cleanup

fix: install sh
feat: add interactive CLI setup wizard for server configuration and log file registration
2026-06-26 10:17:45 +02:00 · 2026-06-25 20:42:05 +02:00 · 2026-06-25 20:41:52 +02:00 · 2026-06-25 18:34:34 +02:00
10 changed files with 510 additions and 53 deletions
@@ -1,15 +0,0 @@
-{
-  "permissions": {
-    "allow": [
-      "Bash(go get *)",
-      "Bash(go build *)",
-      "Bash(go vet *)",
-      "Read(//usr/lib/**)",
-      "Read(//proc/**)",
-      "Bash(systemctl show *)",
-      "Bash(echo \"exit=$?\")",
-      "Bash(systemctl list-units *)",
-      "Bash(go test *)"
-    ]
-  }
-}
@@ -13,4 +13,5 @@ config.yml

 CLAUDE.md

-minisign.key
+minisign.key
+./.claude
@@ -19,7 +19,7 @@ Functionality is organized into **modules**. Each module owns a slice of the
 API and declares its own permission vocabulary.

 - **System** - Dashboard overview (OS/kernel, CPU, memory, disks, load, uptime,
-  network interfaces, temperatures); get/set hostname; time, timezone, and NTP;
+  network interfaces, GPU, temperatures); get/set hostname; time, timezone, and NTP;
  locale and console keymap; reboot and power off.
 - **Services** - List and inspect systemd units; start / stop / restart / enable
  / disable; read service logs from the journal or an allowlisted file, as a
@@ -29,10 +29,11 @@ API and declares its own permission vocabulary.
 - **Groups** - List, inspect, create, and delete local groups.
 - **Packages** - List installed packages and available updates; install, remove,
  and upgrade - streamed live over SSE. Auto-detects `dnf`, `apt`, or `pacman`.
- **Networking** - List network interfaces, routing tables, and DNS settings; configure IPv4 settings with temporary applying and safety auto-rollback; bring interfaces up or down.
+- **Networking** - List network interfaces, routing tables, and DNS settings; configure IPv4 settings with temporary applying and safety auto-rollback; bring interfaces up or down; edit `/etc/hosts`.
+- **Storage** - List active mounts and `/etc/fstab` entries; add, edit, and delete fstab entries; mount and unmount filesystems.
 - **Audit** - Read-only trail of every privileged write (who, what, when, result).
 - **Meta** - Self-description for clients: `/api/_modules`, `/api/whoami`,
-  `/api/health`.
+  `/api/health`; trigger a self-update via `POST /api/update`.

 ### Security model at a glance

@@ -159,14 +160,18 @@ assigns the admin role to the installing user.

 | Command                                          | Effect                                                                      |
 | ------------------------------------------------ | --------------------------------------------------------------------------- |
-| `nadir [run] [-d]`                               | Start the server. `-d` / `--detach` runs it in the background.              |
-| `nadir --save-config`                            | Save the default configuration template to the target path and exit.        |
-| `nadir install`                                  | Install + enable the systemd service (starts now and on boot).              |
-| `nadir uninstall`                                | Stop, disable, and remove the systemd service.                              |
-| `nadir start` \| `stop` \| `restart` \| `status` | Control the running service.                                                |
-| `nadir enable` \| `disable`                      | Toggle start-on-boot without removing the unit.                             |
-| `nadir logs`                                     | Follow logs - journald if installed as a service, otherwise the detach log. |
-| `nadir help`                                     | Show usage.                                                                 |
+| `nadir [run] [-d]`                               | Start the server. `-d` / `--detach` runs it in the background.                                            |
+| `nadir --save-config`                            | Save the default configuration template to the target path and exit.                                      |
+| `nadir install`                                  | Install + enable the systemd service (starts now and on boot).                                            |
+| `nadir uninstall`                                | Stop, disable, and remove the systemd service.                                                            |
+| `nadir start` \| `stop` \| `restart` \| `status` | Control the running service.                                                                             |
+| `nadir enable` \| `disable`                      | Toggle start-on-boot without removing the unit.                                                           |
+| `nadir logs`                                     | Follow logs - journald if installed as a service, otherwise the detach log.                               |
+| `nadir update [--check] [--force]`               | Download and install the latest release (requires `server.release_repo` in config). `--check` reports the available version without downloading; `--force` re-downloads even when already current. |
+| `nadir token add <name>`                         | Mint a machine API token (shown once, not stored in plain text).                                          |
+| `nadir token rm <name>`                          | Revoke a token immediately (no restart needed).                                                           |
+| `nadir token ls`                                 | List token names (not the raw keys).                                                                      |
+| `nadir help`                                     | Show usage.                                                                                               |

 Most commands need root.

