Aleksandr Bogomiakov
90ef5cefec
Offline devices reserved names and IPv4 addresses indefinitely, leaving stale members in the device history. Give every authenticated network participant the same right to retire another device's observed signed claim without an owner, quorum, permanent identity ban or secret change. Add a separately signed retirement containing the target's original signed claim as proof. Verify both signatures and network scopes, retain the strongest proved cutoff per target with deterministic ties, and suppress claim versions at or below it in membership, hostname ownership and IPv4 allocation. Keep the old signed records and range agreement, and retain markers after a newer claim returns so old replicas cannot resurrect obsolete reservations. A returning device signs a newer claim, allocates against current holders and resets its previous hostname priority rather than taking a reused name with an old time. Commit local retirement and the requester's author counter atomically before announcing it; receivers persist proofs before applying them. Bound history to 1024 targets by default, reject additional targets without evicting protection, and treat corrupt or over-limit mandatory storage as an explicit error. Repeat bounded marker batches with a per-session rotating cursor so queue pressure cannot indefinitely starve later markers. Exclude retired claims and cached hints from history-derived discovery without preventing a fresh authenticated device from returning. Expose Agent::forget_node through the shared local control API. CLI command tsng network forget-node accepts exact network names, full ids or unique id prefixes, and exact node hostnames, ids, prefixes or overlay IPv4 addresses. Ambiguous selections change nothing. Recheck the observed version and refuse self, online nodes, the selected exit node and absent or changed claims. Add Forget with an identity confirmation to offline tray device rows. Widen Show devices to 1440x640 with an 800x400 minimum to fit its contents. Control ALPN/version advances to tsunagi/ctrl/7 and local IPC to 21; upgrade all peers and restart the agent with matching CLI/tray clients. Data ALPN 5, network identity derivation, handshake encoding and existing record signing bytes remain unchanged. SQLite schema 9 adds retirement storage while retaining identity, secrets, settings and signed claims. Document behavior and limits. Validation on Linux through nix develop: - cargo fmt --all -- --check - cargo clippy --locked --workspace --all-targets -- -D warnings - cargo test --locked --workspace --all-targets: 519 passed, 2 existing ignored tests; inherited TSUNAGI_CONTROL_SOCKET and TSUNAGI_CONTROL_GROUP were unset - git diff --check and the built CLI forget-node help Tests use real iroh authentication and SQLite. Cover restart, stale-replica recovery, equal rights and network isolation, hostname reuse and return, twelve marker batches through a two-message queue, malformed signatures and proofs, invented future cutoffs, deterministic merge, capacity, migration and injected counter-write rollback, selectors and cleanup through two real CLI processes. Earlier iterations caught a private test API, a collapsible tray conditional and a crash-based CLI test waiting for QUIC's idle timeout; corrected before the final successful suite. Manual tray rendering, macOS/Windows, privileged TUNs and public-network traversal were not run. The requested RC tag is v0.1.0-rc.31; no push was requested.
7.0 KiB
Userspace mesh routing
local OS → TUN → IP plugin (encrypt) → router → transport peer link
transport peer link → router → transport peer link (transit)
transport peer link → router → IP plugin (decrypt) → TUN → local OS
The TUN is the boundary of the local host. The kernel sees one overlay route; it never forwards transit traffic. WireGuard encryption remains end to end. The core router sees only an opaque payload and routing metadata. iroh is the current transport adapter, with fragmentation below the routing envelope so the default 1280-byte interface MTU also works on small QUIC paths.
Graph and lifecycle
Each authenticated control session advertises that peer's own direct data links. There is one table for each network, whatever protocols run over it: a link is a way to reach a neighbour, and a protocol is an end-to-end tunnel to a peer. The envelope names the protocol, so the two are independent. A neighbour reached over several kinds of transport (UDP, TCP, QUIC datagrams) is sent to over the best open one, in the order the protocols were configured, and over the next if that one fails or cannot carry the frame. Every authenticated member enters the graph; a member in the middle needs no protocol in common with either end, only a link to each. A snapshot refreshes on the maintenance interval, expires after 90 seconds, and is withdrawn when the session ends. Link arrival, closure and changed announcements rebuild routes immediately. Unchanged announcements refresh their age without republishing the forwarding snapshot.
