02:00