@@ -187,6 +192,7 @@ server:
  # tls_key:  /etc/nadir/tls/key.pem
  hostname: 100.64.0.189
  port: 9999
+  # release_repo: https://gitea.example.com/owner/nadir  # enables `nadir update`

 # Quote "*" - bare * is YAML alias syntax and fails to parse.
 roles:
@@ -216,6 +222,7 @@ log_files:
 | `tls_cert`, `tls_key` | -       | PEM paths. When both are set (and `trust_proxy` is off), nadir terminates TLS with this pair.                                                                         |
 | `hostname`            | -       | Address to bind. Use `127.0.0.1` for local-only, or an overlay/VPN address to expose nadir only on that interface.                                                    |
 | `port`                | -       | TCP port to listen on.                                                                                                                                                |
+| `release_repo`        | -       | Gitea repo URL (`https://host/owner/repo`). When set, enables `nadir update` and `POST /api/update`. Must be `https://`.                                              |

 TLS selection is covered in [Deployment note 2](#2-tls-three-modes).

@@ -389,7 +396,27 @@ forwarded headers are trustworthy. Without step 1 you're trusting every peer on
 the overlay - fine for a single-tenant network you fully control, risky on a
 shared one.

-### 4. Connecting a dashboard (machine clients)
+### 4. Self-update
+
+When `server.release_repo` points at a Gitea repo, nadir can update itself:
+
+```bash
+sudo nadir update           # download + install latest, restart service
+sudo nadir update --check   # report available version, do nothing
+sudo nadir update --force   # re-download even if already at latest
+```
+
+The updater:
+1. Fetches the latest release from the Gitea API.
+2. Downloads the binary for the host's architecture (`linux-amd64`, `linux-arm64`, …).
+3. Verifies the release: checks the minisign signature on `sha256sums.txt`, then checks the binary's SHA-256 against it. Refuses to install if either check fails.
+4. Atomically replaces the running binary (`os.Rename` on the same filesystem) and runs `systemctl restart nadir`.
+
+The same flow is also reachable via `POST /api/update` (requires the admin wildcard role), which runs the updater detached and returns 202 immediately. Poll `GET /api/health` to confirm the new version is running after the restart drops in-flight connections.
+
+`release_repo` must use `https://` — the update downloads and executes the binary, and a plaintext URL would expose the host to on-path replacement.
+
+### 5. Connecting a dashboard (machine clients)

 To manage one or more Nadir instances via a central dashboard or non-interactive client, authenticate requests using a static Bearer token rather than interactive PAM credentials.

@@ -439,22 +466,23 @@ To connect a browser-based dashboard hosted on a different origin, choose one of
 ## Layout

 ```
-cmd/                  process entry point + CLI (run / install / logs …), TLS, service wiring
-internal/auth         PAM auth, sessions, login/logout, login throttle, PAM service install
+cmd/                  process entry point + CLI (run / install / update / token / logs …), TLS, service wiring
+internal/auth         PAM auth, sessions, login/logout, login throttle, bearer tokens, PAM service install
+internal/auditlog     SQLite-backed audit log writer
 internal/config       config.yaml loader + startup validation
-internal/meta         /api/_modules, /api/whoami, /api/health discovery endpoints
+internal/meta         /api/_modules, /api/whoami, /api/health, /api/update discovery + update endpoints
 internal/module       the Module interface
 internal/modules      concrete modules:
-                        system   - info, hostname, time/timezone/NTP, locale/keymap, power
-                        services - systemd unit control + journal/file logs (snapshot + SSE)
-                        users    - local accounts
-                        groups   - local groups
-                        packages - dnf/apt/pacman install/remove/upgrade (streamed)
-                        audit    - read-only audit trail
-                        networking - network interfaces, routing tables, DNS, and IP configurations
+                        system     - info, hostname, time/timezone/NTP, locale/keymap, power
+                        services   - systemd unit control + journal/file logs (snapshot + SSE)
+                        users      - local accounts
+                        groups     - local groups
+                        packages   - dnf/apt/pacman install/remove/upgrade (streamed)
+                        networking - interfaces, routing tables, DNS, IP config, /etc/hosts
+                        storage    - active mounts, /etc/fstab read/write, mount/unmount
+internal/mounts       /proc/mounts parser (used by storage module)
 internal/oscmd        shared command runner (timeouts, stderr surfacing) + helpers
 internal/rbac         roles, permissions ("*" wildcards), HTTP middleware (RBAC + CSRF)
-internal/audit        SQLite-backed audit log writer
 ```

