Reworks the Linux packaging so the service runs as an unprivileged, dedicated "tsunagi" user instead of root — the point of a native package is to set things up properly for that OS's users — and adds Arch Linux packages beside the deb. Both package families are built from the already-compiled static binary; nothing is recompiled. Service model (shared systemd unit, dist/linux/tsunagi.service): - User=tsunagi with AmbientCapabilities=CAP_NET_ADMIN CAP_NET_BIND_SERVICE, so it creates the TUN and binds low ports without root. - StateDirectory/CacheDirectory/RuntimeDirectory give it /var/lib/tsunagi, /var/cache/tsunagi and /run/tsunagi, and the agent is pointed at them (and the socket at /run/tsunagi/agent.sock) through TSUNAGI_* env in the unit. - The polkit rule now grants the resolve1 actions to the "tsunagi" group; the service user is in it, and admins can add themselves. CLI: when no state dir is chosen, the control commands prefer the system service's socket at /run/tsunagi/agent.sock when it exists, so `sudo tsunagi status`/`join` reach the packaged daemon with no flags. A per-user agent (its socket absent from /run) still uses its own. Packaging: - deb (build-deb.sh): postinst now creates the tsunagi system user+group; unit installed to /lib/systemd/system, binary to /usr/bin. - Arch (new dist/arch/build-pkg.sh): a dependency-free .pkg.tar.zst assembled with bsdtar + zstd (no makepkg, no Arch host) — .PKGINFO, a .INSTALL scriptlet that creates the user and reloads systemd, and a gzipped .MTREE; unit to /usr/lib/systemd/system. - The polkit rule moved to dist/linux (shared by deb, Arch and the tarball), and the tarball install.sh now creates the user and installs the unit + rule. - Both release workflows build the deb and the Arch package on the two Linux matrix entries (deb_arch/pkg_arch), install libarchive-tools+zstd for it, and attach both to the release (Gitea API loop / gh). Not exercised in this session: the .deb and .pkg.tar.zst build and install only on Linux (no dpkg-deb/bsdtar+pacman locally); CLI build, clippy, fmt and tests pass, and all shell/YAML is syntax-checked. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2.4 KiB
Running tsunagi on Linux
This archive holds a static tsunagi binary (musl), so it runs on any
distribution — Arch, Alpine, NixOS, old glibc systems — with no shared-library
dependencies. Alongside it are a systemd unit, a polkit rule, and an installer.
Creating the TUN interface needs CAP_NET_ADMIN, and configuring
systemd-resolved is decided by polkit. There are three ways to run it; pick one.
On Debian/Ubuntu and Arch, prefer the native package (
.deb/.pkg.tar.zst) from the release — it does all of the below for you.
Quick install — systemd service as a dedicated user
sudo ./install.sh
sudo systemctl enable --now tsunagi
The service runs as an unprivileged tsunagi user with just CAP_NET_ADMIN
and CAP_NET_BIND_SERVICE (granted by systemd), not root. systemd creates its
state (/var/lib/tsunagi), cache (/var/cache/tsunagi) and runtime
(/run/tsunagi) directories, and the polkit rule lets it configure the
resolver. Control it with sudo — the CLI finds the running service's socket at
/run/tsunagi/agent.sock with no flags:
sudo tsunagi status
sudo tsunagi join <network-name> <tsn1…secret>
As your own user — one capability, no service
sudo install -m 0755 tsunagi /usr/local/bin/tsunagi
sudo setcap cap_net_admin,cap_net_bind_service+p /usr/local/bin/tsunagi
sudo usermod -aG tsunagi "$USER" # so the resolver works; log in again after
tsunagi up
Nothing runs as root. The capability is lost on every rebuild or copy of the
binary, so re-run setcap after replacing it. The agent uses your per-user
state directory and socket, so plain tsunagi status (no sudo) reaches it.
Without touching the OS
tsunagi up --no-tun
Tunnels form and handshake between agents, but no interface, address or route is created, so traffic never reaches the operating system. Needs no privileges.
What the agent installs
Everything is tagged to its own interface, computed from local state, and reversible — removed when a network leaves, broadcast is turned off, or the agent stops:
- a TUN interface configured over netlink (addresses, MTU, up);
- while broadcast is on, a
255.255.255.255route through the interface and an inbound-UDP firewall allowance from the overlay range (iptables, tagged); - the systemd-resolved setting that sends the overlay zones' questions to the local resolver.
Nothing taken from a remote peer ever becomes a path, a command argument or an OS setting.