* feat(aquamarine): make the iGPU usable for media and OpenVINO
The R9700s stay reserved for inference: VA-API/QSV transcoding moves to the iGPU media
engine, and `openvino-gpu` works around ROCm's OpenCL loader shadowing the plugin's
`ocl-icd` (WA-038).
Assisted-by: DeepSeek Harness 0.2.1-rc.2 (deepseek-v4.1-flash)
* docs(aquamarine): document OpenVINO on the iGPU
How to run models on the iGPU: the `openvino-gpu` wrapper, the encoder staticization Whisper needs
on `GPU`, compile/steady timings and memory, and the language and cache caveats.
Assisted-by: DeepSeek Harness 0.2.1-rc.2 (deepseek-v4.1-flash)
On a tmpfs root the flake/fetch cache is empty at every boot, so
`nix run <flake-ref>` re-downloads its input and fails until DNS is up: a reboot
test showed llama-swap's first start failing on that. Other caches write into
RAM meanwhile.
Assisted-by: DeepSeek Harness (deepseek-flash)
The second NVMe and its LUKS volume leave with the Z890 board, so the data-disk disko module is deleted; `ai` must activate this change before the disk is pulled, or its initrd blocks on the missing volume.
Assisted-by: DeepSeek Harness (deepseek-flash)
The Z890 board and its root SSD move here from `ai`, and the retired VM's name and 192.168.5.101 are reused.
Assisted-by: DeepSeek Harness (deepseek-flash)
The worker MicroVMs ran `home-manager-ryan.service`, which failed on two of them
(`/nix/var/nix/temproots` permission denied, and no `/home/ryan`), so
`HostSystemdServiceCrashed` fired for days. Mirror the masters and extend their
regression test to all six nodes.
Assisted-by: DeepSeek Harness 0.2.0-rc.2 (deepseek-flash)
* fix(postgres): require a password for superuser over TCP
* test: assert the postgres loopback auth rules
* fix(darwin): keep root-only agenix secrets root-only
* docs(security): record the local auth boundaries and retire WA-015
* fix(k3s): deny credential-bearing host ports to pod sources
* test: assert the k3s pod firewall drops sensitive ports
* docs(security): record the pod firewall boundary
dsh rewrites a profile's cordis.patch.yml at runtime, so linking the web
profile put GUI state in the checkout and let runtime writes clobber manual
edits. Link the stable policy as the $DSH_HOME home layer instead: dsh applies
it after every profile and only reads it.
The host-plane compaction-basic row is disabled on the web surface and each
agent preset owns its own copy, so the top-level override was inert. Drop it;
the model contextWindow/maxTokens are the effective knobs.
- add home/base/tui/agents/dsh/cordis.patch.yml, linked from default.nix
- drop the ~/.dsh/profiles/web link and the tracked web/ profile
- keep runtime state and private/local model routes in the profile patch
- update the eval test and README
Vinput's streaming Zipformer models have unbounded dynamic shapes, so
OpenVINO cannot compile them for the GPU or the NPU; the `openvino`
provider and ai's Intel NPU enablement could never run on the NPU (noted
as unverified in WORKAROUNDS WA-017). Remove the backend option, the
daemon's OpenVINO library path, and the intel-npu module enable, and
update the eval test, README and WORKAROUNDS (drop WA-016 and WA-017).
Keep Vinput's stock package and its bundled sherpa-onnx (1.13.4): the
fork-based and nixpkgs-master sherpa overrides existed only to provide the
`openvino` provider, so drop package.nix with them.
Assisted-by: opencode (DeepSeek V4.1 Flash)
* fix(desktop): tie ai's link DNS takeover to mihomo's lifecycle
sing-tun normally points systemd-resolved at the TUN DNS by running
resolvectl (dns/domain/default-route). Under nixpkgs' hardened
DynamicUser mihomo unit that hand-off fails silently: the unit's
RestrictAddressFamilies blocks resolvectl's D-Bus socket and resolve1
requires privileged authorization, so resolved keeps querying the
physical-link DNS, fake-ip is bypassed, and domain rules degrade to
MATCH.
Do the hand-off as '+' ExecStartPost/ExecStopPost hooks instead: '+'
commands run outside the sandbox (systemd.exec: address-family
restrictions do not apply to them) and ExecStopPost reverts even on
crashes, so a dead mihomo leaves DNS usable.
Signed-off-by: Ryan Yin <xiaoyin_c@qq.com>
* docs(mihomo): record sing-tun's resolvectl hand-off as the leak cause
Signed-off-by: Ryan Yin <xiaoyin_c@qq.com>
* refactor(desktop): keep mihomo only on ai and shoukei
ruby/kana never ran the core (service inactive live), and their
resolved + TUN combo would reintroduce the resolver leak once enabled.
