remove noisy debug print message

Contribution for install, and uninstall llama-swap in linux.

README.md

@@ -46,14 +46,14 @@ llama-swap's configuration is purposefully simple.
 models:
   "qwen2.5":
     proxy: "http://127.0.0.1:9999"
-    cmd: >
+    cmd: |
       /app/llama-server
       -hf bartowski/Qwen2.5-0.5B-Instruct-GGUF:Q4_K_M
       --port 9999
 
   "smollm2":
     proxy: "http://127.0.0.1:9999"
-    cmd: >
+    cmd: |
       /app/llama-server
       -hf bartowski/SmolLM2-135M-Instruct-GGUF:Q4_K_M
       --port 9999
@@ -82,7 +82,7 @@ startPort: 10001
 models:
   "llama":
     # multiline for readability
-    cmd: >
+    cmd: |
       llama-server --port 8999
       --model path/to/Qwen2.5-1.5B-Instruct-Q4_K_M.gguf
 
@@ -123,12 +123,18 @@ models:
   # Docker Support (v26.1.4+ required!)
   "docker-llama":
     proxy: "http://127.0.0.1:${PORT}"
-    cmd: >
+    cmd: |
       docker run --name dockertest
       --init --rm -p ${PORT}:8080 -v /mnt/nvme/models:/models
       ghcr.io/ggerganov/llama.cpp:server
       --model '/models/Qwen2.5-Coder-0.5B-Instruct-Q4_K_M.gguf'
 
+    # use a custom command to stop the model when swapping. By default
+    # this is SIGTERM on POSIX systems, and taskkill on Windows systems
+    # the ${PID} variable can be used in cmdStop, it will be automatically replaced
+    # with the PID of the running model
+    cmdStop: docker stop dockertest
+
 # Groups provide advanced controls over model swapping behaviour. Using groups
 # some models can be kept loaded indefinitely, while others are swapped out.
 #
@@ -247,11 +253,11 @@ Pre-built binaries are available for Linux, FreeBSD and Darwin (OSX). These are
 1. Create a configuration file, see [config.example.yaml](config.example.yaml)
 1. Download a [release](https://github.com/mostlygeek/llama-swap/releases) appropriate for your OS and architecture.
 1. Run the binary with `llama-swap --config path/to/config.yaml`.
-  Available flags:
-    - `--config`: Path to the configuration file (default: `config.yaml`).
-    - `--listen`: Address and port to listen on (default: `:8080`).
-    - `--version`: Show version information and exit.
-    - `--watch-config`: Automatically reload the configuration file when it changes. This will wait for in-flight requests to complete then stop all running models (default: `false`).
+   Available flags:
+   - `--config`: Path to the configuration file (default: `config.yaml`).
+   - `--listen`: Address and port to listen on (default: `:8080`).
+   - `--version`: Show version information and exit.
+   - `--watch-config`: Automatically reload the configuration file when it changes. This will wait for in-flight requests to complete then stop all running models (default: `false`).
 
 ### Building from source
 

config.example.yaml

@@ -15,7 +15,7 @@ groups:
 
 models:
   "llama":
-    cmd: >
+    cmd: |
       models/llama-server-osx
       --port ${PORT}
       -m models/Llama-3.2-1B-Instruct-Q4_0.gguf
@@ -38,7 +38,7 @@ models:
   # Embedding example with Nomic
   # https://huggingface.co/nomic-ai/nomic-embed-text-v1.5-GGUF
   "nomic":
-    cmd: >
+    cmd: |
       models/llama-server-osx --port ${PORT}
       -m models/nomic-embed-text-v1.5.Q8_0.gguf
       --ctx-size 8192
@@ -51,7 +51,7 @@ models:
   # Reranking example with bge-reranker
   # https://huggingface.co/gpustack/bge-reranker-v2-m3-GGUF
   "bge-reranker":
-    cmd: >
+    cmd: |
       models/llama-server-osx --port ${PORT}
       -m models/bge-reranker-v2-m3-Q4_K_M.gguf
       --ctx-size 8192
@@ -59,7 +59,7 @@ models:
 