 ## API docs
@@ -1,6 +1,7 @@
 package main

 import (
+	"bufio"
 	"flag"
 	"fmt"
 	"io"
@@ -8,8 +9,11 @@ import (
 	"os/exec"
 	"path/filepath"
 	"strconv"
+	"strings"
 	"syscall"

+	"github.com/mattn/go-isatty"
+
 	"nadir/internal/auth"
 	"nadir/internal/config"
 )
@@ -137,6 +141,125 @@ func installService(args []string) error {
 	isUnsecure := *unsecureOpt || optCount == 0
 	isTrustProxy := *trustProxyOpt

+	cfgPath, err := resolveConfigPath()
+	if err != nil {
+		return err
+	}
+
+	shouldWriteConfig := false
+	if _, err := os.Stat(cfgPath); os.IsNotExist(err) {
+		shouldWriteConfig = true
+	}
+
+	username := getUsername()
+	var logFiles map[string][]string
+
+	if fs.NFlag() == 0 && (isatty.IsTerminal(os.Stdin.Fd()) || isatty.IsCygwinTerminal(os.Stdin.Fd())) {
+		reader := bufio.NewReader(os.Stdin)
+
+		if !shouldWriteConfig {
+			fmt.Printf("Configuration file already exists at %s. Overwrite? [y/N] (default n): ", cfgPath)
+			overwriteInput, _ := reader.ReadString('\n')
+			overwriteInput = strings.ToLower(strings.TrimSpace(overwriteInput))
+			if overwriteInput != "y" && overwriteInput != "yes" {
+				fmt.Println("Keeping existing configuration. Proceeding with installation...")
+				if existingCfg, loadErr := config.Load(cfgPath); loadErr == nil {
+					*hostnameOpt = existingCfg.Server.Hostname
+					if p, err := strconv.Atoi(existingCfg.Server.Port); err == nil {
+						*portOpt = p
+					}
+					isTLS = existingCfg.Server.TLSCert != "" && existingCfg.Server.TLSKey != ""
+					isTrustProxy = existingCfg.Server.TrustProxy
+					isUnsecure = !isTLS && !isTrustProxy
+				}
+				goto skipConfigPrompt
+			}
+			shouldWriteConfig = true
+		}
+
+		fmt.Println("Configuring Nadir installation:")
+		fmt.Println("  1) Serve plaintext HTTP directly (unsecure) [default]")
+		fmt.Println("  2) Generate persistent self-signed TLS cert/key and enable HTTPS (tls)")
+		fmt.Println("  3) Serve plaintext HTTP behind a trusted TLS-terminating reverse proxy (trust-proxy)")
+		fmt.Print("Enter choice [1-3] (default 1): ")
+		choice, _ := reader.ReadString('\n')
+		choice = strings.TrimSpace(choice)
+		if choice == "" || choice == "1" {
+			isUnsecure = true
+			isTLS = false
+			isTrustProxy = false
+		} else if choice == "2" {
+			isTLS = true
+			isUnsecure = false
+			isTrustProxy = false
+		} else if choice == "3" {
+			isTrustProxy = true
+			isTLS = false
+			isUnsecure = false
+		} else {
+			return fmt.Errorf("invalid choice: %q", choice)
+		}
+
+		fmt.Printf("Enter hostname to bind to (default %s): ", *hostnameOpt)
+		hostChoice, _ := reader.ReadString('\n')
+		hostChoice = strings.TrimSpace(hostChoice)
+		if hostChoice != "" {
+			*hostnameOpt = hostChoice
+		}
+
+		fmt.Printf("Enter port to bind to (default %d): ", *portOpt)
+		portChoice, _ := reader.ReadString('\n')
+		portChoice = strings.TrimSpace(portChoice)
+		if portChoice != "" {
+			p, err := strconv.Atoi(portChoice)
+			if err != nil || p <= 0 || p > 65535 {
+				return fmt.Errorf("invalid port: %q", portChoice)
+			}
+			*portOpt = p
+		}
+
+		fmt.Printf("Enter main admin username (default %s): ", username)
+		userChoice, _ := reader.ReadString('\n')
+		userChoice = strings.TrimSpace(userChoice)
+		if userChoice != "" {
+			username = userChoice
+		}
+
+		fmt.Print("Would you like to expose any log files to the Nadir UI? [y/N] (default n): ")
+		logInput, _ := reader.ReadString('\n')
+		logInput = strings.ToLower(strings.TrimSpace(logInput))
+		if logInput == "y" || logInput == "yes" {
+			logFiles = make(map[string][]string)
+			for {
+				fmt.Print("  Enter service/unit name (e.g. nginx): ")
+				unit, _ := reader.ReadString('\n')
+				unit = strings.TrimSpace(unit)
+				if unit == "" {
+					fmt.Println("  Service name cannot be empty. Skipping.")
+					continue
+				}
+
+				fmt.Printf("  Enter absolute path to log file for %s: ", unit)
+				path, _ := reader.ReadString('\n')
+				path = strings.TrimSpace(path)
+				if path == "" {
+					fmt.Println("  Path cannot be empty. Skipping.")
+					continue
+				}
+
+				logFiles[unit] = append(logFiles[unit], path)
+
+				fmt.Print("  Add another log file? [y/N] (default n): ")
+				another, _ := reader.ReadString('\n')
+				another = strings.ToLower(strings.TrimSpace(another))
+				if another != "y" && another != "yes" {
+					break
+				}
+			}
+		}
+	}
+
+skipConfigPrompt:
 	// Provision the PAM service the server authenticates against, so it exists
 	// before the unit starts rather than appearing on first login. Idempotent:
 	// EnsurePAMService leaves an existing /etc/pam.d/nadir untouched. runServer
@@ -181,11 +304,6 @@ func installService(args []string) error {
 		}
 	}