Drop the enable and the explicit module import from computer-use; the
eval test now pins mihomo=false for both guests.
---------
Signed-off-by: Ryan Yin <xiaoyin_c@qq.com>
Add SECURITY.md covering the threat model, trust boundaries,
proportionate kernel/process hardening, the update/reboot routine and
remaining prioritized work. Record the dated read-only audit findings
(listeners, privileged workloads, effective RBAC, secret metadata and
sandbox/SSH boundaries) separately from planned controls and runtime
verification.
Define agent-first diagnosis: scoped root read-only access on homelab
hosts, and normal-user diagnostics on core desktops with no new sudo
grants. Add canonical domain-based k3s API endpoint regression tests,
record the unverified Darwin secret-ownership concern as WA-015, and
correct comments claiming mode 0000 blocks privileged root.
* fix(colmena): expose colmenaHive for the new evaluator
nixpkgs now ships the nix-community Colmena rewrite, whose default
evaluator reads an evaluated colmenaHive flake output instead of the raw
colmena hive. Without it every colmena apply recipe failed with
"does not provide attribute colmenaHive".
Add a colmena input pinned to the same v0.5.0 as nixpkgs and expose
colmenaHive = colmena.lib.makeHive <raw hive>, with the raw hive kept as
an internal binding rather than a flake output of its own.
* chore: clean up stale colmena and aquamarine references
- point the Colmena links at nix-community/colmena after the move
- rename the security-aquamarine-metrics test to youko-metrics (it
already tested youko) and update the outputs README
- drop the commented aquamarine remote-builder example
Hosts used to trust the router as the only boundary (servers ran with the
firewall off; service ports were opened globally), so any LAN device could
reach every host service. Move filtering to the hosts as defence in depth
behind the router, which keeps its firewall on and uses a port mapping for
the few ports we actually need.
- base: nftables backend; deny inbound by default; trust the home LAN by
source and the tailnet by interface (restrict the tailnet with ACLs);
suppress per-service `openFirewall` global ports; allow node_exporter only
from the monitoring host. The local podman bridge, the tailnet and the pod
CIDRs are accepted only after the node_exporter drop.
- servers: stop disabling the firewall; drop redundant global port opens
(caddy 80/443, NFS 2049, Jellyfin, localsend/test ports); delete the
redundant desktop firewall module; aarch64/riscv64 import the firewall
explicitly (akane was missing it).
- k3s: Cilium pods (10.0.0.0/8, fd00::/8) may reach host services; drop the
flannel-era cni0/flannel.1 trust.
- youko: postgres listens on localhost + the LAN with pg_hba limited to the
LAN; the NFS export stays LAN-scoped; remove the unauthenticated v2ray
proxy.
- tests: `security-firewall` asserts every host runs the shared firewall, has
the LAN CIDR rule and no global ports.
Keep the six node configuration intact while removing the failing user activation from the three control-plane MicroVMs. Update the regression test and current K3s configuration notes.
- run two autocutsel instances so the X PRIMARY/CLIPBOARD selections stay in
sync, making copy/paste work in the session and through VNC
- drop --disable-blink-features=AutomationControlled: nothing drives the
browser over CDP, so navigator.webdriver is already false and the flag only
marked the setup as automated
- document why the driver stays on real X11/AT-SPI input and is never granted
a CDP attach, and note the SSH-tunnel path for manual VNC access
- extend the computer-use eval test
Give the idols-ruby and idols-kana VMs a headless computer-use
environment: Xvfb plus i3, driven by cua-driver (and computer-use-linux
as a fallback). Includes the system module (AT-SPI, fonts, Mesa, Clash
Verge proxy, proxy-region timezone), the home i3 module (Xvfb/i3/x11vnc
services, browser flags), host wiring and an eval test.
Drop the now-meaningless kubevirt- prefix and move the hosts out of
hosts/k8s/, following the existing 12kingdoms-shoukei layout: the
directory and output carry the theme prefix while the configuration,
hostname and colmena target are the bare name (shoryu, shushou, youko).