   # Docker Support (v26.1.4+ required!)
   "dockertest":
-    cmd: >
+    cmd: |
       docker run --name dockertest
       --init --rm -p ${PORT}:8080 -v /mnt/nvme/models:/models
       ghcr.io/ggerganov/llama.cpp:server

proxy/config.go

@@ -17,6 +17,7 @@ const DEFAULT_GROUP_ID = "(default)"
 
 type ModelConfig struct {
 	Cmd           string   `yaml:"cmd"`
+	CmdStop       string   `yaml:"cmdStop"`
 	Proxy         string   `yaml:"proxy"`
 	Aliases       []string `yaml:"aliases"`
 	Env           []string `yaml:"env"`
@@ -135,7 +136,6 @@ func LoadConfigFromReader(r io.Reader) (Config, error) {
 		}
 	}
 
-	// iterate over the models and replace any ${PORT} with the next available port
 	// Get and sort all model IDs first, makes testing more consistent
 	modelIds := make([]string, 0, len(config.Models))
 	for modelId := range config.Models {
@@ -143,10 +143,10 @@ func LoadConfigFromReader(r io.Reader) (Config, error) {
 	}
 	sort.Strings(modelIds) // This guarantees stable iteration order
 
-	// iterate over the sorted models
 	nextPort := config.StartPort
 	for _, modelId := range modelIds {
 		modelConfig := config.Models[modelId]
+		// iterate over the models and replace any ${PORT} with the next available port
 		if strings.Contains(modelConfig.Cmd, "${PORT}") {
 			modelConfig.Cmd = strings.ReplaceAll(modelConfig.Cmd, "${PORT}", strconv.Itoa(nextPort))
 			if modelConfig.Proxy == "" {
@@ -160,6 +160,7 @@ func LoadConfigFromReader(r io.Reader) (Config, error) {
 			return Config{}, fmt.Errorf("model %s requires a proxy value when not using automatic ${PORT}", modelId)
 		}
 	}
+
 	config = AddDefaultGroupToConfig(config)
 	// check that members are all unique in the groups
 	memberUsage := make(map[string]string) // maps member to group it appears in

proxy/config_windows_test.go

@@ -38,5 +38,4 @@ func TestConfig_SanitizeCommand(t *testing.T) {
 	args, err = SanitizeCommand("")
 	assert.Error(t, err)
 	assert.Nil(t, args)
-
 }

proxy/process.go

@@ -8,6 +8,7 @@ import (
 	"net/http"
 	"net/url"
 	"os/exec"
+	"runtime"
 	"strconv"
 	"strings"
 	"sync"
@@ -80,9 +81,8 @@ func NewProcess(ID string, healthCheckTimeout int, config ModelConfig, processLo
 	concurrentLimit := 10
 	if config.ConcurrencyLimit > 0 {
 		concurrentLimit = config.ConcurrencyLimit
-	} else {
-		proxyLogger.Debugf("Concurrency limit for model %s not set, defaulting to 10", ID)
 	}
+
 	return &Process{
 		ID:                      ID,
 		config:                  config,
@@ -148,7 +148,9 @@ func isValidTransition(from, to ProcessState) bool {
 		return to == StateStopping
 	case StateStopping:
 		return to == StateStopped || to == StateShutdown
-	case StateFailed, StateShutdown:
+	case StateFailed:
+		return to == StateStopping
+	case StateShutdown:
 		return false // No transitions allowed from these states
 	}
 	return false
@@ -358,12 +360,19 @@ func (p *Process) StopImmediately() {
 		return
 	}
 
-	p.proxyLogger.Debugf("<%s> Stopping process", p.ID)
+	p.proxyLogger.Debugf("<%s> Stopping process, current state: %s", p.ID, p.CurrentState())
+	currentState := p.CurrentState()
 
-	// calling Stop() when state is invalid is a no-op
-	if curState, err := p.swapState(StateReady, StateStopping); err != nil {
-		p.proxyLogger.Infof("<%s> Stop() Ready -> StateStopping err: %v, current state: %v", p.ID, err, curState)
-		return
+	if currentState == StateFailed {
+		if curState, err := p.swapState(StateFailed, StateStopping); err != nil {
+			p.proxyLogger.Infof("<%s> Stop() Failed -> StateStopping err: %v, current state: %v", p.ID, err, curState)
+			return
+		}
+	} else {
+		if curState, err := p.swapState(StateReady, StateStopping); err != nil {
+			p.proxyLogger.Infof("<%s> Stop() Ready -> StateStopping err: %v, current state: %v", p.ID, err, curState)
+			return
+		}
 	}
 
