Pin compose example to v0.3.1

Extends the existing per-tick nvidia-smi query with temperature.gpu and
fan.speed (no extra call); N/A values (cards without fan telemetry) are
simply omitted from the GPU line.

cmd/gpu-turnstile/main.go

@@ -813,7 +813,7 @@ func run(ctx context.Context, cfg config.Config, log *slog.Logger, logOut io.Wri
 	// AUTO_UPDATE.
 	if isService {
 		serveControl(ctx, log, u, exePath, applyStaged,
-			statusProvider(cfg, lk, comfySup, health, started, gw),
+			statusProvider(cfg, lk, comfySup, health, started, gw, ollamaClient),
 			reloadHandler(cfg, configPath, restartWhenIdle))
 	}
 
@@ -847,14 +847,17 @@ type gpuWatch struct {
 	enabled bool
 	usedMB  int
 	total   int
+	tempC   int
+	fanPct  int
 	known   bool
 	foreign string // last Check result: external holders, "" when none
 	at      time.Time
 }
 
-func (g *gpuWatch) setVRAM(used, total int) {
+func (g *gpuWatch) setVRAM(st game.GPUStats) {
 	g.mu.Lock()
-	g.usedMB, g.total, g.known = used, total, true
+	g.usedMB, g.total = st.UsedMB, st.TotalMB
+	g.tempC, g.fanPct, g.known = st.TempC, st.FanPct, true
 	g.mu.Unlock()
 }
 
@@ -864,14 +867,15 @@ func (g *gpuWatch) setCheck(foreign string) {
 	g.mu.Unlock()
 }
 
-func (g *gpuWatch) get() (used, total int, known bool, foreign string, ageS int64) {
+func (g *gpuWatch) get() (st game.GPUStats, known bool, foreign string, ageS int64) {
 	g.mu.Lock()
 	defer g.mu.Unlock()
 	ageS = -1
 	if !g.at.IsZero() {
 		ageS = int64(time.Since(g.at).Seconds())
 	}
-	return g.usedMB, g.total, g.known, g.foreign, ageS
+	return game.GPUStats{UsedMB: g.usedMB, TotalMB: g.total, TempC: g.tempC, FanPct: g.fanPct},
+		g.known, g.foreign, ageS
 }
 
 // gameLoop polls for foreign GPU holders (a game, another ML job). While one
@@ -894,8 +898,8 @@ func gameLoop(ctx context.Context, cfg config.Config, log *slog.Logger, det *gam
 			log.Warn("game detection failed", "err", err)
 		}
 		gw.setCheck(summarizeHolders(holders))
-		if used, total, verr := game.QueryVRAMMB(ctx); verr == nil {
-			gw.setVRAM(used, total)
+		if st, verr := game.QueryGPUStats(ctx); verr == nil {
+			gw.setVRAM(st)
 		}
 		switch {
 		case len(holders) > 0 && !held:
@@ -1106,6 +1110,16 @@ type statusDownstream struct {
 	URL     string `json:"url"`
 	Up      bool   `json:"up"`
 	Managed string `json:"managed,omitempty"`
+	// Models lists Ollama's loaded models with their VRAM footprint. Null
+	// when unknown (query failed / not applicable); [] means none loaded —
+	// deliberately no omitempty so the two stay distinguishable.
+	Models []statusModel `json:"models"`
+}
+
+// statusModel is one loaded Ollama model.
+type statusModel struct {
+	Name   string `json:"name"`
+	VRAMMB int64  `json:"vram_mb"` // 0 = resident in RAM, not VRAM
 }
 
 type statusLock struct {
@@ -1138,7 +1152,11 @@ type statusGPU struct {
 	Enabled bool `json:"enabled"`
 	UsedMB  int  `json:"used_mb"`
 	TotalMB int  `json:"total_mb"`
-	Known   bool `json:"known"`
+	// TempC/FanPct are -1 when unknown (never sampled or nvidia-smi
+	// reported N/A).
+	TempC  int  `json:"temp_c"`
+	FanPct int  `json:"fan_pct"`
+	Known  bool `json:"known"`
 	// Foreign is the last detector finding (external GPU holders), empty
 	// when the last check found none.
 	Foreign string `json:"foreign,omitempty"`
@@ -1183,23 +1201,41 @@ func diffConfig(a, b config.Config) []string {
 	return out
 }
 