-	cfgPath, err := resolveConfigPath()
-	if err != nil {
-		return err
-	}
-
 	// Construct configuration template content based on installation options
 	secureTLSVal := "true"
 	trustProxyLine := "# trust_proxy: false"
@@ -204,11 +322,21 @@ func installService(args []string) error {
 		keyLine = "# tls_key:  /var/lib/nadir/tls/key.pem"
 	}

-	username := getUsername()
 	configContent := fmt.Sprintf(configTemplateBase, secureTLSVal, trustProxyLine, certLine, keyLine, *hostnameOpt, *portOpt, username)
+	if len(logFiles) > 0 {
+		var logFilesSection strings.Builder
+		logFilesSection.WriteString("\nlog_files:\n")
+		for unit, paths := range logFiles {
+			logFilesSection.WriteString(fmt.Sprintf("  %s:\n", unit))
+			for _, path := range paths {
+				logFilesSection.WriteString(fmt.Sprintf("    - %s\n", path))
+			}
+		}
+		configContent += logFilesSection.String()
+	}

-	// Ensure default config file exists
-	if _, err := os.Stat(cfgPath); os.IsNotExist(err) {
+	// Ensure default config file exists or we explicitly overwrote it
+	if shouldWriteConfig {
 		if err := saveDefaultConfig(cfgPath, configContent); err != nil {
 			return err
 		}
@@ -16,7 +16,7 @@ require (
 	github.com/dustin/go-humanize v1.0.1 // indirect
 	github.com/google/uuid v1.6.0 // indirect
 	github.com/jedisct1/go-minisign v0.0.0-20260527172527-a09352b57a22
-	github.com/mattn/go-isatty v0.0.21 // indirect
+	github.com/mattn/go-isatty v0.0.21
 	github.com/ncruces/go-strftime v1.0.0 // indirect
 	github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec // indirect
 	golang.org/x/crypto v0.52.0 // indirect
@@ -99,7 +99,11 @@ do_install() {