 	// stop the process with a graceful exit timeout
@@ -379,8 +388,14 @@ func (p *Process) StopImmediately() {
 // is in the state of starting, it will cancel it and shut it down. Once a process is in
 // the StateShutdown state, it can not be started again.
 func (p *Process) Shutdown() {
+	if !isValidTransition(p.CurrentState(), StateStopping) {
+		return
+	}
+
 	p.shutdownCancel()
 	p.stopCommand(p.gracefulStopTimeout)
+
+	// just force it to this state since there is no recovery from shutdown
 	p.state = StateShutdown
 }
 
@@ -400,8 +415,38 @@ func (p *Process) stopCommand(sigtermTTL time.Duration) {
 		return
 	}
 
-	if err := p.terminateProcess(); err != nil {
-		p.proxyLogger.Debugf("<%s> Process already terminated: %v (normal during shutdown)", p.ID, err)
+	// if err := p.terminateProcess(); err != nil {
+	// 	p.proxyLogger.Debugf("<%s> Process already terminated: %v (normal during shutdown)", p.ID, err)
+	// }
+	// the default cmdStop to taskkill /f /t /pid ${PID}
+	if runtime.GOOS == "windows" && strings.TrimSpace(p.config.CmdStop) == "" {
+		p.config.CmdStop = "taskkill /f /t /pid ${PID}"
+	}
+
+	if p.config.CmdStop != "" {
+		// replace ${PID} with the pid of the process
+		stopArgs, err := SanitizeCommand(strings.ReplaceAll(p.config.CmdStop, "${PID}", fmt.Sprintf("%d", p.cmd.Process.Pid)))
+		if err != nil {
+			p.proxyLogger.Errorf("<%s> Failed to sanitize stop command: %v", p.ID, err)
+			return
+		}
+
+		p.proxyLogger.Debugf("<%s> Executing stop command: %s", p.ID, strings.Join(stopArgs, " "))
+
+		stopCmd := exec.Command(stopArgs[0], stopArgs[1:]...)
+		stopCmd.Stdout = p.processLogger
+		stopCmd.Stderr = p.processLogger
+		stopCmd.Env = p.config.Env
+
+		if err := stopCmd.Run(); err != nil {
+			p.proxyLogger.Errorf("<%s> Failed to exec stop command: %v", p.ID, err)
+			return
+		}
+	} else {
+		if err := p.cmd.Process.Signal(syscall.SIGTERM); err != nil {
+			p.proxyLogger.Errorf("<%s> Failed to send SIGTERM to process: %v", p.ID, err)
+			return
+		}
 	}
 
 	select {

proxy/process_stop.go

@@ -1,9 +0,0 @@
-//go:build !windows
-
-package proxy
-
-import "syscall"
-
-func (p *Process) terminateProcess() error {
-	return p.cmd.Process.Signal(syscall.SIGTERM)
-}

proxy/process_stop_windows.go

@@ -1,14 +0,0 @@
-//go:build windows
-
-package proxy
-
-import (
-	"fmt"
-	"os/exec"
-)
-
-func (p *Process) terminateProcess() error {
-	pid := fmt.Sprintf("%d", p.cmd.Process.Pid)
-	cmd := exec.Command("taskkill", "/f", "/t", "/pid", pid)
-	return cmd.Run()
-}

proxy/process_test.go

@@ -449,3 +449,22 @@ func TestProcess_ForceStopWithKill(t *testing.T) {
 	// the request should have been interrupted by SIGKILL
 	<-waitChan
 }
+
+func TestProcess_StopCmd(t *testing.T) {
+	config := getTestSimpleResponderConfig("test_stop_cmd")
+
+	if runtime.GOOS == "windows" {
+		config.CmdStop = "taskkill /f /t /pid ${PID}"
+	} else {
+		config.CmdStop = "kill -TERM ${PID}"
+	}
+
+	process := NewProcess("testStopCmd", 2, config, debugLogger, debugLogger)
+	defer process.Stop()
+
+	err := process.start()
+	assert.Nil(t, err)
+	assert.Equal(t, process.CurrentState(), StateReady)
+	process.StopImmediately()
+	assert.Equal(t, process.CurrentState(), StateStopped)
+}

proxy/proxymanager.go

@@ -352,6 +352,8 @@ func (pm *ProxyManager) upstreamIndex(c *gin.Context) {
 					stateStr = "Failed"
 				case StateShutdown:
 					stateStr = "Shutdown"
+				case StateStopped:
+					stateStr = "Stopped"
 				default:
 					stateStr = "Unknown"
 				}