-// statusProvider assembles the one-line JSON snapshot for CmdStatus.
-func statusProvider(cfg config.Config, lk *lock.Lock, comfySup *supervise.Process, health *healthTracker, started time.Time, gw *gpuWatch) func() string {
+// statusProvider assembles the one-line JSON snapshot for CmdStatus. The
+// loaded-model query to Ollama gets a short timeout so a wedged upstream
+// cannot stall the status channel for long.
+func statusProvider(cfg config.Config, lk *lock.Lock, comfySup *supervise.Process, health *healthTracker, started time.Time, gw *gpuWatch, ollamaClient *ollama.Client) func() string {
 	return func() string {
 		snap := statusSnapshot{
 			Version: version,
 			UptimeS: int64(time.Since(started).Seconds()),
 		}
-		used, total, known, foreign, ageS := gw.get()
+		st, known, foreign, ageS := gw.get()
 		snap.GPU = statusGPU{
 			Enabled: gw.enabled,
-			UsedMB:  used, TotalMB: total, Known: known,
+			UsedMB:  st.UsedMB, TotalMB: st.TotalMB, Known: known,
+			TempC: -1, FanPct: -1,
 			Foreign: foreign, AgeS: ageS,
 		}
+		if known {
+			snap.GPU.TempC, snap.GPU.FanPct = st.TempC, st.FanPct
+		}
 		if cfg.OllamaURL != "" {
-			snap.Downstreams = append(snap.Downstreams, statusDownstream{
+			d := statusDownstream{
 				Name: "ollama", URL: cfg.OllamaURL, Up: health.get("ollama"),
-			})
+			}
+			if ollamaClient != nil {
+				mctx, cancel := context.WithTimeout(context.Background(), 2*time.Second)
+				models, err := ollamaClient.LoadedModelDetails(mctx)
+				cancel()
+				if err == nil {
+					d.Models = make([]statusModel, 0, len(models))
+					for _, m := range models {
+						d.Models = append(d.Models, statusModel{Name: m.Name, VRAMMB: m.SizeVRAM / (1024 * 1024)})
+					}
+				}
+			}
+			snap.Downstreams = append(snap.Downstreams, d)
 		}
 		if cfg.ComfyURL != "" {
 			d := statusDownstream{Name: "comfy", URL: cfg.ComfyURL, Up: health.get("comfy")}
@@ -1208,15 +1244,15 @@ func statusProvider(cfg config.Config, lk *lock.Lock, comfySup *supervise.Proces
 			}
 			snap.Downstreams = append(snap.Downstreams, d)
 		}
-		st := lk.Status()
+		lst := lk.Status()
 		snap.Lock = statusLock{
-			State:       string(st.State),
-			Detail:      st.Detail,
-			LLMInflight: st.LLMInflight,
-			LLMWaiting:  st.LLMWaiting,
-			ImageQueue:  st.ImageQueue,
-			External:    st.External,
-			SinceS:      int64(time.Since(st.Since).Seconds()),
+			State:       string(lst.State),
+			Detail:      lst.Detail,
+			LLMInflight: lst.LLMInflight,
+			LLMWaiting:  lst.LLMWaiting,
+			ImageQueue:  lst.ImageQueue,
+			External:    lst.External,
+			SinceS:      int64(time.Since(lst.Since).Seconds()),
 		}
 		b, err := json.Marshal(snap)
 		if err != nil {

cmd/gpu-turnstile/monitor.go

@@ -22,13 +22,14 @@ const (
 )
 
 // hotkeysLine is the monitor's footer.
-const hotkeysLine = " " + cDim + "q quit · u update now" + cReset + "\x1b[K\n"
+const hotkeysLine = " " + cDim + "q quit · u update now · r reload config" + cReset + "\x1b[K\n"
 