  echo "binary installed at /usr/local/bin/nadir"
  echo "installing as a systemd service ..."
-  /usr/local/bin/nadir install
+  if [ -c /dev/tty ]; then
+    /usr/local/bin/nadir install < /dev/tty
+  else
+    /usr/local/bin/nadir install
+  fi

  echo
  echo "done. check status with: nadir status"
@@ -0,0 +1,187 @@
+package system
+
+import (
+	"os"
+	"path/filepath"
+	"strconv"
+	"strings"
+
+	"nadir/internal/oscmd"
+)
+
+type GPUInfo struct {
+	Model             string  `json:"model" example:"AMD Radeon RX 7900 XTX" doc:"GPU model name, or vendor:device hex ID if lspci unavailable"`
+	Vendor            string  `json:"vendor" example:"1002" doc:"PCI vendor ID (hex)"`
+	DeviceID          string  `json:"device_id" example:"744c" doc:"PCI device ID (hex)"`
+	Driver            string  `json:"driver" example:"amdgpu" doc:"Kernel driver in use"`
+	MemoryTotalBytes  uint64  `json:"memory_total_bytes" example:"8589934592" doc:"Total VRAM in bytes (driver-dependent; 0 if unavailable)"`
+	MemoryUsedBytes   uint64  `json:"memory_used_bytes" example:"27267072" doc:"Used VRAM in bytes (driver-dependent; 0 if unavailable)"`
+	UtilizationPct    float64 `json:"utilization_pct" example:"23.5" doc:"GPU compute utilization percentage (driver-dependent; 0 if unavailable)"`
+	MemUtilizationPct float64 `json:"mem_utilization_pct" example:"15.0" doc:"GPU memory controller utilization percentage (driver-dependent; 0 if unavailable)"`
+}
+
+func gpuInfo() []GPUInfo {
+	return readGPUsFromSysfs("/sys/class/drm")
+}
+
+func readGPUsFromSysfs(drmRoot string) []GPUInfo {
+	entries, err := os.ReadDir(drmRoot)
+	if err != nil {
+		return nil
+	}
+
+	seen := map[string]bool{}
+	var gpus []GPUInfo
+
+	for _, e := range entries {
+		if !isGPUCard(e.Name()) {
+			continue
+		}
+
+		pciAddr, vendor, device, driver := readGPUFromCard(drmRoot, e.Name())
+		if pciAddr == "" || seen[pciAddr] {
+			continue
+		}
+		seen[pciAddr] = true
+
+		model := vendor + ":" + device
+		if m := lookupGPUName(pciAddr); m != "" {
+			model = m
+		}
+
+		gpu := GPUInfo{
+			Model:    model,
+			Vendor:   vendor,
+			DeviceID: device,
+			Driver:   driver,
+		}
+
+		devPath := filepath.Join(drmRoot, e.Name(), "device")
+		enrichGPUInfo(&gpu, devPath, pciAddr, driver)
+		gpus = append(gpus, gpu)
+	}
+
+	return gpus
+}
+
+func enrichGPUInfo(gpu *GPUInfo, devPath, pciAddr, driver string) {
+	switch driver {
+	case "amdgpu":
+		enrichAMDGPU(gpu, devPath)
+	case "nvidia":
+		enrichNvidiaGPU(gpu, pciAddr)
+	}
+}
+
+func enrichAMDGPU(gpu *GPUInfo, devPath string) {
+	if total := readUint64FromFile(filepath.Join(devPath, "mem_info_vram_total")); total > 0 {
+		gpu.MemoryTotalBytes = total
+		gpu.MemoryUsedBytes = readUint64FromFile(filepath.Join(devPath, "mem_info_vram_used"))
+	}
+	if pct := readIntFromFile(filepath.Join(devPath, "gpu_busy_percent")); pct >= 0 {
+		gpu.UtilizationPct = float64(pct)
+	}
+	if pct := readIntFromFile(filepath.Join(devPath, "mem_busy_percent")); pct >= 0 {
+		gpu.MemUtilizationPct = float64(pct)
+	}
+}
+
+func enrichNvidiaGPU(gpu *GPUInfo, pciAddr string) {
+	out, err := oscmd.Run("nvidia-smi",
+		"-i", pciAddr,
+		"--query-gpu=memory.total,memory.used,utilization.gpu,utilization.memory",
+		"--format=csv,noheader,nounits",
+	)
+	if err != nil {
+		return
+	}
+	parts := strings.Split(strings.TrimSpace(out), ", ")
+	if len(parts) < 4 {
+		return
+	}
+	if total, err := strconv.ParseUint(parts[0], 10, 64); err == nil && total > 0 {
+		gpu.MemoryTotalBytes = total * 1024 * 1024
+		if used, err := strconv.ParseUint(parts[1], 10, 64); err == nil {
+			gpu.MemoryUsedBytes = used * 1024 * 1024
+		}
+	}
+	if pct, err := strconv.ParseFloat(parts[2], 64); err == nil {
+		gpu.UtilizationPct = pct
+	}
+	if pct, err := strconv.ParseFloat(parts[3], 64); err == nil {
+		gpu.MemUtilizationPct = pct
+	}
+}
+
+func isGPUCard(name string) bool {
+	if !strings.HasPrefix(name, "card") {
+		return false
+	}
+	if len(name) == 4 {
+		return false
+	}
+	ch := name[4]
+	return ch >= '0' && ch <= '9'
+}
+
+func readGPUFromCard(drmRoot, name string) (pciAddr, vendor, device, driver string) {
+	cardPath := filepath.Join(drmRoot, name)
+	devPath := filepath.Join(cardPath, "device")
+
+	resolved, err := filepath.EvalSymlinks(cardPath)
+	if err != nil {
+		return "", "", "", ""
+	}
+
+	parts := strings.Split(resolved, "/")
+	for i, p := range parts {
+		if p == "drm" && i > 0 {
+			pciAddr = parts[i-1]
+			break
+		}
+	}
+	if pciAddr == "" {
+		return "", "", "", ""
+	}
+
+	vendor = strings.TrimPrefix(readTrim(filepath.Join(devPath, "vendor")), "0x")
+	device = strings.TrimPrefix(readTrim(filepath.Join(devPath, "device")), "0x")
+	driver = readDriver(devPath)
+	return
+}
+
+func readDriver(devPath string) string {
+	target, err := os.Readlink(filepath.Join(devPath, "driver"))
+	if err != nil {
+		return ""
+	}
+	return filepath.Base(target)
+}
+
+func lookupGPUName(pciAddr string) string {
+	out, err := oscmd.Run("lspci", "-nns", pciAddr)
+	if err != nil {
+		return ""
+	}
+	_, rest, ok := strings.Cut(out, " ")
+	if !ok {
+		return ""
+	}
+	return strings.TrimSpace(rest)
+}
+
+func readUint64FromFile(path string) uint64 {
+	v, err := strconv.ParseUint(readTrim(path), 10, 64)
+	if err != nil {
+		return 0
+	}
+	return v
+}
+
+func readIntFromFile(path string) int {
+	v, err := strconv.Atoi(readTrim(path))
+	if err != nil {
+		return -1
+	}
+	return v
+}
@@ -11,7 +11,7 @@ import (
 )