scripts/install.sh

@@ -0,0 +1,189 @@
+#!/bin/sh
+# This script installs llama-swap on Linux.
+# It detects the current operating system architecture and installs the appropriate version of llama-swap.
+
+set -eu
+
+red="$( (/usr/bin/tput bold || :; /usr/bin/tput setaf 1 || :) 2>&-)"
+plain="$( (/usr/bin/tput sgr0 || :) 2>&-)"
+
+status() { echo ">>> $*" >&2; }
+error() { echo "${red}ERROR:${plain} $*"; exit 1; }
+warning() { echo "${red}WARNING:${plain} $*"; }
+
+available() { command -v $1 >/dev/null; }
+require() {
+    local MISSING=''
+    for TOOL in $*; do
+        if ! available $TOOL; then
+            MISSING="$MISSING $TOOL"
+        fi
+    done
+
+    echo $MISSING
+}
+
+SUDO=
+if [ "$(id -u)" -ne 0 ]; then
+    if ! available sudo; then
+        error "This script requires superuser permissions. Please re-run as root."
+    fi
+
+    SUDO="sudo"
+fi
+
+NEEDS=$(require curl tee jq tar)
+if [ -n "$NEEDS" ]; then
+    status "ERROR: The following tools are required but missing:"
+    for NEED in $NEEDS; do
+        echo "  - $NEED"
+    done
+    exit 1
+fi
+
+[ "$(uname -s)" = "Linux" ] || error 'This script is intended to run on Linux only.'
+
+ARCH=$(uname -m)
+case "$ARCH" in
+    x86_64) ARCH="amd64" ;;
+    aarch64|arm64) ARCH="arm64" ;;
+    *) error "Unsupported architecture: $ARCH" ;;
+esac
+
+IS_WSL2=false
+
+KERN=$(uname -r)
+case "$KERN" in
+    *icrosoft*WSL2 | *icrosoft*wsl2) IS_WSL2=true;;
+    *icrosoft) error "Microsoft WSL1 is not currently supported. Please use WSL2 with 'wsl --set-version <distro> 2'" ;;
+    *) ;;
+esac
+
+download_binary() {
+    ASSET_NAME="linux_$ARCH"
+
+    # Fetch the latest release info and extract the matching asset URL
+    DL_URL=$(curl -s "https://api.github.com/repos/mostlygeek/llama-swap/releases/latest" | \
+        jq -r --arg name "$ASSET_NAME" \
+        '.assets[] | select(.name | contains($name)) | .browser_download_url')
+
+    # Check if a URL was successfully extracted
+    if [ -z "$DL_URL" ]; then
+        error "No matching asset found with name containing '$ASSET_NAME'."
+    fi
+
+    status "Downloading Linux $ARCH binary"
+    curl -s -L "$DL_URL" | $SUDO tar -xzf - -C /usr/local/bin llama-swap
+}
+download_binary
+
+configure_systemd() {
+    if ! id llama-swap >/dev/null 2>&1; then
+        status "Creating llama-swap user..."
+        $SUDO useradd -r -s /bin/false -U -m -d /usr/share/llama-swap llama-swap
+    fi
+    if getent group render >/dev/null 2>&1; then
+        status "Adding llama-swap user to render group..."
+        $SUDO usermod -a -G render llama-swap
+    fi
+    if getent group video >/dev/null 2>&1; then
+        status "Adding llama-swap user to video group..."
+        $SUDO usermod -a -G video llama-swap
+    fi
+    if getent group docker >/dev/null 2>&1; then
+        status "Adding llama-swap user to docker group..."
+        $SUDO usermod -a -G docker llama-swap
+    fi
+
+    status "Adding current user to llama-swap group..."
+    $SUDO usermod -a -G llama-swap $(whoami)
+
+    if [ ! -f "/usr/share/llama-swap/config.yaml" ]; then
+        status "Creating default config.yaml..."
+        cat <<EOF | $SUDO -u llama-swap tee /usr/share/llama-swap/config.yaml >/dev/null
+# default 15s likely to fail for default models due to downloading models
+healthCheckTimeout: 60
+
+models:
+  "qwen2.5":
+    cmd: |
+      docker run
+        --rm
+        -p \${PORT}:8080
+        --name qwen2.5
+      ghcr.io/ggml-org/llama.cpp:server
+        -hf bartowski/Qwen2.5-0.5B-Instruct-GGUF:Q4_K_M
+    cmdStop: docker stop qwen2.5
+
+  "smollm2":
+    cmd: |
+      docker run
+        --rm
+        -p \${PORT}:8080
+        --name smollm2
+      ghcr.io/ggml-org/llama.cpp:server
+        -hf bartowski/SmolLM2-135M-Instruct-GGUF:Q4_K_M
+    cmdStop: docker stop smollm2
+EOF
+    fi
+
+    status "Creating llama-swap systemd service..."
+    cat <<EOF | $SUDO tee /etc/systemd/system/llama-swap.service >/dev/null
+[Unit]
+Description=llama-swap
+After=network.target
+
+[Service]
+User=llama-swap
+Group=llama-swap
+
+# set this to match your environment
+ExecStart=/usr/local/bin/llama-swap --config /usr/share/llama-swap/config.yaml --watch-config
+
+Restart=on-failure
+RestartSec=3
+StartLimitBurst=3
+StartLimitInterval=30
+
+[Install]
+WantedBy=multi-user.target
+EOF
+    SYSTEMCTL_RUNNING="$(systemctl is-system-running || true)"
+    case $SYSTEMCTL_RUNNING in
+        running|degraded)
+            status "Enabling and starting llama-swap service..."
+            $SUDO systemctl daemon-reload
+            $SUDO systemctl enable llama-swap
+
+            start_service() { $SUDO systemctl restart llama-swap; }
+            trap start_service EXIT
+            ;;
+        *)
+            warning "systemd is not running"
+            if [ "$IS_WSL2" = true ]; then
+                warning "see https://learn.microsoft.com/en-us/windows/wsl/systemd#how-to-enable-systemd to enable it"
+            fi
+            ;;
+    esac
+}
+
+if available systemctl; then
+    configure_systemd
+fi
+
+install_success() {
+    status 'The llama-swap API is now available at 127.0.0.1:8080.'
+    status 'Customize the config file at /usr/share/llama-swap/config.yaml.'
+    status 'Install complete.'
+}
+
+# WSL2 only supports GPUs via nvidia passthrough
+# so check for nvidia-smi to determine if GPU is available
+if [ "$IS_WSL2" = true ]; then
+    if available nvidia-smi && [ -n "$(nvidia-smi | grep -o "CUDA Version: [0-9]*\.[0-9]*")" ]; then
+        status "Nvidia GPU detected."
+    fi
+    exit 0
+fi
+
+install_success