 // monitorCommand renders a live status view of the running service,
 // refreshed every second from the control channel. When the service
 // reports a different version and the executable on disk changed (the
 // updater replaced it), the monitor restarts itself onto the new binary.
-// Hotkeys: q quits, u triggers an update check on the service.
+// Hotkeys: q quits, u triggers an update check on the service, r asks the
+// service to reload its config file.
 func monitorCommand() int {
 	if !stdoutIsTerminal() {
 		fmt.Fprintln(os.Stderr, "gpu-turnstile: --monitor needs an interactive terminal")
@@ -57,7 +58,27 @@ func monitorCommand() int {
 	var note string
 	var noteAt time.Time
 	noteCh := make(chan string, 1)
-	updatePending := false
+	askPending := false
+
+	// ask sends a one-shot command to the service and reports the reply in
+	// the note line. Only one ask runs at a time.
+	ask := func(cmd, busy, label string) {
+		if askPending {
+			return
+		}
+		askPending = true
+		note, noteAt = busy, time.Now()
+		go func() {
+			reply, err := control.Ask(cmd)
+			if err != nil {
+				noteCh <- label + ": no answer from the service"
+				return
+			}
+			msg := strings.TrimPrefix(reply, "OK ")
+			msg = strings.TrimPrefix(msg, "ERR ")
+			noteCh <- label + ": " + msg
+		}()
+	}
 
 	poll := func() string {
 		frame := renderWaiting()
@@ -104,23 +125,12 @@ func monitorCommand() int {
 			case 'q', 'Q', 3: // q or Ctrl+C (raw mode delivers it as a byte)
 				return 0
 			case 'u', 'U':
-				if !updatePending {
-					updatePending = true
-					note, noteAt = "checking for updates…", time.Now()
-					go func() {
-						reply, err := control.Ask(control.CmdUpdateNow)
-						if err != nil {
-							noteCh <- "update: no answer from the service"
-							return
-						}
-						msg := strings.TrimPrefix(reply, "OK ")
-						msg = strings.TrimPrefix(msg, "ERR ")
-						noteCh <- "update: " + msg
-					}()
-				}
+				ask(control.CmdUpdateNow, "checking for updates…", "update")
+			case 'r', 'R':
+				ask(control.CmdReloadEnv, "reloading config…", "reload")
 			}
 		case n := <-noteCh:
-			updatePending = false
+			askPending = false
 			note, noteAt = n, time.Now()
 		}
 	}
@@ -188,7 +198,9 @@ func renderMonitor(snap statusSnapshot, width int) string {
 	b.WriteString(cDim + " " + strings.Repeat("─", width-2) + cReset + "\x1b[K\n")
 
 	for _, d := range snap.Downstreams {
-		b.WriteString(renderDownstream(d) + "\x1b[K\n")
+		busy := (d.Name == "ollama" && snap.Lock.State == "llm" && snap.Lock.LLMInflight > 0) ||
+			(d.Name == "comfy" && snap.Lock.State == "image")
+		b.WriteString(renderDownstream(d, busy) + "\x1b[K\n")
 	}
 	b.WriteString("\x1b[K\n")
 	b.WriteString(renderLock(snap.Lock) + "\x1b[K\n")
@@ -213,6 +225,12 @@ func renderGPU(g statusGPU) string {
 	s := " GPU: "
 	if g.Known {
 		s += renderVRAM(g.UsedMB, g.TotalMB)
+		if g.TempC >= 0 {
+			s += fmt.Sprintf(" · %d°C", g.TempC)
+		}
+		if g.FanPct >= 0 {
+			s += fmt.Sprintf(" · fan %d%%", g.FanPct)
+		}
 	} else {
 		s += cDim + "VRAM unknown (nvidia-smi not answering)" + cReset
 	}
@@ -227,25 +245,46 @@ func renderGPU(g statusGPU) string {
 	return s
 }
 