 // SystemInfoBody is the dashboard overview: OS identity plus live CPU, memory,
-// disk, load, network, and temperature readings. Every section is best-effort —
+// disk, load, network, GPU, and temperature readings. Every section is best-effort —
 // a source that's unavailable (e.g. no thermal zones in a VM) yields a zero
 // value or empty list rather than failing the whole call.
 type SystemInfoBody struct {
@@ -24,6 +24,7 @@ type SystemInfoBody struct {
 	Disks             []DiskInfo     `json:"disks" doc:"Mounted block-device filesystems"`
 	NetworkInterfaces []NetInterface `json:"network_interfaces" doc:"Network interfaces and their addresses"`
 	Temperatures      []Temperature  `json:"temperatures" doc:"Thermal sensor readings in Celsius"`
+	GPUs              []GPUInfo      `json:"gpus" doc:"Graphics processors detected via DRM sysfs"`
 }

 type GetInfoOutput struct{ Body SystemInfoBody }
@@ -36,9 +37,9 @@ func registerInfo(api huma.API, sampler *Sampler) {
 		Summary:     "Get system information",
 		Description: "Returns an overview for a dashboard: OS/kernel identity, CPU, " +
 			"memory and swap, mounted disks, load averages, uptime, network " +
-			"interfaces, and temperatures. All values come from cheap local reads " +
-			"(/proc, /sys, syscalls) with no D-Bus dependency; each section is " +
-			"best-effort.",
+			"interfaces, temperatures, and GPU information. All values come from cheap " +
+			"local reads (/proc, /sys, syscalls) with no D-Bus dependency; each " +
+			"section is best-effort.",
 		Tags:     []string{tagSystem},
 		Metadata: op("read"),
 		Errors:   readErrors,
@@ -54,6 +55,7 @@ func registerInfo(api huma.API, sampler *Sampler) {
 			Disks:             diskInfo(),
 			NetworkInterfaces: netInfo(),
 			Temperatures:      tempInfo(),
+			GPUs:              gpuInfo(),
 		}}, nil
 	})
 }
@@ -6,6 +6,106 @@ import (
 	"testing"
 )

+func TestReadGPUsFromSysfs(t *testing.T) {
+	root := t.TempDir()
+
+	// card0 — AMD GPU with VRAM and utilization files
+	pciDev := filepath.Join(root, "devices/pci0000:00/0000:00:02.0")
+	mkdirAll(t, pciDev)
+	write(t, pciDev, ".", "vendor", "0x1002")
+	write(t, pciDev, ".", "device", "0x7480")
+	write(t, pciDev, ".", "mem_info_vram_total", "8573157376")
+	write(t, pciDev, ".", "mem_info_vram_used", "27267072")
+	write(t, pciDev, ".", "gpu_busy_percent", "23")
+	write(t, pciDev, ".", "mem_busy_percent", "15")
+
+	driverDir := filepath.Join(root, "bus/pci/drivers/amdgpu")
+	mkdirAll(t, driverDir)
+	mustSymlink(t, driverDir, filepath.Join(pciDev, "driver"))
+
+	cardTarget := filepath.Join(pciDev, "drm", "card0")
+	mkdirAll(t, cardTarget)
+	mustSymlink(t, pciDev, filepath.Join(cardTarget, "device"))
+	mustSymlink(t, cardTarget, filepath.Join(root, "card0"))
