proxy: add unload of single model (#318)

This adds a new API endpoint, /api/models/unload/*model, that unloads a single model. In the UI when a model is in a ReadyState it will have a new button to unload it. 

Fixes #312

proxy/processgroup.go

@@ -86,6 +86,29 @@ func (pg *ProcessGroup) HasMember(modelName string) bool {
 	return slices.Contains(pg.config.Groups[pg.id].Members, modelName)
 }
 
+func (pg *ProcessGroup) StopProcess(modelID string, strategy StopStrategy) error {
+	pg.Lock()
+
+	process, exists := pg.processes[modelID]
+	if !exists {
+		pg.Unlock()
+		return fmt.Errorf("process not found for %s", modelID)
+	}
+
+	if pg.lastUsedProcess == modelID {
+		pg.lastUsedProcess = ""
+	}
+	pg.Unlock()
+
+	switch strategy {
+	case StopImmediately:
+		process.StopImmediately()
+	default:
+		process.Stop()
+	}
+	return nil
+}
+
 func (pg *ProcessGroup) StopProcesses(strategy StopStrategy) {
 	pg.Lock()
 	defer pg.Unlock()

proxy/proxymanager.go

@@ -228,7 +228,6 @@ func (pm *ProxyManager) setupGinEngine() {
 		c.Redirect(http.StatusFound, "/ui/models")
 	})
 	pm.ginEngine.Any("/upstream/*upstreamPath", pm.proxyToUpstream)
-
 	pm.ginEngine.GET("/unload", pm.unloadAllModelsHandler)
 	pm.ginEngine.GET("/running", pm.listRunningProcessesHandler)
 	pm.ginEngine.GET("/health", func(c *gin.Context) {

proxy/proxymanager_api.go

@@ -3,8 +3,10 @@ package proxy
 import (
 	"context"
 	"encoding/json"
+	"fmt"
 	"net/http"
 	"sort"
+	"strings"
 
 	"github.com/gin-gonic/gin"
 	"github.com/mostlygeek/llama-swap/event"
@@ -23,6 +25,7 @@ func addApiHandlers(pm *ProxyManager) {
 	apiGroup := pm.ginEngine.Group("/api")
 	{
 		apiGroup.POST("/models/unload", pm.apiUnloadAllModels)
+		apiGroup.POST("/models/unload/*model", pm.apiUnloadSingleModelHandler)
 		apiGroup.GET("/events", pm.apiSendEvents)
 		apiGroup.GET("/metrics", pm.apiGetMetrics)
 	}
@@ -202,3 +205,25 @@ func (pm *ProxyManager) apiGetMetrics(c *gin.Context) {
 	}
 	c.Data(http.StatusOK, "application/json", jsonData)
 }
+
+func (pm *ProxyManager) apiUnloadSingleModelHandler(c *gin.Context) {
+	requestedModel := strings.TrimPrefix(c.Param("model"), "/")
+	realModelName, found := pm.config.RealModelName(requestedModel)
+	if !found {
+		pm.sendErrorResponse(c, http.StatusNotFound, "Model not found")
+		return
+	}
+
+	processGroup := pm.findGroupByModelName(realModelName)
+	if processGroup == nil {
+		pm.sendErrorResponse(c, http.StatusInternalServerError, fmt.Sprintf("process group not found for model %s", requestedModel))
+		return
+	}
+
+	if err := processGroup.StopProcess(realModelName, StopImmediately); err != nil {
+		pm.sendErrorResponse(c, http.StatusInternalServerError, fmt.Sprintf("error stopping process: %s", err.Error()))
+		return
+	} else {
+		c.String(http.StatusOK, "OK")
+	}
+}

proxy/proxymanager_test.go

@@ -401,11 +401,65 @@ func TestProxyManager_Unload(t *testing.T) {
 	assert.Equal(t, http.StatusOK, w.Code)
 	assert.Equal(t, w.Body.String(), "OK")
 
-	// give it a bit of time to stop
-	<-time.After(time.Millisecond * 250)
+	select {
+	case <-proxy.processGroups[DEFAULT_GROUP_ID].processes["model1"].cmdWaitChan:
+		// good
+	case <-time.After(2 * time.Second):
+		t.Fatal("timeout waiting for model1 to stop")
+	}
 	assert.Equal(t, proxy.processGroups[DEFAULT_GROUP_ID].processes["model1"].CurrentState(), StateStopped)
 }
 
+func TestProxyManager_UnloadSingleModel(t *testing.T) {
+	const testGroupId = "testGroup"
+	config := AddDefaultGroupToConfig(Config{
+		HealthCheckTimeout: 15,
+		Models: map[string]ModelConfig{
+			"model1": getTestSimpleResponderConfig("model1"),
+			"model2": getTestSimpleResponderConfig("model2"),
+		},
+		Groups: map[string]GroupConfig{
+			testGroupId: {
+				Swap:    false,
+				Members: []string{"model1", "model2"},
+			},
+		},
+		LogLevel: "error",
+	})
+
+	proxy := New(config)
+	defer proxy.StopProcesses(StopImmediately)
+
+	// start both model
+	for _, modelName := range []string{"model1", "model2"} {
+		reqBody := fmt.Sprintf(`{"model":"%s"}`, modelName)
+		req := httptest.NewRequest("POST", "/v1/chat/completions", bytes.NewBufferString(reqBody))
+		w := httptest.NewRecorder()
+		proxy.ServeHTTP(w, req)
+	}
+
+	assert.Equal(t, StateReady, proxy.processGroups[testGroupId].processes["model1"].CurrentState())
+	assert.Equal(t, StateReady, proxy.processGroups[testGroupId].processes["model2"].CurrentState())
+
+	req := httptest.NewRequest("POST", "/api/models/unload/model1", nil)
+	w := httptest.NewRecorder()
+	proxy.ServeHTTP(w, req)
+	assert.Equal(t, http.StatusOK, w.Code)
+	if !assert.Equal(t, w.Body.String(), "OK") {
+		t.FailNow()
+	}
+
+	select {
+	case <-proxy.processGroups[testGroupId].processes["model1"].cmdWaitChan:
+		// good
+	case <-time.After(2 * time.Second):
+		t.Fatal("timeout waiting for model1 to stop")
+	}
+
+	assert.Equal(t, proxy.processGroups[testGroupId].processes["model1"].CurrentState(), StateStopped)
+	assert.Equal(t, proxy.processGroups[testGroupId].processes["model2"].CurrentState(), StateReady)
+}
+
 // Test issue #61 `Listing the current list of models and the loaded model.`
 func TestProxyManager_RunningEndpoint(t *testing.T) {
 	// Shared configuration