Which protocol a pair of members runs their tunnel in is settled without asking: the member with the lower endpoint id decides, by the order it announces its protocols in, among those both have enabled at the same wire version. Both ends read the same answer from what they already hold. The tunnel is created as soon as the peer is reachable by any path, and it does not change when a direct link comes or goes underneath it: as soon as a direct link comes up it is shorter than the path through others, and traffic moves to it with nothing to redo. The user sees only a lower round-trip time.
Breadth-first search computes the shortest directed paths and every equal-cost first hop in sorted order. An actual direct link always wins. The first row is always supplied by local transport state, never by a remote claim. Routes stop at 16 links. A packet also carries a decreasing hop limit, bounding temporary loops if different nodes have not yet received the same topology update.
The existing control plane still forms authenticated pairwise sessions. Routing provides multihop data paths among these members; it does not add control-plane flooding or carry control sessions through the data plane, so a member that has no control session with another has no data path to it either. Announcements are volatile, not durable membership or availability guarantees.
What "online" means
One definition, computed by the agent and reported identically by tsng status
and the tray: a member is online when this agent can reach it by any path — a
live authenticated control session (QUIC keep-alive every 5 seconds, so a
vanished peer is noticed within about 15), or a data path to it through other
members. It is never inferred from ping, which a firewall may block, or from a
tunnel's last handshake, which says only when traffic last flowed.
How it is reached is reported separately: direct over a named transport,
relayed through a named member over N hops, or no data path yet while one is
being set up. Being relayed is not being degraded in any way but distance, and
is shown so it is not a surprise.
Packet path
The control loop binds next hops to transport handles in an immutable table
and publishes it through ArcSwap. Readers do not acquire the topology mutex.
The transit routine validates the fixed 82-byte envelope, looks up the source
and destination, selects a cached next hop, decrements one byte, and calls the
transport's synchronous datagram send. It never searches the graph, formats an
endpoint string, validates an Ed25519 key, decrypts, or touches TUN.
An exclusively owned receive buffer is reused when decrementing the hop limit.
Shared buffers require a copy through the safe bytes API. There is no added
transit queue or timer; every raw transport has its own reader. Readers yield
after 64 ready packets to avoid starving other runtime tasks. Local delivery
uses a bounded 256-datagram inbox and drops a new packet when full. Transport
queues and fragment reassembly retain their own limits.
PacketLink::send_flow accepts an opaque 64-bit flow identifier. IP plugins
derive it before encryption from source/destination addresses, protocol and
TCP/UDP ports. WireGuard keeps tags alongside its bounded pending queue, so
the first application packets retain their flow identity after a handshake.
Transit preserves the tag; it never hashes changing ciphertext. Handshake and
keepalive frames use flow zero. Fragmented IP traffic uses a coarse address/
protocol hash because later fragments lack ports; fragmented flows between
the same addresses coalesce. A transition between fragmented and unfragmented
traffic can change its path. The default overlay MTU avoids needing IP
fragmentation for ordinary host TCP traffic.
ECMP is deterministic for a source and flow while the table is unchanged. Closed next hops are skipped until the control loop replaces the table. A topology change can move a flow; datagrams remain unreliable and unordered, and there is no promise to preserve order across a physical path failure. Logical peer links and end-to-end encryption state survive these changes.
Carrier health and switching
Carrier setup failures and persistent encrypted delivery loss advance the initiator's ordered fallback cycle. Logical end-to-end links survive while physical links are retired explicitly. The AWG carrier negotiates a fresh private pair profile on every reentry, without changing topology's durable state. See amnezia.md for thresholds and bounds.
Verification and measurement
The default offline suite exercises four real agents, iroh transports and WireGuard tunnels with memory TUNs, plus graph/forwarder unit tests. It needs neither administrative rights nor public relays/DHT.
Run the forwarding microbenchmark explicitly:
cargo test --release -p tsunagi-core --lib forwarding_benchmark -- --ignored --nocapture
It processes one million 64-byte frames and one million 1280-byte frames with two equal next hops. Frame construction is outside the timed section; the actual ingress routine, header validation, snapshot lookup, ECMP, TTL update, buffer release and mock transport submission are inside. The mock transport counts sends without storing frames. Results are CPU forwarding cost, not end-to-end network latency or VPN throughput; encryption, fragmentation, sockets, congestion and scheduling contribute separately.
Wire compatibility: control ALPN tsunagi/ctrl/7, data ALPN tsunagi/data/5.
Upgrade every participant together; saved identities and network state persist.