+
+	// Distractors
+	write(t, root, ".", "card0-HDMI-1", "distract")
+	write(t, root, ".", "renderD128", "distract")
+
+	gpus := readGPUsFromSysfs(root)
+	if len(gpus) != 1 {
+		t.Fatalf("want 1 GPU, got %d: %+v", len(gpus), gpus)
+	}
+	if gpus[0].Vendor != "1002" {
+		t.Errorf("vendor = %q, want 1002", gpus[0].Vendor)
+	}
+	if gpus[0].DeviceID != "7480" {
+		t.Errorf("device_id = %q, want 7480", gpus[0].DeviceID)
+	}
+	if gpus[0].Driver != "amdgpu" {
+		t.Errorf("driver = %q, want amdgpu", gpus[0].Driver)
+	}
+	if gpus[0].MemoryTotalBytes != 8573157376 {
+		t.Errorf("MemoryTotalBytes = %d, want 8573157376", gpus[0].MemoryTotalBytes)
+	}
+	if gpus[0].MemoryUsedBytes != 27267072 {
+		t.Errorf("MemoryUsedBytes = %d, want 27267072", gpus[0].MemoryUsedBytes)
+	}
+	if gpus[0].UtilizationPct != 23.0 {
+		t.Errorf("UtilizationPct = %f, want 23.0", gpus[0].UtilizationPct)
+	}
+	if gpus[0].MemUtilizationPct != 15.0 {
+		t.Errorf("MemUtilizationPct = %f, want 15.0", gpus[0].MemUtilizationPct)
+	}
+}
+
+func TestReadGPUsFromSysfsNoEnrichment(t *testing.T) {
+	root := t.TempDir()
+
+	// i915 GPU with no VRAM or utilization files
+	pciDev := filepath.Join(root, "devices/pci0000:00/0000:00:02.0")
+	mkdirAll(t, pciDev)
+	write(t, pciDev, ".", "vendor", "0x8086")
+	write(t, pciDev, ".", "device", "0x46a6")
+
+	driverDir := filepath.Join(root, "bus/pci/drivers/i915")
+	mkdirAll(t, driverDir)
+	mustSymlink(t, driverDir, filepath.Join(pciDev, "driver"))
+
+	cardTarget := filepath.Join(pciDev, "drm", "card0")
+	mkdirAll(t, cardTarget)
+	mustSymlink(t, pciDev, filepath.Join(cardTarget, "device"))
+	mustSymlink(t, cardTarget, filepath.Join(root, "card0"))
+
+	gpus := readGPUsFromSysfs(root)
+	if len(gpus) != 1 {
+		t.Fatalf("want 1 GPU, got %d", len(gpus))
+	}
+	if gpus[0].MemoryTotalBytes != 0 || gpus[0].MemoryUsedBytes != 0 {
+		t.Errorf("expected 0 VRAM for i915, got total=%d used=%d", gpus[0].MemoryTotalBytes, gpus[0].MemoryUsedBytes)
+	}
+	if gpus[0].UtilizationPct != 0 || gpus[0].MemUtilizationPct != 0 {
+		t.Errorf("expected 0 utilization for i915, got gpu=%f mem=%f", gpus[0].UtilizationPct, gpus[0].MemUtilizationPct)
+	}
+}
+
+func TestReadGPUsFromSysfsMissingDir(t *testing.T) {
+	gpus := readGPUsFromSysfs("/nonexistent/drm")
+	if gpus != nil {
+		t.Errorf("expected nil, got %+v", gpus)
+	}
+}
+
+func TestReadGPUsFromSysfsSkipsNonGPU(t *testing.T) {
+	root := t.TempDir()
+	write(t, root, ".", "renderD128", "x")
+	write(t, root, ".", "card0-HDMI-1", "x")
+	gpus := readGPUsFromSysfs(root)
+	if len(gpus) != 0 {
+		t.Errorf("expected 0 GPUs, got %d", len(gpus))
+	}
+}
+
 func TestReadHwmonTemps(t *testing.T) {
 	root := t.TempDir()
 	// k10temp: CPU, labelled Tctl.
@@ -33,6 +133,20 @@ func TestReadHwmonTemps(t *testing.T) {
 	}
 }