scripts/uninstall.sh

@@ -0,0 +1,68 @@
+#!/bin/sh
+# This script uninstalls llama-swap on Linux.
+# It removes the binary, systemd service, config.yaml (optional), and llama-swap user and group.
+
+set -eu
+
+red="$( (/usr/bin/tput bold || :; /usr/bin/tput setaf 1 || :) 2>&-)"
+plain="$( (/usr/bin/tput sgr0 || :) 2>&-)"
+
+status() { echo ">>> $*" >&2; }
+error() { echo "${red}ERROR:${plain} $*"; exit 1; }
+warning() { echo "${red}WARNING:${plain} $*"; }
+
+available() { command -v $1 >/dev/null; }
+
+SUDO=
+if [ "$(id -u)" -ne 0 ]; then
+    if ! available sudo; then
+        error "This script requires superuser permissions. Please re-run as root."
+    fi
+
+    SUDO="sudo"
+fi
+
+configure_systemd() {
+    status "Stopping llama-swap service..."
+    $SUDO systemctl stop llama-swap
+
+    status "Disabling llama-swap service..."
+    $SUDO systemctl disable llama-swap
+}
+if available systemctl; then
+    configure_systemd
+fi
+
+if available llama-swap; then
+    status "Removing llama-swap binary..."
+    $SUDO rm $(which llama-swap)
+fi
+
+if [ -f "/usr/share/llama-swap/config.yaml" ]; then
+    while true; do
+        printf "Delete config.yaml (/usr/share/llama-swap/config.yaml)? [y/N] " >&2
+        read answer
+        case "$answer" in
+            [Yy]* ) 
+                $SUDO rm -r /usr/share/llama-swap
+                break
+                ;;
+            [Nn]* | "" ) 
+                break
+                ;;
+            * ) 
+                echo "Invalid input. Please enter y or n."
+                ;;
+        esac
+    done
+fi
+
+if id llama-swap >/dev/null 2>&1; then
+    status "Removing llama-swap user..."
+    $SUDO userdel llama-swap
+fi
+
+if getent group llama-swap >/dev/null 2>&1; then
+    status "Removing llama-swap group..."
+    $SUDO groupdel llama-swap
+fi