+// formatMB renders MiB as "4.2 GiB" (or MiB below 1 GiB).
+func formatMB(mb int64) string {
+	if mb >= 1024 {
+		return fmt.Sprintf("%.1f GiB", float64(mb)/1024)
+	}
+	return fmt.Sprintf("%d MiB", mb)
+}
+
 // renderVRAM renders "4.2 / 16.0 GiB used" (or MiB below 1 GiB).
 func renderVRAM(used, total int) string {
-	format := func(mb int) string {
-		if mb >= 1024 {
-			return fmt.Sprintf("%.1f GiB", float64(mb)/1024)
-		}
-		return fmt.Sprintf("%d MiB", mb)
-	}
 	if total > 0 {
-		return format(used) + " / " + format(total) + " used"
+		return formatMB(int64(used)) + " / " + formatMB(int64(total)) + " used"
 	}
-	return format(used) + " used"
+	return formatMB(int64(used)) + " used"
 }
 
 // printableLen counts characters without ANSI escapes (ASCII-only content).
 func printableLen(s string) int { return len(s) }
 
-func renderDownstream(d statusDownstream) string {
+func renderDownstream(d statusDownstream, busy bool) string {
 	url := cDim + d.URL + cReset
+	detail := ""
+	if d.Models != nil {
+		if len(d.Models) == 0 {
+			detail = cDim + " · no models loaded" + cReset
+		} else {
+			parts := make([]string, 0, len(d.Models))
+			for _, m := range d.Models {
+				if m.VRAMMB > 0 {
+					parts = append(parts, fmt.Sprintf("%s (%s VRAM)", m.Name, formatMB(m.VRAMMB)))
+				} else {
+					parts = append(parts, m.Name+" (in RAM)")
+				}
+			}
+			detail = " · " + strings.Join(parts, ", ")
+		}
+	}
+	if busy {
+		detail += " · " + cCyan + "busy" + cReset
+	}
 	switch d.Managed {
 	case "stopped":
 		return fmt.Sprintf(" %s○%s %-8s %sstopped (managed — starts on demand)%s  %s",
@@ -259,7 +298,7 @@ func renderDownstream(d statusDownstream) string {
 		suffix = " (external)"
 	}
 	if d.Up {
-		return fmt.Sprintf(" %s●%s %-8s %sUP%s%s   %s", cGreen, cReset, d.Name, cGreen, cReset, suffix, url)
+		return fmt.Sprintf(" %s●%s %-8s %sUP%s%s%s   %s", cGreen, cReset, d.Name, cGreen, cReset, suffix, detail, url)
 	}
 	return fmt.Sprintf(" %s●%s %-8s %sDOWN%s %s", cRed, cReset, d.Name, cRed, cReset, url)
 }

cmd/gpu-turnstile/monitor_test.go

@@ -32,9 +32,9 @@ func TestRenderMonitor(t *testing.T) {
 		}
 	}
 
-	snap.GPU = statusGPU{Enabled: true, Known: true, UsedMB: 4300, TotalMB: 16384, Foreign: "cyberpunk2077.exe (pid 1234)", AgeS: 12}
+	snap.GPU = statusGPU{Enabled: true, Known: true, UsedMB: 4300, TotalMB: 16384, TempC: 55, FanPct: 42, Foreign: "cyberpunk2077.exe (pid 1234)", AgeS: 12}
 	frame = renderMonitor(snap, 80)
-	for _, want := range []string{"GPU:", "4.2 GiB / 16.0 GiB used", "external:", "checked 12s ago"} {
+	for _, want := range []string{"GPU:", "4.2 GiB / 16.0 GiB used", "55°C", "fan 42%", "external:", "checked 12s ago"} {
 		if !strings.Contains(frame, want) {
 			t.Errorf("frame missing %q:\n%s", want, frame)
 		}
@@ -47,6 +47,25 @@ func TestRenderMonitor(t *testing.T) {
 			t.Errorf("frame missing %q:\n%s", want, frame)
 		}
 	}
+
+	snap.Downstreams[0].Models = []statusModel{{Name: "llama3.1:8b", VRAMMB: 4900}, {Name: "embed", VRAMMB: 0}}
+	snap.Lock = statusLock{State: "llm", LLMInflight: 1}
+	frame = renderMonitor(snap, 80)
+	for _, want := range []string{"llama3.1:8b (4.8 GiB VRAM)", "embed (in RAM)", "busy"} {
+		if !strings.Contains(frame, want) {
+			t.Errorf("frame missing %q:\n%s", want, frame)
+		}
+	}
+
+	snap.Downstreams[0].Models = []statusModel{}
+	snap.Lock = statusLock{State: "idle"}
+	frame = renderMonitor(snap, 80)
+	if !strings.Contains(frame, "no models loaded") {
+		t.Errorf("frame missing %q:\n%s", "no models loaded", frame)
+	}
+	if strings.Contains(frame, "busy") {
+		t.Errorf("idle lock still shows busy:\n%s", frame)
+	}
 }
 