- hosts/k8s/kubevirt-* -> hosts/12kingdoms-*; the shared disko config
becomes host-disko-fs.nix; the now-different relative imports are fixed
- outputs, vars/networking.nix keys, the Justfile recipes/tags, the eval
tests and the docs follow
- colmena reaches these hosts by IP (a new targetHost override), since
their DNS name changes with the hostname and is not resolvable until
the router learns it
Aquamarine's services now run directly on kubevirt-youko, so the guest
definition is no longer needed:
- move its service modules to hosts/k8s/kubevirt-youko/aquamarine/ and
drop the explicit import list, leaving aquamarine-services.nix with
just its IP and secret groups
- delete the guest base, journald override, output entry and home config
- repoint the eval tests that read the guest's config to kubevirt-youko
- drop the now-dead 'aqua' colmena recipe and its image upload
The aqua KubeVirt VM itself is deleted separately; the name is kept in
the docs for a possible later test microvm/qemu-vm on idols-ai.
It was never bootstrapped (no flux deploy key, VMs commented out since
2024-10) and only carried a stale duplicate of the test cluster's config.
- hosts/k8s/k3s-prod-1-{master,worker}-{1,2,3} + flake outputs + the home module
- the k3s-prod-1-token agenix entry (the .age file lives in the private secrets repo)
- the k3s-prod-1-* IPs, the offlineHosts entries, the Justfile recipes and the docs
* feat(btrbk): make snapshots explicit and guarded
Replace the always-on, hardcoded module with modules.btrbk options (enable/volume/subvolume/snapshotDir/target/onCalendar), and assert the btrfs top-level subvolume is mounted so a missing mount fails evaluation instead of silently failing at runtime.
Store snapshots in the @snapshots subvolume (snapshot_dir=@snapshots) rather than next to the live subvolume, make the off-host target explicit and off by default, and document that local snapshots are not off-host backups.
Enable the feature explicitly on idols-ai, shoukei and the three kubevirt hosts; qemu-guest now disables it through the option. Add eval tests for shoukei and ai.
* feat(monitoring): alert on btrbk snapshot failure
Add HostBtrbkSnapshotFailed on node_systemd_unit_state for btrbk-btrbk.service, so a broken snapshot schedule alerts instead of only surfacing in the journal.
- render Niri on the Intel iGPU while retaining NVIDIA-connected physical outputs
- encode Sunshine streams with Intel VAAPI and wait for a Niri output before startup
- load VKMS persistently for headless sessions and configure `Virtual-1` at 2560x1600@60
- let Moonlight request the stream bitrate instead of enforcing a 20 Mbps host cap
- remove unsupported Dynamic Boost and global NVIDIA session overrides
- grant Sunshine access to virtual input devices
- add a Nushell recovery helper that starts a temporary greetd/Niri session from SSH
- identify and safely replace normal, standalone, or previous transient Niri sessions
- launch the transient service in the background and verify greetd, Niri, and Sunshine before returning
- ignore local `.worktrees` directories and keep eval regression coverage focused on structural behavior
* security: make firewall secure-by-default, disable explicitly on servers
Flip the base firewall default to ON so new hosts are protected by default. Servers keep the firewall off (trusted internal LAN, WAN protected at the router) via explicit overrides in modules/nixos/server/{server,server-aarch64}.nix. Behavior-preserving for all 18 current hosts; adds a security-firewall eval test guarding the per-host state.
* test: fix security-firewall eval test to scalar per-host values
* security: enable AppArmor (complain) on all Linux hosts
* security: bind aquamarine metrics exporters to loopback
* security: restrict k3s kubeconfig file mode to 600
* security: disable SSH X11 forwarding on servers, keep on desktops
* security: add conservative systemd hardening to aquamarine services
* docs(apparmor): correct FHS-alias comment (nixpkgs has no broad FHS layer)
nixpkgs only creates /run/current-system, /usr/bin/env and /bin/sh; it does not
provide a broad FHS->store alias tree, so FHS-oriented abstractions are incomplete
on NixOS. They are safe only because the profiles run in complain mode.
* fix(aquamarine): scrape same-host exporters over loopback
The v2ray/postgres/sftpgo exporters were bound to 127.0.0.1 but VictoriaMetrics
scraped them via the host's routable IP, so those scrapes refused connections.
Point the three same-host scrape targets at 127.0.0.1 (VM runs on the same host).
node-exporter keeps the host IP since it binds 0.0.0.0.
* Revert "security: add conservative systemd hardening to aquamarine services"
This reverts commit f9cf99dadb.
Setting nix.settings.sandbox-paths replaces Nix compiled sandbox defaults. Keeping only /dev/net made local builds lose the sandbox shell that normally provides /bin/sh, which broke upstream scripts with /bin/sh shebangs in packages such as X11-fonts and openldap.
Use extra-sandbox-paths for /dev/net instead. This keeps Nix daemon defaults intact, including the configured sandbox shell, while still allowing builds that need /dev/net.
Add eval coverage for both Linux output sets so future changes keep /dev/net in extra-sandbox-paths and do not reintroduce an explicit sandbox-paths override.