+func mkdirAll(t *testing.T, path string) {
+	t.Helper()
+	if err := os.MkdirAll(path, 0o755); err != nil {
+		t.Fatal(err)
+	}
+}
+
+func mustSymlink(t *testing.T, target, link string) {
+	t.Helper()
+	if err := os.Symlink(target, link); err != nil {
+		t.Fatal(err)
+	}
+}
+
 func write(t *testing.T, root, chip, file, val string) {
 	t.Helper()
 	dir := filepath.Join(root, chip)
@@ -34,6 +34,14 @@ func TestSystemHandlers(t *testing.T) {
 		}
 		return oscmd.MockCommand{ExitCode: 1}
 	})
+	// Mock lspci to prevent real calls in case the test host has GPUs.
+	oscmd.SetMock("lspci", func(args []string) oscmd.MockCommand {
+		return oscmd.MockCommand{ExitCode: 1}
+	})
+	// Mock nvidia-smi: return failure so enrichment is a no-op.
+	oscmd.SetMock("nvidia-smi", func(args []string) oscmd.MockCommand {
+		return oscmd.MockCommand{ExitCode: 1}
+	})
 	defer oscmd.ClearMocks()

 	// 1. Test GET /api/system/info
Author	SHA1	Message	Date
urania	bb5cc8268f	cleanup build-and-release / release (push) Successful in 1m10s Details	2026-06-26 10:17:45 +02:00
urania	cad6c1f421	fix: install sh build-and-release / release (push) Successful in 2m32s Details	2026-06-25 20:42:05 +02:00
urania	22e6812d4b	feat: add interactive CLI setup wizard for server configuration and log file registration	2026-06-25 20:41:52 +02:00
urania	0415e905af	Feat: GPU detection via DRM sysfs, integrated into system info endpoint build-and-release / release (push) Successful in 2m34s Details Adds GPUInfo struct and readGPUsFromSysfs parsing DRM card entries (/sys/class/drm/card*). Supports: - AMD GPUs (amdgpu driver): VRAM totals/utilization from sysfs files - NVIDIA GPUs: enrichment via nvidia-smi query - Intel/other: basic PCI vendor/device/driver identification Includes full test coverage for AMD enrichment, i915 fallback, missing sysfs dir, and non-GPU DRM entry filtering.	2026-06-25 18:34:34 +02:00