 func TestFmtDur(t *testing.T) {

compose.yaml.example

@@ -6,7 +6,7 @@
 # ComfyUI --listen 0.0.0.0 --port 8189).
 services:
   gpu-turnstile:
-    image: git.rambossek.at/public/gpu-turnstile:v0.2.9
+    image: git.rambossek.at/public/gpu-turnstile:v0.3.1
     restart: unless-stopped
     environment:
       # Each consumer is enabled by setting its URL; leave one unset to

internal/game/game.go

@@ -186,27 +186,54 @@ func queryComputeApps(ctx context.Context) ([]computeApp, error) {
 	return parseComputeApps(string(out))
 }
 
-// QueryVRAMMB returns used and total GPU VRAM in MiB via nvidia-smi.
-// Unlike the per-process list this works under WDDM too.
-func QueryVRAMMB(ctx context.Context) (used, total int, err error) {
+// GPUStats is one nvidia-smi reading of the whole card.
+type GPUStats struct {
+	UsedMB  int
+	TotalMB int
+	TempC   int // -1 when nvidia-smi reports N/A
+	FanPct  int // -1 when N/A (some cards don't expose the fan)
+}
+
+// QueryGPUStats returns VRAM usage, temperature and fan speed via
+// nvidia-smi. Unlike the per-process list this works under WDDM too.
+func QueryGPUStats(ctx context.Context) (GPUStats, error) {
 	out, err := exec.CommandContext(ctx, "nvidia-smi",
-		"--query-gpu=memory.used,memory.total", "--format=csv,noheader,nounits").Output()
+		"--query-gpu=memory.used,memory.total,temperature.gpu,fan.speed", "--format=csv,noheader,nounits").Output()
 	if err != nil {
-		return 0, 0, err
+		return GPUStats{}, err
 	}
-	usedStr, totalStr, ok := strings.Cut(strings.TrimSpace(string(out)), ",")
-	if !ok {
-		return 0, 0, fmt.Errorf("nvidia-smi: unexpected output %q", strings.TrimSpace(string(out)))
+	return parseGPUStats(string(out))
+}
+
+// parseGPUStats parses one "used, total, temp, fan" CSV line (MiB, °C,
+// percent). The memory fields must be numeric; temperature and fan fall
+// back to -1 on "N/A" and friends.
+func parseGPUStats(out string) (GPUStats, error) {
+	fields := strings.Split(strings.TrimSpace(out), ",")
+	if len(fields) != 4 {
+		return GPUStats{}, fmt.Errorf("nvidia-smi: unexpected output %q", strings.TrimSpace(out))
 	}
-	used, err = strconv.Atoi(strings.TrimSpace(usedStr))
-	if err != nil {
-		return 0, 0, fmt.Errorf("nvidia-smi: unexpected used memory in %q", strings.TrimSpace(string(out)))
+	num := func(s string) (int, error) {
+		return strconv.Atoi(strings.TrimSpace(s))
 	}
-	total, err = strconv.Atoi(strings.TrimSpace(totalStr))
-	if err != nil {
-		return 0, 0, fmt.Errorf("nvidia-smi: unexpected total memory in %q", strings.TrimSpace(string(out)))
+	optional := func(s string) int {
+		n, err := num(s)
+		if err != nil {
+			return -1
+		}
+		return n
 	}
-	return used, total, nil
+	var st GPUStats
+	var err error
+	if st.UsedMB, err = num(fields[0]); err != nil {
+		return GPUStats{}, fmt.Errorf("nvidia-smi: unexpected used memory in %q", strings.TrimSpace(out))
+	}
+	if st.TotalMB, err = num(fields[1]); err != nil {
+		return GPUStats{}, fmt.Errorf("nvidia-smi: unexpected total memory in %q", strings.TrimSpace(out))
+	}
+	st.TempC = optional(fields[2])
+	st.FanPct = optional(fields[3])
+	return st, nil
 }
 
 // parseComputeApps parses "pid, used_memory" CSV lines (no header, MiB

internal/game/game_test.go

@@ -116,6 +116,31 @@ func TestParseGPUEngineInstance(t *testing.T) {
 	}
 }
 
+func TestParseGPUStats(t *testing.T) {
+	st, err := parseGPUStats("4300, 16384, 55, 42\n")
+	if err != nil {
+		t.Fatal(err)
+	}
+	if st.UsedMB != 4300 || st.TotalMB != 16384 || st.TempC != 55 || st.FanPct != 42 {
+		t.Errorf("got %+v", st)
+	}
+
+	// Cards that don't expose temperature/fan report N/A.
+	st, err = parseGPUStats("1024, 16384, N/A, N/A")
+	if err != nil {
+		t.Fatal(err)
+	}
+	if st.TempC != -1 || st.FanPct != -1 {
+		t.Errorf("got %+v, want -1 for N/A fields", st)
+	}
+
+	for _, bad := range []string{"", "1, 2", "x, 16384, 55, 42", "1024, x, 55, 42", "1, 2, 3, 4, 5"} {
+		if _, err := parseGPUStats(bad); err == nil {
+			t.Errorf("%q parsed, want failure", bad)
+		}
+	}
+}
+
 func TestProcessesLive(t *testing.T) {
 	if runtime.GOOS != "windows" && runtime.GOOS != "linux" {
 		t.Skip("no process listing on this platform")

internal/ollama/client.go

@@ -60,13 +60,21 @@ func (c *Client) Probe(ctx context.Context) error {
 
 type psResponse struct {
 	Models []struct {
-		Name  string `json:"name"`
-		Model string `json:"model"`
+		Name     string `json:"name"`
+		Model    string `json:"model"`
+		SizeVRAM int64  `json:"size_vram"` // bytes resident in VRAM (0 = RAM-only)
 	} `json:"models"`
 }
 
-// LoadedModels returns the names of models currently held in memory.
-func (c *Client) LoadedModels(ctx context.Context) ([]string, error) {
+// LoadedModel is one model currently held in memory.
+type LoadedModel struct {
+	Name     string
+	SizeVRAM int64 // bytes resident in VRAM; 0 when the model sits in RAM
+}
+
+// LoadedModelDetails returns the models currently held in memory with
+// their VRAM footprint.
+func (c *Client) LoadedModelDetails(ctx context.Context) ([]LoadedModel, error) {
 	req, err := http.NewRequestWithContext(ctx, http.MethodGet, c.base+"/api/ps", nil)
 	if err != nil {
 		return nil, err
@@ -84,13 +92,26 @@ func (c *Client) LoadedModels(ctx context.Context) ([]string, error) {
 	if err := json.NewDecoder(resp.Body).Decode(&ps); err != nil {
 		return nil, err
 	}
-	models := make([]string, 0, len(ps.Models))
+	models := make([]LoadedModel, 0, len(ps.Models))
 	for _, m := range ps.Models {
-		if m.Name != "" {
-			models = append(models, m.Name)
-		} else {
-			models = append(models, m.Model)
+		name := m.Name
+		if name == "" {
+			name = m.Model
 		}
+		models = append(models, LoadedModel{Name: name, SizeVRAM: m.SizeVRAM})
+	}
+	return models, nil
+}
+
+// LoadedModels returns the names of models currently held in memory.
+func (c *Client) LoadedModels(ctx context.Context) ([]string, error) {
+	details, err := c.LoadedModelDetails(ctx)
+	if err != nil {
+		return nil, err
+	}
+	models := make([]string, 0, len(details))
+	for _, m := range details {
+		models = append(models, m.Name)
 	}
 	return models, nil
 }